Re: 700/23 vs 700/25 tires ?
Ryan Cousineau
<44c69398-5e8e-474b...@e3g2000vbe.googlegroups.com>,
Frank Krygowski <frkr...@gmail.com> wrote:
> On Feb 19, 9:48 pm, Ryan Cousineau <rcous...@gmail.com> wrote:
> > In article
> > <420e77f3-aeef-4290-9b9c-2c3b1712a...@o36g2000yqh.googlegroups.com>,
> > Frank Krygowski <frkry...@gmail.com> wrote:
> >
> >
> >
> > > On Feb 18, 9:47 pm, Ryan Cousineau <rcous...@gmail.com> wrote:
> >
> > > > If I could just interject at this point, most or all of the "extreme
> > > > Cervelo tri- or TT- bike" (any P3, P4, take your pick) are not UCI-legal
> > > > for road competition. Not just the aerobars or the wheels, the frames
> > > > themselves....
> >
> > > > I don't expect people to know this stuff. Nonetheless, please let it be
> > > > a lesson to those who are ignorant of UCI rules not to make ignorant
> > > > claims about what happens in races. Questions and research work better.
> >
> > > > (Specifically, I'm referring to "It's just another case where racers do
> > > > not grasp every advantage. IOW, they disagree with posters here.")
> >
> > > I think we all know the difference between a legal-for-time-trial bike
> > > and a legal-for-mass-start bike.
> >
> > Oh really?
>
> Well, I _thought_ so. Maybe some don't.
>
> >
> > > However, Cervelo does make road
> > > frames. They've been shown in some of the photos I've posted. And
> > > they tout aerodynamics of their road frames: "Outstanding
> > > aerodynamics Airfoil shapes designed by Cervélo for bicycle speeds,
> > > not for airplanes. Features that work, no aero gimmicks that work on a
> > > TT bike but not on a road bike."
> > > Therefore, my point remains: top level road racers still ride bikes
> > > without Cervelo's "outstanding aerodynamics." These racers do not
> > > grasp every possible advantage, but instead treat some as negligible.
> > > IOW, they disagree with posters here.
> >
> > Wait, Frank, what are you arguing here? Because before you were arguing
> > this...
> >
> > "Again, Chung pointed to an extreme Cervelo tri- or TT- bike,
> > supposedly most aero in the world, as proof that aero matters. Yet
> > it's very rarely used in road races. It's just another case where
> > racers do not grasp every advantage. IOW, they disagree with posters
> > here."
> >
> > <http://groups.google.com/group/rec.bicycles.tech/msg/092dbf57c5a8365d>
> >
> > To which I reasonably pointed out that they don't use Cervelo's TT
> > frames because they're illegal in mass-start road races (1.3.020,
> > sigh...)
>
> Ah yes, since this is Usenet, every word must be precisely correct.
> The overall correctness of an idea is invalid if one word might be
> pointed to as an error. Now I remember.
>
> Let me change, then, to "Chung, in his continued tendency to flee to
> the subject of time trials, pointed to an extreme Cervelo tri- or TT-
> bike, supposedly most aero in the world, as proof that aero matters.
> But Cervelo also makes road bikes that are legal for road racing, and
> are touted for their "outstanding aerodynamics." Yet those legal-for-
> road-racing frames are very rarely used in road races. It's just
> another case where racers do not grasp every advantage. IOW, they
> disagree with posters here."
>
> How's that? Same point, and I think all the details are up to
> exacting Usenet standards.
> > But most Cervelo-sponsored pros do use their most aero road frame in
> > road races.
>
> And most Cannondale-sponsored pros use Cannondale's top frame, and
> most Trek-sponsored pros use Trek's top frame. Not as aero, but those
> riders seem to get by.
Now you're on shakier ground, for two reasons. First, the point of pro
riders is to get paid*, and thus they almost always use their sponsor's
gear. Second, the narrower design parameters of UCI-legal road frames
(versus TT frames) mean there is likely to be less variance in frame
performance. Third, I know of very good comparisons of TT-frame aero
performance, but I haven't read a good comparison of road-frame aero
performance, so unlike you I'm not going to claim that the Cervelo S3 is
more aero than the Trek Madone 6.9.
> Now let's move away from sponsored pros to the thousands upon
> thousands of other road racers. They very rarely use Cervelo frames.
> As I said, it's just another case where racers do not grasp every
> advantage. IOW, they disagree with posters here.
Frank, seriously? Because they're amateurs. And like amateur golfers, a
lot of them, any other considerations notwithstanding, can't afford or
refuse to spend the price of the latest and greatest.
Personally, I'd rather eat food and race a bike with a $100 frame than
make the lifestyle compromises (starvation and divorce) required for me
to afford a $3000+ frame. Others, notably Cervelo owners, are either
richer, take cycling more seriously, or are already single. There may be
other explanations.
> Personally, I think it's crazy to continue pretending that racers must
> never (or never do) dismiss an advantage as negligible. That point
> has been so easily, and so often, disproven that the "nothing is
> negligible" crew should formally concede.
Frank, I'm just sitting here, eating popcorn, and calling you on your
more hilarious errors. I think about the best you'll be able to pin on
the posters in this thread is that some of them think racers should care
about smaller details than actual racers do care about. Of course, that
leads you into another loop, where you'd now be arguing that racer
behavior is definitive evidence of what kinds of details do matter to
performance.
> We can then concentrate on the "small changes can add up to big
> changes" idea. That's more interesting, since there are some
> situations in which that has validity and some where it doesn't.
That'd be great! Don't change the thread title, though: we're going for
the record.
*no shame in that, mind.
--
Ryan Cousineau rcou...@gmail.com http://www.wiredcola.com/
"In other newsgroups, they killfile trolls."
"In rec.bicycles.racing, we coach them."
carl...@comcast.net
wrote:
[snip]
>That'd be great! Don't change the thread title, though: we're going for
>the record.
Dear Ryan,
Pack a lunch. A recent helmet thread ran over 5,000 posts.
But every post helps!
Cheers,
Carl Fogel
Robert Chung
> Frank Krygowski <frkr...@gmail.com> wrote:
>> Ryan Cousineau wrote:
>>> But most Cervelo-sponsored pros do use their most aero road frame in
>>> road races.
>>
>> And most Cannondale-sponsored pros use Cannondale's top frame, and
>> most Trek-sponsored pros use Trek's top frame. Not as aero, but
>> those riders seem to get by.
>
> Now you're on shakier ground, for two reasons. First, the point of pro
> riders is to get paid*, and thus they almost always use their
> sponsor's gear.
I've pointed this out to Frank several times. What makes you think he'll
suddenly become rational this time?
>> Now let's move away from sponsored pros to the thousands upon
>> thousands of other road racers. They very rarely use Cervelo frames.
>> As I said, it's just another case where racers do not grasp every
>> advantage. IOW, they disagree with posters here.
>
> Frank, seriously? Because they're amateurs.
Ryan, I think it's become clear that Frank doesn't understand the
difference. He appears to confuse TT bikes with road bikes, he's unaware of
UCI rules, and he thinks pro riders can use whatever bikes and equipment
they want.
>> Personally, I think it's crazy to continue pretending that racers
>> must never (or never do) dismiss an advantage as negligible. That
>> point has been so easily, and so often, disproven that the "nothing
>> is negligible" crew should formally concede.
Personally, I think it's crazy for Frank to continue pretending that he's
got a defensible argument based on "pro riders in road races don't use
Cervelo TT bikes" --but it's kinda entertaining watching him try.
On the other hand, we appear to have broken Google Groups' ability to track
the thread count.
Frank Krygowski
> Frank Krygowski <frkry...@gmail.com> wrote:
>
>
> > Let me change, then, to "Chung, in his continued tendency to flee to
> > the subject of time trials, pointed to an extreme Cervelo tri- or TT-
> > bike, supposedly most aero in the world, as proof that aero matters.
> > But Cervelo also makes road bikes that are legal for road racing, and
> > are touted for their "outstanding aerodynamics." Yet those legal-for-
> > road-racing frames are very rarely used in road races. It's just
> > another case where racers do not grasp every advantage. IOW, they
> > disagree with posters here."
>
> > How's that? Same point, and I think all the details are up to
> > exacting Usenet standards.
> > > But most Cervelo-sponsored pros do use their most aero road frame in
> > > road races.
>
> > And most Cannondale-sponsored pros use Cannondale's top frame, and
> > most Trek-sponsored pros use Trek's top frame. Not as aero, but those
> > riders seem to get by.
>
> Now you're on shakier ground, for two reasons. First, the point of pro
> riders is to get paid*, and thus they almost always use their sponsor's
> gear. Second, the narrower design parameters of UCI-legal road frames
> (versus TT frames) mean there is likely to be less variance in frame
> performance. Third, I know of very good comparisons of TT-frame aero
> performance, but I haven't read a good comparison of road-frame aero
> performance, so unlike you I'm not going to claim that the Cervelo S3 is
> more aero than the Trek Madone 6.9.
Admittedly, I can't prove that the Cervelo is more slippery than the
Trek. However, since Cervelo built its reputation almost entirely on
aero, I think most potential buyers would assume the Cervelo was more
aero. And that's good enough for us to examine racers' motivation and
equipment strategy.
Now, as to the first point above: Sponsored racers use their sponsors
equipment. (Well, most of the time. We know about re-badging,
repainting, etc. But we also know it's going to be impossible to
repaint a Cervelo into a Trek.) Still, I've got an easy job here:
I'm arguing against absolutist positions that require only one
counterexample. I'm arguing against "No detail is negligible" and "A
racer who dismisses a potential advantage as negligible has already
lost the race."
So if (as out TT-fixated friend thinks) any tiny bit of aero drag ,
even finger position, is critically important; and if a Trek sponsored
rider accepts more aero drag from his free Trek frame, then indeed,
that Trek rider is deciding that the slight difference in aero drag on
his frame is negligible. Or at least, he's comparing tiny benefits
versus real detriments and making a rational decision, which is what
I've been advocating all along.
>
> > Now let's move away from sponsored pros to the thousands upon
> > thousands of other road racers. They very rarely use Cervelo frames.
> > As I said, it's just another case where racers do not grasp every
> > advantage. IOW, they disagree with posters here.
>
> Frank, seriously? Because they're amateurs. And like amateur golfers, a
> lot of them, any other considerations notwithstanding, can't afford or
> refuse to spend the price of the latest and greatest.
Of course! But by doing so, they are adding to the chorus that agrees
with me. They're saying "Oh, it's not likely to matter
significantly." And every time someone wins a little (or big) race on
a non-Cervelo frame, they prove they were right.
> > Personally, I think it's crazy to continue pretending that racers must
> > never (or never do) dismiss an advantage as negligible. That point
> > has been so easily, and so often, disproven that the "nothing is
> > negligible" crew should formally concede.
>
> Frank, I'm just sitting here, eating popcorn, and calling you on your
> more hilarious errors. I think about the best you'll be able to pin on
> the posters in this thread is that some of them think racers should care
> about smaller details than actual racers do care about.
Well, that, plus they're wildly overstating their positions, using
either inappropriate absolute statements, or giving TT examples of
wind tunnel finger positions, or retreating to "hahaha you just don't
understand."
> > We can then concentrate on the "small changes can add up to big
> > changes" idea. That's more interesting, since there are some
> > situations in which that has validity and some where it doesn't.
>
> That'd be great! Don't change the thread title, though: we're going for
> the record.
OK. Separate post, though.
- Frank Krygowski
Tim McNamara
"Robert Chung" <anonymou...@invalid.address> wrote:
> Ryan Cousineau wrote:
> > Frank Krygowski <frkr...@gmail.com> wrote:
> >> Ryan Cousineau wrote:
>
> >>> But most Cervelo-sponsored pros do use their most aero road frame
> >>> in road races.
> >>
> >> And most Cannondale-sponsored pros use Cannondale's top frame, and
> >> most Trek-sponsored pros use Trek's top frame. Not as aero, but
> >> those riders seem to get by.
> >
> > Now you're on shakier ground, for two reasons. First, the point of
> > pro riders is to get paid*, and thus they almost always use their
> > sponsor's gear.
>
> I've pointed this out to Frank several times. What makes you think
> he'll suddenly become rational this time?
Do either of you guys pay any attention to pro bike racing? There are
many, many examples of top pros *not* using the sponsor's equipment and
using their own instead. Lance Armstrong and Andy Hampsten rode
Lightspeed frames instead of Merckx frames, painted up in the team
colors. Armstrong rode a Lightspeed Blade instead of a Trek in TTs for
a while. This stuff still goes on.
Frank Krygowski
> Frank Krygowski <frkry...@gmail.com> wrote:
>
> > Personally, I think it's crazy to continue pretending that racers must
> > never (or never do) dismiss an advantage as negligible. That point
> > has been so easily, and so often, disproven that the "nothing is
> > negligible" crew should formally concede.
>
> > changes" idea. That's more interesting, since there are some
> > situations in which that has validity and some where it doesn't.
>
> That'd be great! Don't change the thread title, though: we're going for
> the record.
OK, regarding small changes and big changes, there is a sort of
paradox to deal with. We know (well, most people know) that reducing
bike+rider mass, rolling resistance, or aero drag by 0.5% will never
be shown to improve a rider's results in a road race, crit race, or
matched sprint. It will certainly never add significantly to the
person's enjoyment of a recreational ride.
Yet most cyclists are sure that reducing any of those three parameters
by, say, 20% will provide detectable benefit in racing placement. If
each 0.5% is negligible, how do we resolve the somewhat simplistic
idea that adding up 40 negligibles becomes non-negligible? This might
be the Zeno's paradox of bike technology.
Here's an imperfect analogy to explain it. It's based on the fact a
race has hundreds of tiny events, hundreds of factors that affect the
outcome. Many are based on training, or genetics (can you grind out
one more good sprint, even though you had to catch up after
flatting?). Some are based on the course (does a hilly course give
advantage to your lighter body weight, or is the course flat, giving
advantage to your competitor's muscular power?). Some are based on
equipment (Is your expensive frame really a bit more aero? Did you
really luck out by picking better tires for the course, and set them
to optimum tire pressure both front and back?) Many are completely
random (like, when you were in the center of the pack, did you run
over the thumbtack or did the guy to your left? Did your teammate
lead you out nicely, or was he stuck behind three other guys at the
wrong time?)
You might model this by rolling dice. Hundreds of dice. In this
thought experiment, the total of all the dice values rolled indicates
your chance of winning, and if your competitor has a higher score, he
beats you.
But how do we account for the relative importance of different
factors? Perhaps by assigning more dice to them. If we knew for sure
that genetics was 25% of what determined racing success, we could say
250 dice out of a total of 1000 were genetic dice. If training was
another 25%, there'd be 250 training dice. Say 100 strategy dice...
and so on.
Now you get to shave the dice. You put your effort into trimming dice
faces so they come up with numbers in your favor. Of course, some you
can't shave, because they've been pre-shaved; those would be the
genetic ones. They'd also be the ones representing course contour and
length, and size of field. And others are unshaveable - the ones that
represent truly random events that matter in every mass-start race.
These are the untoucheable "luck" dice.
Where do the equipment dice fit in this thought experiment? There may
be (um... we can argue about this) 75 aerodynamics dice. One of them
might represent the drag on your water bottle. By shaving that die,
you're increasing your chance of winning.... by a little bit. You can
also shave the die that represents the mass of your cassette cluster,
or the size of your wristwatch.
Each time you shave one of those equipment dice, you don't guarantee
it's going to come up a "six." You're slightly increasing the odds it
will. The more you shave, the better its particular odds. But the
effect on the overall odds of all the diceis not great. All those
genetic dice, training dice and strategy dice still matter a whole lot
more.
If you do serious shaving of ALL the equipment dice, you might get a
significant total benefit. But it's still a statistical sort of
benefit. Sometimes it will matter, when all the other factors
affecting you and your competitor are (by amazingly low probability)
exactly even. Unfortunately, there are so many "luck" dice in a road
race that you'll have to perform thousands of rolls to detect that
tiny change in odds.
Now for our TT-fixated friend: Time trials have way, way fewer luck
dice. And because of that, the genetic dice and training dice have
much more consistent and observable effects. It's much easier to tell
who's a great time trialist; the results are much more consistent than
those of road races, even (or especially) down in the amateur ranks.
Since there are fewer "luck" dice, enough shaving of the equipment
dice can have a more easily detected effect. But if there's any room
left to shave all the "training" dice, it's still more productive to
work on them.
But even in a TT, shaving one die - say, the "finger position on the
bar" die - is not likely to be detectable in the final total. Doing
so in the road race is just distracting yourself; it's allowing the
"strategy" dice to go completely random, and there are lots more of
them.
OK, that's my imperfect attempt to explain by analogy. There are lots
of random or semi-random factors involved in race placement. Whether
any one factor affects placing is a statistical event. Adding lots of
slight changes in probability may, possibly, cause a different total
result, but far from guaranteed. Making one or two tiny changes in
probability is negligible.
(Oh, and the OP? While we're all shaving dice, he's out for a
recreational ride with his girlfriend. He put the 25s on her bike,
saying "I learned they're more comfortable _and_ a little easier
rolling than the 23s you had, so I bought them for you." See, he'd
actually like them to finish at the same time. ;-)
- Frank Krygowski
Robert Chung
That's why Ryan used the words "almost always," and why I almost always
point out that they use the equipment dictated by their team unless they
have a separate individual deal, or get a special dispensation. You don't
see Cancellara painting up a P4 to look like a Transition (though he may
have lucked out if rumors are true: the Transition is supposedly quite
aero). Last year Garmin-Chipotle allowed some of their riders to ride
without the Garmin 705 on their bikes for certain stages but in general as a
lead sponsor Garmin likes the riders to use Garmin products.
A Muzi
>>>>> But most Cervelo-sponsored pros do use their most aero road frame
>>>>> in road races.
>>> Frank Krygowski <frkr...@gmail.com> wrote:
>>>> And most Cannondale-sponsored pros use Cannondale's top frame, and
>>>> most Trek-sponsored pros use Trek's top frame. Not as aero, but
>>>> those riders seem to get by.
>> Ryan Cousineau wrote:
>>> Now you're on shakier ground, for two reasons. First, the point of
>>> pro riders is to get paid*, and thus they almost always use their
>>> sponsor's gear.
> "Robert Chung" <anonymou...@invalid.address> wrote:
>> I've pointed this out to Frank several times. What makes you think
>> he'll suddenly become rational this time?
Tim McNamara wrote:
> Do either of you guys pay any attention to pro bike racing? There are
> many, many examples of top pros *not* using the sponsor's equipment and
> using their own instead. Lance Armstrong and Andy Hampsten rode
> Lightspeed frames instead of Merckx frames, painted up in the team
> colors. Armstrong rode a Lightspeed Blade instead of a Trek in TTs for
> a while. This stuff still goes on.
So, in Mr Armstrong and Mr Hampsten's expert opinion, small equipment
differences matter.
--
Andrew Muzi
<www.yellowjersey.org/>
Open every day since 1 April, 1971
A Muzi
>>> Personally, I think it's crazy to continue pretending that racers must
>>> never (or never do) dismiss an advantage as negligible. That point
>>> has been so easily, and so often, disproven that the "nothing is
>>> negligible" crew should formally concede.
>>> We can then concentrate on the "small changes can add up to big
>>> changes" idea. That's more interesting, since there are some
>>> situations in which that has validity and some where it doesn't.
> Ryan Cousineau <rcous...@gmail.com> wrote:
>> That'd be great! Don't change the thread title, though: we're going for
>> the record.
Your dice analogy is good in that it pretty much reflects reality. One
has limited control of some variables which may or may not be critical
at any given moment or in any given event.
Chalo
>
> On the other hand, we appear to have broken Google Groups' ability to track
> the thread count.
U broke teh internets!
John Forrest Tomlinson
<frkr...@gmail.com> wrote:
[tiny difference in rolling resistance]
>It will certainly never add significantly to the
>person's enjoyment of a recreational ride.
Thanks for clearing that up for us again Frank.
I was getting the urge to post about how critical tire choice is to
happiness on rec rides, but you're keeping me honest.
Thanks.
Tim McNamara
<acdf924c-dd7e-4a20...@o36g2000yqh.googlegroups.com>,
Frank Krygowski <frkr...@gmail.com> wrote:
> On Feb 20, 1:39 am, Ryan Cousineau <rcous...@gmail.com> wrote:
> > In article
> > <44c69398-5e8e-474b-8c19-21daac1ed...@e3g2000vbe.googlegroups.com>,
> > Frank Krygowski <frkry...@gmail.com> wrote:
> >
> > > Personally, I think it's crazy to continue pretending that racers
> > > must never (or never do) dismiss an advantage as negligible.
> > > That point has been so easily, and so often, disproven that the
> > > "nothing is negligible" crew should formally concede.
> >
> > > We can then concentrate on the "small changes can add up to big
> > > changes" idea. That's more interesting, since there are some
> > > situations in which that has validity and some where it doesn't.
> >
> > That'd be great! Don't change the thread title, though: we're going
> > for the record.
>
> OK, regarding small changes and big changes, there is a sort of
> paradox to deal with. We know (well, most people know) that reducing
> bike+rider mass, rolling resistance, or aero drag by 0.5% will never
> be shown to improve a rider's results in a road race, crit race, or
> matched sprint.
The data provided in one of Robert's links suggested that a significant
difference could be found in something like match sprinting, which is
often separated by .0x seconds. In roads and crits, the difference
would be equivocal due to the aerodynamic and psychological effects of
pack riding. The latter is something that isn't addressable in this
thread and may not be "modelable."
> It will certainly never add significantly to the
> person's enjoyment of a recreational ride.
Probably not, but the focus of the benefits under discussion is really
in the competitive arena. A more comfortable saddle and properly
adjusted bike fit are probably the biggest improvements that can be
provided to recreational cyclists.
Tim McNamara
A Muzi <a...@yellowjersey.org> wrote:
I bet that's true. Hampsten even used Pino Morroni skewers; Pino made
his entire career out of splitting hairs.
Ryan Cousineau
carl...@comcast.net wrote:
Just what makes that little old ant, think he'll move that rubber tree
plant?
Wait, what was the question?
Ryan Cousineau
<2e0a1c83-fc24-4f80...@d32g2000yqe.googlegroups.com>,
Chalo <chalo....@gmail.com> wrote:
First, I hoped rbr could be as useful as rbt. Then, I hoped that rbt
could be as decorous as rbr. Now, I've decided that by combining the
most ridiculous attributes of each newsgroup, we can create something
even better.
rbr: you have to argue about helmets now.
rbt: start addressing each other as "dumbass."
Thanks,
Usenet Police, rb* task force
Ryan Cousineau
"Robert Chung" <anonymou...@invalid.address> wrote:
> Ryan Cousineau wrote:
> > Frank Krygowski <frkr...@gmail.com> wrote:
> >> Ryan Cousineau wrote:
>
> >>> But most Cervelo-sponsored pros do use their most aero road frame in
> >>> road races.
> >>
> >> And most Cannondale-sponsored pros use Cannondale's top frame, and
> >> most Trek-sponsored pros use Trek's top frame. Not as aero, but
> >> those riders seem to get by.
> >
> > Now you're on shakier ground, for two reasons. First, the point of pro
> > riders is to get paid*, and thus they almost always use their
> > sponsor's gear.
>
> I've pointed this out to Frank several times. What makes you think he'll
> suddenly become rational this time?
Because I'm so nice.
> >> Now let's move away from sponsored pros to the thousands upon
> >> thousands of other road racers. They very rarely use Cervelo frames.
> >> As I said, it's just another case where racers do not grasp every
> >> advantage. IOW, they disagree with posters here.
> >
> > Frank, seriously? Because they're amateurs.
>
> Ryan, I think it's become clear that Frank doesn't understand the
> difference. He appears to confuse TT bikes with road bikes, he's unaware of
> UCI rules, and he thinks pro riders can use whatever bikes and equipment
> they want.
>
> >> Personally, I think it's crazy to continue pretending that racers
> >> must never (or never do) dismiss an advantage as negligible. That
> >> point has been so easily, and so often, disproven that the "nothing
> >> is negligible" crew should formally concede.
>
> Personally, I think it's crazy for Frank to continue pretending that he's
> got a defensible argument based on "pro riders in road races don't use
> Cervelo TT bikes" --but it's kinda entertaining watching him try.
I've got a lot of popcorn.
> On the other hand, we appear to have broken Google Groups' ability to track
> the thread count.
I have not, as a result, lost any respect for rbt or rbr.
A Muzi
Tim McNamara wrote:
> I bet that's true. Hampsten even used Pino Morroni skewers; Pino made
> his entire career out of splitting hairs.
The world is poorer without him. What a guy!
Ryan Cousineau
<034339c6-bb87-4e42...@v5g2000prm.googlegroups.com>,
Frank Krygowski <frkr...@gmail.com> wrote:
Frank, nobody else in this thread thought they'd have to point out that
pro bike racers are keen on getting paid. Also, do you know the
difference between "negligible" and "worthwhile compromise?"
A hangnail is a negligible injury. The agony of chemotherapy is a
worthwhile compromise.
Riders routinely use the best equipment possible, given the limits of
imperfect knowledge and imperfect economics. Within the parameters of
sponsorship, rules, economics, and their own limitations, the riders and
their teams seek the best possible equipment. They don't always make the
best choices.
> > > Now let's move away from sponsored pros to the thousands upon
> > > thousands of other road racers. They very rarely use Cervelo frames.
> > > As I said, it's just another case where racers do not grasp every
> > > advantage. IOW, they disagree with posters here.
> >
> > Frank, seriously? Because they're amateurs. And like amateur golfers, a
> > lot of them, any other considerations notwithstanding, can't afford or
> > refuse to spend the price of the latest and greatest.
>
> Of course! But by doing so, they are adding to the chorus that agrees
> with me. They're saying "Oh, it's not likely to matter
> significantly." And every time someone wins a little (or big) race on
> a non-Cervelo frame, they prove they were right.
None of these three paragraphs mean what you think they mean.
> > > Personally, I think it's crazy to continue pretending that racers must
> > > never (or never do) dismiss an advantage as negligible. That point
> > > has been so easily, and so often, disproven that the "nothing is
> > > negligible" crew should formally concede.
> >
> > Frank, I'm just sitting here, eating popcorn, and calling you on your
> > more hilarious errors. I think about the best you'll be able to pin on
> > the posters in this thread is that some of them think racers should care
> > about smaller details than actual racers do care about.
>
> Well, that, plus they're wildly overstating their positions, using
> either inappropriate absolute statements, or giving TT examples of
> wind tunnel finger positions, or retreating to "hahaha you just don't
> understand."
I stand by my previous statement. And now I need more popcorn.
Robert Chung
> I stand by my previous statement. And now I need more popcorn.
See, this is what I meant about Frank being kinda entertaining, in a Pee Wee
Herman froth-at-the-mouth sort of way.
bjwe...@gmail.com
>
> Well, that, plus they're wildly overstating their positions, using
> either inappropriate absolute statements, or giving TT examples of
> wind tunnel finger positions, or retreating to "hahaha you just don't
> understand."
>
You demanded a quantified example from me.
I provided it using tire Crr and a road race (not a TT).
Which of these categories did you fit my clearly
worked out example into?
You had said:
http://groups.google.com/group/rec.bicycles.tech/msg/ab190b0b70849274
> It takes a big change in weight or rolling resistance to be
> detectable.
(I think you meant detectable in terms of speed/results,
here.)
http://groups.google.com/group/rec.bicycles.tech/msg/84e006ffea2f1bb1
> I know you're writing to Carl, but (to use your phrasing): What I've
> said is that something that has a tiny or extremely difficult to
> measure effect has negligible effect. IOW, any benefits would be lost
> in the noise. IOW, you could not pick their effect out of the race
> results. IOW, any scientific analysis would say "Not worth it."
I think what I demonstrated is that a very small
difference in rolling resistance, which requires
painstaking efforts to measure, can have
non-negligible effect. You may choose to believe
that this Crr difference was not tiny nor difficult
to measure. In that case, your expression would be
tautologous: once an effect is demonstrated, it is
asserted to be not tiny, therefore your expression is
always right, at the cost of conveying zero information.
I don't _like_ that equipment or wind tunnel testing
makes a (small) difference. In my ideal world,
I wouldn't have to worry about tire choices, and
the only way I could improve my results would be
to train harder and not waste time arguing on
Usenet. However, it isn't my ideal world, and
I've been convinced by data of that. You don't
seem willing to be convinced of anything. Once
I demonstrated it, you had known it all along
and went back to arguing about dimes:
http://groups.google.com/group/rec.bicycles.tech/msg/8d01b2df221a8f32
This is unscientific. It's also lame.
Ben
Robert Chung
> I think what I demonstrated is that a very small
> difference in rolling resistance, which requires
> painstaking efforts to measure, can have
> non-negligible effect. You may choose to believe
> that this Crr difference was not tiny nor difficult
> to measure.
I'm not sure that's what Frank is doing. I think he's actually saying that
the difference really is negligible.
I got into this thread kinda late, with what I thought was just a
clarificatory question to Frank: since he determines negligibility by
detectability (thought I), how small of an effect could he detect? But the
evidence from this thread is that his position is quite different: he's
decided the effect is negligible, so there's no need to detect it.
Michael Press
Tim McNamara <tim...@bitstream.net> wrote:
I hope the skewers hold up better than the hairs.
Come to think of it, the hairs are carbon fibre;
why were they splitting?
--
Michael Press
Frank Krygowski
wrote:
> On Feb 20, 8:55 am, Frank Krygowski <frkry...@gmail.com> wrote:
>
>
>
> > Well, that, plus they're wildly overstating their positions, using
> > either inappropriate absolute statements, or giving TT examples of
> > wind tunnel finger positions, or retreating to "hahaha you just don't
> > understand."
>
> You demanded a quantified example from me.
> I provided it using tire Crr and a road race (not a TT).
> Which of these categories did you fit my clearly
> worked out example into?
> I think what I demonstrated is that a very small
> difference in rolling resistance, which requires
> painstaking efforts to measure, can have
> non-negligible effect.
Did you read my re-do of your calculations, where I used tires that
actually fit the OP's question? What I did was change just one
factor, the tire size, while using the same brand and model.
The upshot was, if we do look at one change in isolation, the
difference the OP asked about was indeed negligible. Not the (IIRC) 9
seconds or so that you claimed, but closer to one second.
Furthermore, I pointed out why even that one second would be reduced,
because it was actually based on the fact that the _wider_ tire had
less rolling resistance. That is, the wider, heavier, less aero tire
would have shortcomings to work against the Crr benefit.
And of course, your calculation method assumes neither rider drafts
anyone, not even each other. I went further to take that into
account, and show yet another reason the small change that started
this discussion is indeed, negligible.
I just re-pasted all that back into a splinter thread, because Robert
Chung is pretending to have missed it. I can paste it here if you
missed it too.
> I don't _like_ that equipment or wind tunnel testing
> makes a (small) difference. In my ideal world,
> I wouldn't have to worry about tire choices...
You don't have to worry much about your tire choice. Scanning a table
ranked by Crr is sufficient to keep you from buying the worst tires on
the market; but nobody should think their place at the finish line is
significantly affected by slight differences in ranking. Like it or
not, if you put your tire-shopping time into interval training, it's
probably going to make a lot more difference.
BTW, it occurs to me that many - probably most - of those reading this
are past their racing prime. We are never going to take another first
place. (Hell, best I ever did was second place.) But let's think for
a minute about the young kids who might be really good someday. Do we
want to be telling them to train and learn tactics? Or do we want to
be telling them that they can never win if they ignore any tiny - and
expensive - potential equipment advantage?
- Frank Krygowski
Tim McNamara
A Muzi <a...@yellowjersey.org> wrote:
> -snip snip- Tim McNamara wrote:
> > I bet that's true. Hampsten even used Pino Morroni skewers; Pino
> > made his entire career out of splitting hairs.
>
> The world is poorer without him. What a guy!
He sounds like he was a character from the people who did meet him.
About 10 years ago, the Italian track team came to the Blaine Velodrome
because it was a very similar track to wherever the Olympics or the
World championships were being held. According to the track director,
some had Pino parts on their bikes and they held him in very high esteem.
Eddy Mercxk had a Pino titanium stem on his Hour Record bike.
http://www.classicrendezvous.com/USA/Pino_Moroni_main.htm
http://chainedrevolution.com/CS/forums/thread/562.aspx
http://www.cyclingutah.com/april/april99/classic.html
Etc. A google search for "Pino Morroni" turns up all kinds of stuff.
carl...@comcast.net
Dear Tim,
Thanks--nice links.
Cheers,
Carl Fogel
carl...@comcast.net
> In article <gnnshl$5f...@news.motzarella.org>,> > > I bet that's true. Hampsten even usedPinoMorroni skewers;Pino
> > > made his entire career out of splitting hairs.
>
> > The world is poorer without him. What a guy!
>
> He sounds like he was a character from the people who did meet him.
> About 10 years ago, the Italian track team came to the Blaine Velodrome
> because it was a very similar track to wherever the Olympics or the
> World championships were being held. According to the track director,
>
> Eddy Mercxk had aPinotitanium stem on his Hour Record bike.
>
> http://www.classicrendezvous.com/USA/Pino_Moroni_main.htm
>
> http://chainedrevolution.com/CS/forums/thread/562.aspx
>
> http://www.cyclingutah.com/april/april99/classic.html
>
Dear Tim,
Since no one else has mentioned it, this was one of the more
interesting passages in the Pino article:
"This bike was also left side drive. The reasons: track races are
ridden counter-clockwise, and putting the crank on the left allowed
the right crankarm to be closer to the frame creating more room for
banking clearance."
"Also, tracks are circular, therefore by placing the extra weight of
the crank to the inside of the circle, a rider pushes the extra weight
a shorter distance each lap than if it were on the right, or outside,
of the circle. 'That’s it!' as one writer put it, 'Pino is about
splitting hairs that nobody else even sees.'"
http://www.cyclingutah.com/april/april99/classic.html
Now I wonder if track racers with right side drive are giving up a
potential advantage and losing the race before it begins?
The left-side drive bike:
http://www.wooljersey.com/gallery/v/broderir/Catalogs-Posters/Pino-Morroni/1974-12/Page+04.jpg.html?g2_imageViewsIndex=1
or http://tinyurl.com/bpke3
And no matter what its aerodynamics were, this would have been fun to
ride around the block:
http://www.classicrendezvous.com/USA/pino/pinos_bulb.htm
Cheers,
Carl Fogel
bjwe...@gmail.com
> On Feb 20, 11:49 pm, "b...@mambo.ucolick.org" <bjwei...@gmail.com>
> wrote:
>
> > I think what I demonstrated is that a very small
> > difference in rolling resistance, which requires
> > painstaking efforts to measure, can have
> > non-negligible effect.
>
> Did you read my re-do of your calculations, where I used tires that
> actually fit the OP's question? What I did was change just one
> factor, the tire size, while using the same brand and model.
>
> The upshot was, if we do look at one change in isolation, the
> difference the OP asked about was indeed negligible. Not the (IIRC) 9
> seconds or so that you claimed, but closer to one second.
> Furthermore, I pointed out why even that one second would be reduced,
> because it was actually based on the fact that the _wider_ tire had
> less rolling resistance. That is, the wider, heavier, less aero tire
> would have shortcomings to work against the Crr benefit.
Frank, you wrote
http://groups.google.com/group/rec.bicycles.tech/msg/84e006ffea2f1bb1
> I know you're writing to Carl, but (to use your phrasing): What I've
> said is that something that has a tiny or extremely difficult to
> measure effect has negligible effect. IOW, any benefits would be lost
> in the noise. IOW, you could not pick their effect out of the race
> results. IOW, any scientific analysis would say "Not worth it."
You generalized the argument here and elsewhere
beyond the OP's question, to a general question of
whether small changes can affect race results.
You didn't quantify detectability. This is
when I started picking on you. You had an earlier
answer to the OP where you said something like there's
no meaningful speed difference between 23 and 25
for most recreational users. I never argued with that,
in fact my intuition tells me it's true. Crr tests appear
to bear out that intuition. For race results, finer hairs
can be split.
So I demonstrated that a Crr effect in a race could
be not negligible, but potentially on the order of 2%
speed or power (linearly proportional since a hillclimb).
It was 1.9% for the tires I chose, which were _not_ cherry
picked to be the best and worst (Carl graciously retracted
that, but you didn't). Your response was to go back to the
OP's question. I took this to be exchanging an
argument you couldn't win on the numbers for one
you could.
> And of course, your calculation method assumes neither rider drafts
> anyone, not even each other. I went further to take that into
> account, and show yet another reason the small change that started
> this discussion is indeed, negligible.
Of course, if the rider with slower tires only keeps up by
drafting behind, he'll still come in second. But that's just
my little joke. More seriously, consider that I assumed total
power 300 watts, found that air drag took 11 watts, and
the tire difference was 1.9% out of the remaining 289 watts,
or 5.5 watts. So the Crr difference was half of the total
aero drag. You can't save 50% of drag by drafting; even at
high speed the best estimates are around 30%. The only
reason you got it to come out to 4% of drag was that you were
using your redefined version of the problem as a small
23 vs 25mm tire difference, rather than the plausible range
of Crr I used. Repeating your assertion without pointing out
that it applies to your problem rather than mine is
intellectually dishonest.
> BTW, it occurs to me that many - probably most - of those reading this
> are past their racing prime. We are never going to take another first
> place. (Hell, best I ever did was second place.) But let's think for
> a minute about the young kids who might be really good someday. Do we
> want to be telling them to train and learn tactics? Or do we want to
> be telling them that they can never win if they ignore any tiny - and
> expensive - potential equipment advantage?
Oh noes, think of the children!!!
When someone makes this argument - that we're
setting a bad example for impressionable youth -
it's generally a sign they have no better argument.
I don't imagine that you look favorably upon it in
h*lmet threads.
Nobody seriously thinks that obsessing over tires or
CdA is a substitute for training smart. However, if you
have already trained smart, and are not an over the hill
Masters Fattie, you might want to think about the details.
It appears that the details, at a high level of competition,
do matter.
I also would still like to know which of the following my
calculation was:
> > > Well, that, plus they're wildly overstating their positions, using
> > > either inappropriate absolute statements, or giving TT examples of
> > > wind tunnel finger positions, or retreating to "hahaha you just don't
> > > understand."
Ben
John Forrest Tomlinson
>On Feb 21, 9:37 am, Frank Krygowski <frkry...@gmail.com> wrote:
>> BTW, it occurs to me that many - probably most - of those reading this
>> are past their racing prime. We are never going to take another first
>> place. (Hell, best I ever did was second place.) But let's think for
>> a minute about the young kids who might be really good someday. Do we
>> want to be telling them to train and learn tactics? Or do we want to
>> be telling them that they can never win if they ignore any tiny - and
>> expensive - potential equipment advantage?
>
>Oh noes, think of the children!!!
Ben, I missed that bit from Frank but have to thank you for quoting it
so I can pile on.
Wow Frank, you're really stretching in an attempt to "win" your
argument. Is anyone here saying a racer can never win if they ignore
any tiny advantage? I haven't heard it.
Has anyone here ever said don't train or learn tactics: obsess about
equipment.
You're flailing. It's hilarious. I'll go all high-school and rude and
say hahahahaha.
Frank Krygowski
wrote:
> >On Feb 21, 9:37 am, Frank Krygowski <frkry...@gmail.com> wrote:
> >> BTW, it occurs to me that many - probably most - of those reading this
> >> are past their racing prime. We are never going to take another first
> >> place. (Hell, best I ever did was second place.) But let's think for
> >> a minute about the young kids who might be really good someday. Do we
> >> want to be telling them to train and learn tactics? Or do we want to
> >> be telling them that they can never win if they ignore any tiny - and
> >> expensive - potential equipment advantage?
>
> >Oh noes, think of the children!!!
>
> Ben, I missed that bit from Frank but have to thank you for quoting it
> so I can pile on.
>
> Wow Frank, you're really stretching in an attempt to "win" your
> argument. Is anyone here saying a racer can never win if they ignore
> any tiny advantage? I haven't heard it.
??? Really? You haven't heard it?
Have you read none of Michael Press's posts? "A racer who dismisses
any potential advantage as negligible has already lost the race"?
He's repeated that many times.
Your responses have been much less specific, tending toward "you just
don't understand racing" or "hahahahaha." IOW, not enough technical
content to bother with!
- Frank Krygowski
Frank Krygowski
wrote:
> On Feb 21, 9:37 am, Frank Krygowski <frkry...@gmail.com> wrote:
>
>
>
> > On Feb 20, 11:49 pm, "b...@mambo.ucolick.org" <bjwei...@gmail.com>
> > wrote:
>
> > > I think what I demonstrated is that a very small
> > > difference in rolling resistance, which requires
> > > painstaking efforts to measure, can have
> > > non-negligible effect.
>
> > Did you read my re-do of your calculations, where I used tires that
> > actually fit the OP's question? What I did was change just one
> > factor, the tire size, while using the same brand and model.
>
> > The upshot was, if we do look at one change in isolation, the
> > difference the OP asked about was indeed negligible. Not the (IIRC) 9
> > seconds or so that you claimed, but closer to one second.
> > Furthermore, I pointed out why even that one second would be reduced,
> > because it was actually based on the fact that the _wider_ tire had
> > less rolling resistance. That is, the wider, heavier, less aero tire
> > would have shortcomings to work against the Crr benefit.
>
> beyond the OP's question, to a general question of
> whether small changes can affect race results.
> You didn't quantify detectability. This is
> when I started picking on you. You had an earlier
> answer to the OP where you said something like there's
> no meaningful speed difference between 23 and 25
> for most recreational users. I never argued with that,
> in fact my intuition tells me it's true.
OK, we have some agreement.
> For race results, finer hairs
> can be split.
>
> So I demonstrated that a Crr effect in a race could
> be not negligible, but potentially on the order of 2%
> speed or power (linearly proportional since a hillclimb).
> It was 1.9% for the tires I chose, which were _not_ cherry
> picked to be the best and worst (Carl graciously retracted
> that, but you didn't). Your response was to go back to the
> OP's question. I took this to be exchanging an
> argument you couldn't win on the numbers for one
> you could.
Interesting. I took your tactic, of grabbing a "quite slow" tire vs.
a "quite fast" tire, to be an attempt at changing the subject by
inflating the differences.
(One begins to see why the Israeli-Palestinian problem is so
difficult.)
> > And of course, your calculation method assumes neither rider drafts
> > anyone, not even each other. I went further to take that into
> > account, and show yet another reason the small change that started
> > this discussion is indeed, negligible.
>
> Of course, if the rider with slower tires only keeps up by
> drafting behind, he'll still come in second.
<sigh> Not at all! Given the choice, I'd _prefer_ to be just behind
a rival when I'm within, oh, 50 yards of the finish line.
> More seriously, consider that I assumed total
> power 300 watts, found that air drag took 11 watts, and
> the tire difference was 1.9% out of the remaining 289 watts,
> or 5.5 watts. So the Crr difference was half of the total
> aero drag. You can't save 50% of drag by drafting; even at
> high speed the best estimates are around 30%.
You need to look for the results when you're behind not just one
person, but more than one - which is the case when there's a peloton
or most breakaways.
> The only
> reason you got it to come out to 4% of drag was that you were
> using your redefined version of the problem as a small
> 23 vs 25mm tire difference, rather than the plausible range
> of Crr I used. Repeating your assertion without pointing out
> that it applies to your problem rather than mine is
> intellectually dishonest.
But how is returning to the original question dishonest? Or to turn
it around, why didn't you put one of your calculation riders on
mountain bike tires?
Look, Ben, nobody is saying it's impossible for a BIG enough change to
affect a difference in the right conditions. I don't think for a
minute that LeMond would have beaten Fignon without the aero bar -
although, once again, that was a time trial.
If you take your average amateur Joe Racer, there are guys who could
beat him in a road race by riding a mountain bike with knobbies.
There are far, far more guys who could beat Joe even if they were on
the "slow" tires you chose and Joe was on the "fast" tires you chose.
But there is nobody that will consistently beat Joe because of a 5%
difference in Crr. Ditto a 2 ounce weight difference, or an aero
water bottle.
Because, once again, it's a probabilistic problem. Re-read the dice
analogy I gave. There are too many other factors, many of them very
random.
Analytic Cycling doesn't even _have_ a way to input number of riders
surrounding you in a pack, position in the peloton when a breakaway
occurs, variation in road roughness across the lane, whether one is
drinking water when someone attacks, etc., let alone time spent doing
interval training or tactics practicing. IOW, it imagines a sterile
world, much more sterile even than a time trial.
You can focus on large changes, the way Robert Chung focuses on time
trials. For those cases, I'll say yes, you have a higher probability
of spotting a difference in the race results. But even then, it's a
probability, not a certainty. If that were not so, nobody would
bother to actually race.
(Never mind, I retract that. Las Vegas proves millions of people will
bet heavily against the obvious.)
>
> > BTW, it occurs to me that many - probably most - of those reading this
> > are past their racing prime. We are never going to take another first
> > place. (Hell, best I ever did was second place.) But let's think for
> > a minute about the young kids who might be really good someday. Do we
> > want to be telling them to train and learn tactics? Or do we want to
> > be telling them that they can never win if they ignore any tiny - and
> > expensive - potential equipment advantage?
>
> Oh noes, think of the children!!!
>
> When someone makes this argument - that we're
> setting a bad example for impressionable youth -
> it's generally a sign they have no better argument.
> I don't imagine that you look favorably upon it in
> h*lmet threads.
>
> Nobody seriously thinks that obsessing over tires or
> CdA is a substitute for training smart.
Do you think we'd have more, or fewer, young novices if we quashed the
idea that you must have a new, high tech bike to be competitive?
Frankly, it doesn't matter to me. I'd be just as happy seeing kids
riding for fun and utility rather than racing. But those who want
racing to grow should think about that. I have little doubt that the
average inner-city poor kid could ride the wheels off the average
suburban prince or princess, if given just a few hours coaching and a
fifteen-year-old road bike.
- Frank Krygowski
John Forrest Tomlinson
<frkr...@gmail.com> wrote:
>Your responses have been much less specific, tending toward "you just
>don't understand racing" or "hahahahaha." IOW, not enough technical
>content to bother with!
Not technical enough to bother with, but I have said things that you
have (laughably) disputed, such as that tiny differences in bikes can
matter, and you've said not true.
I've said that aerodynamics matter a lot in a race, even while riding
in a pack, and you've disputed that in a variety of ways.
And more.
No need to repeat, by me, unless you revise, in which case I'll call
you out.
John Forrest Tomlinson
<frkr...@gmail.com> wrote:
>
>But how is returning to the original question dishonest?
It's dishonest when you posit it as an argument with someone, in a
attempt to show you are right and the other guy is wrong, when the
other person isn't arguing with you about that.
PS: It's lame too.
Tim McNamara
John Forrest Tomlinson <usenet...@jt10000.com> wrote:
> On Mon, 23 Feb 2009 02:13:15 -0800 (PST), "b...@mambo.ucolick.org"
> <bjwe...@gmail.com> wrote:
>
> >On Feb 21, 9:37 am, Frank Krygowski <frkry...@gmail.com> wrote:
>
> >> BTW, it occurs to me that many - probably most - of those reading
> >> this are past their racing prime. We are never going to take
> >> another first place. (Hell, best I ever did was second place.)
> >> But let's think for a minute about the young kids who might be
> >> really good someday. Do we want to be telling them to train and
> >> learn tactics? Or do we want to be telling them that they can
> >> never win if they ignore any tiny - and expensive - potential
> >> equipment advantage?
> >
> >Oh noes, think of the children!!!
>
> Ben, I missed that bit from Frank but have to thank you for quoting
> it so I can pile on.
>
> Wow Frank, you're really stretching in an attempt to "win" your
> argument. Is anyone here saying a racer can never win if they ignore
> any tiny advantage? I haven't heard it.
> > In article <rubrum-BC5A63....@news.sf.sbcglobal.net>,
> > Michael Press <rub...@pacbell.net> wrote:
> >
> > > 5) A racer who dismisses a potential advantage as negligible has
> > > already lost race.
You seem to be missing a lot of posts, JT. Perhaps you should contact
your ISP to make sure their NNTP server is functioning correctly and
not dropping posts out of the spool.
Michael Press
<805447cb-6597-440b...@o36g2000yqh.googlegroups.com>,
Frank Krygowski <frkr...@gmail.com> wrote:
> On Feb 23, 6:50 am, John Forrest Tomlinson <usenetrem...@jt10000.com>
> wrote:
> > On Mon, 23 Feb 2009 02:13:15 -0800 (PST), "b...@mambo.ucolick.org"
> >
> > <bjwei...@gmail.com> wrote:
> > >On Feb 21, 9:37 am, Frank Krygowski <frkry...@gmail.com> wrote:
> > >> BTW, it occurs to me that many - probably most - of those reading this
> > >> are past their racing prime. We are never going to take another first
> > >> place. (Hell, best I ever did was second place.) But let's think for
> > >> a minute about the young kids who might be really good someday. Do we
> > >> want to be telling them to train and learn tactics? Or do we want to
> > >> be telling them that they can never win if they ignore any tiny - and
> > >> expensive - potential equipment advantage?
> >
> > >Oh noes, think of the children!!!
> >
> > Ben, I missed that bit from Frank but have to thank you for quoting it
> > so I can pile on.
> >
> > Wow Frank, you're really stretching in an attempt to "win" your
> > argument. Is anyone here saying a racer can never win if they ignore
> > any tiny advantage? I haven't heard it.
>
> ??? Really? You haven't heard it?
>
> Have you read none of Michael Press's posts? "A racer who dismisses
> any potential advantage as negligible has already lost the race"?
Not the same, is it?
--
Michael Press
bjwe...@gmail.com
> On Feb 23, 5:13 am, "b...@mambo.ucolick.org" <bjwei...@gmail.com>
> wrote:
>
> > So I demonstrated that a Crr effect in a race could
> > be not negligible, but potentially on the order of 2%
> > speed or power (linearly proportional since a hillclimb).
> > It was 1.9% for the tires I chose, which were _not_ cherry
> > picked to be the best and worst (Carl graciously retracted
> > that, but you didn't). Your response was to go back to the
> > OP's question. I took this to be exchanging an
> > argument you couldn't win on the numbers for one
> > you could.
>
> Interesting. I took your tactic, of grabbing a "quite slow" tire vs.
> a "quite fast" tire, to be an attempt at changing the subject by
> inflating the differences.
>
> (One begins to see why the Israeli-Palestinian problem is so
> difficult.)
Franklin Mint,
Your conversation becomes tiresome, as Mike
Myers's character Dieter says.
I explicitly documented what I was doing. Not answering
the OP's question, but the question of whether small
differences in tires could make a race-significant
difference. The two tires spanned 60% of the range of tires
in Al's table of RR tests - and his tests were mostly pretty
lightweight racing tires.
I also tilted the problem against finding a difference
by considering only the last 1.5 miles of a 40 mile race.
The actual extra effort a slow tire costs over the whole
race is more, but I wanted to keep the example clear.
Your reference to the Israeli-Palestinian problem
is inflammatory and the jury will disregard it.
> > Of course, if the rider with slower tires only keeps up by
> > drafting behind, he'll still come in second.
>
> <sigh> Not at all! Given the choice, I'd _prefer_ to be just behind
> a rival when I'm within, oh, 50 yards of the finish line.
It's a 7 percent uphill grade, so distance is more important
than drafting. If you're behind, that's about 3 extra yards
you have to pedal out of 50. A disadvantage of 6%.
Aero drag was 11 watts out of 300, or 3.7%. If you save
_all_ the aero drag by drafting (impossible), you're still
losing on the extra distance. Does your innumeracy
frequently lose you bets?
> > More seriously, consider that I assumed total
> > power 300 watts, found that air drag took 11 watts, and
> > the tire difference was 1.9% out of the remaining 289 watts,
> > or 5.5 watts. So the Crr difference was half of the total
> > aero drag. You can't save 50% of drag by drafting; even at
> > high speed the best estimates are around 30%.
>
> You need to look for the results when you're behind not just one
> person, but more than one - which is the case when there's a peloton
> or most breakaways.
Have you ever ridden up a hill? On a steep hill
finish, you don't generally see a pack crossing the
line together. Most numbers for power saved by drafting
come from riding at high speeds rather than slogging
up a 7% hill in a Fat Masters race, where I suspect the
percentage of aero drag saved is very small.
Again, you are trying to change the problem to win an
argument.
>
> > The only
> > reason you got it to come out to 4% of drag was that you were
> > using your redefined version of the problem as a small
> > 23 vs 25mm tire difference, rather than the plausible range
> > of Crr I used. Repeating your assertion without pointing out
> > that it applies to your problem rather than mine is
> > intellectually dishonest.
>
> But how is returning to the original question dishonest? Or to turn
> it around, why didn't you put one of your calculation riders on
> mountain bike tires?
>
> Look, Ben, nobody is saying it's impossible for a BIG enough change to
> affect a difference in the right conditions. I don't think for a
> minute that LeMond would have beaten Fignon without the aero bar -
> although, once again, that was a time trial.
Quantify "big enough." That was what I asked you
way back when.
Lemond used not only aero bars, but an early aero helmet.
Did the aero hat make 8 seconds worth of difference? Some
people think so, but others think it might actually not have
helped. I don't know if Lemond tested it with/without.
Fignon had access to aero bars and (AFAIK) aero hat,
but did not use them. Perhaps this illustrates Michael
Press's adage about dismissing potential advantages
as negligible. I once heard Jim Gentes say that he had
had to fly back from that Tour before seeing the last day,
but if he had known Fignon was going to ride bareheaded,
he would have stayed. (Just to keep this on track, he was
talking about aero benefits rather than foam hat safety
issues, of course.)
> If you take your average amateur Joe Racer, there are guys who could
> beat him in a road race by riding a mountain bike with knobbies.
> There are far, far more guys who could beat Joe even if they were on
> the "slow" tires you chose and Joe was on the "fast" tires you chose.
> But there is nobody that will consistently beat Joe because of a 5%
> difference in Crr. Ditto a 2 ounce weight difference, or an aero
> water bottle.
>
> Because, once again, it's a probabilistic problem. Re-read the dice
> analogy I gave. There are too many other factors, many of them very
> random.
You don't understand probabilistic problems. You are
conflating random and systematic effects. If Joe and
Bob are evenly matched, and Joe has a 2% advantage
on uphills due to Crr, then sometimes Bob will win because
Joe didn't get enough sleep, or Bob had better team tactics,
or whatever. But on the average, Joe will win more. If
they had the same tires but Bob had a 2% advantage in
power to weight ratio, Bob will win more, and you'd
point out that training better is more important than
fussing over tires. Which of course it is in the long run;
but improving your power by even 2% takes a fair amount
of work. It isn't something to brush off.
> Analytic Cycling doesn't even _have_ a way to input number of riders
> surrounding you in a pack, position in the peloton when a breakaway
> occurs, variation in road roughness across the lane, whether one is
> drinking water when someone attacks, etc., let alone time spent doing
> interval training or tactics practicing. IOW, it imagines a sterile
> world, much more sterile even than a time trial.
It imagines a simplified world. This is how one
does physics problems, for example, you analyze a
still-wind case before trying to understand side-winds.
The rest of the stuff like keeping good position is
up to the racer. Staying alert in a pack is part
of preparation - but so is making sure that you
understand the benefits and limits of your
training and equipment.
> You can focus on large changes, the way Robert Chung focuses on time
> trials. For those cases, I'll say yes, you have a higher probability
> of spotting a difference in the race results. But even then, it's a
> probability, not a certainty. If that were not so, nobody would
> bother to actually race.
>
> (Never mind, I retract that. Las Vegas proves millions of people will
> bet heavily against the obvious.)
You talk of things which refute your point.
Las Vegas makes millions by taking only a small
advantage on each bet. If the house won every
game, no one would play. But over time, the
house always wins, and so does the person with
the 2% advantage.
> > Oh noes, think of the children!!!
>
> > When someone makes this argument - that we're
> > setting a bad example for impressionable youth -
> > it's generally a sign they have no better argument.
> > I don't imagine that you look favorably upon it in
> > h*lmet threads.
>
> > Nobody seriously thinks that obsessing over tires or
> > CdA is a substitute for training smart.
>
> Do you think we'd have more, or fewer, young novices if we quashed the
> idea that you must have a new, high tech bike to be competitive?
>
> Frankly, it doesn't matter to me. I'd be just as happy seeing kids
> riding for fun and utility rather than racing. But those who want
> racing to grow should think about that. I have little doubt that the
> average inner-city poor kid could ride the wheels off the average
> suburban prince or princess, if given just a few hours coaching and a
> fifteen-year-old road bike.
How do you know the average inner-city kid would be
better? Should you consider weight, diet, amount of exercise,
the fact that suburban princes and princesses have been
running up and down soccer fields while kids in really bad
neighborhoods are kept inside by their parents? Point is,
neither you nor I know, but there are societal factors
that keep kids off bikes that are much bigger and harder to
deal with than the cost of a starter road bike.
I don't think anybody here has said you _need_ a high tech
bike to be competitive. Another strawman. I'm sure you can
find some dumbass somewhere who says that, but I've
also seen plenty of kids (and some adults) get their
start racing on hand-me-downs or old bikes. Go watch the
variety in the beginner class of an MTB race sometime -
even though there are actual technological differences
between generations of MTBs, much more so than road
bikes.
If you race on a semi-regular basis, you realize that time
and other stuff like gas money are at least as much of an
issue as equipment cost.
Ben
Robert Chung
> It's a 7 percent uphill grade, so distance is more important
> than drafting. If you're behind, that's about 3 extra yards
> you have to pedal out of 50. A disadvantage of 6%.
> Aero drag was 11 watts out of 300, or 3.7%. If you save
> _all_ the aero drag by drafting (impossible), you're still
> losing on the extra distance. Does your innumeracy
> frequently lose you bets?
A rec.bike participant who dismisses calculations as being negligibly
important has already lost.
Frank Krygowski
wrote:
>
> Your conversation becomes tiresome....
... and you and I seem to be into "infinitely increasing post
length." So I'll trim a lot.
> Again, you are trying to change the problem to win an
> argument.
Again, that's an odd thing to say when I "changed" to the original
problem!
> > Look, Ben, nobody is saying it's impossible for a BIG enough change to
> > affect a difference in the right conditions. I don't think for a
> > minute that LeMond would have beaten Fignon without the aero bar -
> > although, once again, that was a time trial.
>
> Quantify "big enough." That was what I asked you
> way back when.
It's been answered several times, several ways. I'll try again,
though.
Whether an equipment change benefit is "big enough" to significantly
affect most race results cannot be answered in specific numerical
terms. That's BECAUSE race results are not governed by simple
deterministic effects. (Time trials come closer, but even then, there
are enough factors at play that the word "probably" or "possibly"
needs to be inserted in front of any predicted benefit.)
I hope we agree that there are huge numbers of factors, including
totally random factors, that affect a road race placement. (Do you
disagree?) Given that, then any guesses that anybody gives on (say)
necessary percentage bike+rider weight reduction, percentage aero Cd
reduction, percentage Crr reduction, etc. are simply guesses. And in
fact, they'd need to be stated in probabilistic terms - like, "A five
pound weight reduction has a 10% +/- 3% chance of increasing you race
placement by one position."
But I emphasize, any such answer would be no more than a guess. And
the answers cannot be verified except by running many, many controlled
tests, where all other factors are kept constant. That's an obvious
impossibility.
I've explained this using the dice-shaving analogy. It's not perfect,
but I think it should get the idea across. You can re-read it if
necessary, and try to say why it isn't a correct representation of the
situation.
> > If you take your average amateur Joe Racer, there are guys who could
> > beat him in a road race by riding a mountain bike with knobbies.
> > There are far, far more guys who could beat Joe even if they were on
> > the "slow" tires you chose and Joe was on the "fast" tires you chose.
> > But there is nobody that will consistently beat Joe because of a 5%
> > difference in Crr. Ditto a 2 ounce weight difference, or an aero
> > water bottle.
>
> > Because, once again, it's a probabilistic problem. Re-read the dice
> > analogy I gave. There are too many other factors, many of them very
> > random.
>
> You don't understand probabilistic problems.
:-) That's another way of saying "You're just wrong."
> You are
> conflating random and systematic effects. If Joe and
> Bob are evenly matched, and Joe has a 2% advantage
> on uphills due to Crr, then sometimes Bob will win because
> Joe didn't get enough sleep, or Bob had better team tactics,
> or whatever. But on the average, Joe will win more.
Your final sentence is correct, but practically meaningless. IOW, I
could turn the tables and ask you to quantify "will win more."
But I think we can both agree that if Bob has a 2% advantage in Crr,
Bob will _certainly_ not beat Joe 2% more often. You don't think that
he will, do you?
He won't because Crr is not the only thing that affects race
placement. In fact, Crr and Cd and weight are not the only things.
They are not even the most important things, unless someone sets up a
completely ridiculous special case, like riding a 40 pound bike with
knobby tires while wearing a sandwich signboard. 2% drop in Crr will
not affect his placement 1/2% of the time. In a racing season, you'd
never detect that difference from looking at results.
> > Analytic Cycling doesn't even _have_ a way to input number of riders
> > surrounding you in a pack, position in the peloton when a breakaway
> > occurs, variation in road roughness across the lane, whether one is
> > drinking water when someone attacks, etc., let alone time spent doing
> > interval training or tactics practicing. IOW, it imagines a sterile
> > world, much more sterile even than a time trial.
>
> It imagines a simplified world. This is how one
> does physics problems, for example, you analyze a
> still-wind case before trying to understand side-winds.
Just a side note: That's also how you do engineering problems; but
engineers are trained not to oversimplify, because being simplistic
won't get the job done. Physicists are sometimes not clear on that.
Anecdote alert: For a while, I worked as a plant engineer at a small
holography company. Producing commercial holograms is an exacting
business. The theory is weird enough, but the practicalities are
tricky indeed - there's much that can go wrong, some of it on a very
small scale, some of it requiring extremely careful process control,
some of it requiring extremely sophisticated machinery.
So one day we were visited by a physics professor. I introduced
myself, and in conversation said something like "Yes, what goes on
here is pretty complicated." He assumed a look of superiority and
said "Oh, it's just a Fourier transform!"
Sure - right. But you don't even have to understand Fourier
transforms to produce a good hologram. To do them as we did, you do
need to understand lasers, optics, process environmental control,
electronic and computer control of complicated machines, geometry,
machining, plating chemistry, etc. etc. (Not to mention economics,
business, OSHA...) There's a LOT of stuff beyond the math theory. But
that guy was interested only in the theory. He could never make a
hologram.
Likewise, in a road race, there's a LOT of stuff beyond "who has the
fastest tires." Changing that will change one tiny portion of the
factors. In any reasonable situation, it will never show up as the
factor that made someone win a race.
. . .
I've suggested before, the crowd that thinks any improvment is
important could try random trials with hidden 2 ounce weight
differences, switching weights between two equally matched racers, and
keeping records through a race season. I predict the small weight
difference will be undetectable in the results.
Or if you want to do more "Analytic Cycling" work, try modeling an
entire race. Use two guys with identical power, one with 2% less
Crr. But do it with, say, 20 other racers, and use dice or other
random devices to account for the positions and moves of all the other
racers. Account for the time Joe is in the middle of the pack, vs. in
front; and the time he's on the rougher section of road, not the
smoother; and the time he's behind the bigger guy who jumps, instead
of behind the smaller guy who doesn't, and so on. Add in about 200
other random events and see what you come up with. But keep in mind,
even those results will be oversimplified.
- Frank Krygowski
carl...@comcast.net
Dear Frank,
Andrew Muzi mentioned "200% of Nothing" earlier.
Inter-library loan came up with a copy.
Here's an interesting comment on drug testing and false positives:
"The general rule is simple: The lower the background level of the
substance being tested for, the more likely that a test of a given
accuracy will fail in the positive direction."
--p.117
In cruder terms, the harder it is to test (or measure) for something,
the more likely it is that the test (or measurement) will produce a
false positive.
In other words, the smaller the theoretical improvement we try to
measure, the more likely we are to come up with over-optimistic
predictions.
I seem to recall that in real bicycyling, predicted effects (good or
bad) often turn out to be noticeably less than expected. Robert works
hard to confirm his theoretical models, so I expect that they're
better than some, but a lot of theories predicting specific effects
turn out to be over-estimates.
Cheers,
Carl Fogel
A Muzi
I bought it when it was published and enjoyed it immensely. Glad you
liked it.
Nick L Plate
On 24 Feb, 20:00, Frank Krygowski <frkry...@gmail.com> wrote:
>
> Likewise, in a road race, there's a LOT of stuff beyond "who has the
> fastest tires." Changing that will change one tiny portion of the
> factors. In any reasonable situation, it will never show up as the
> factor that made someone win a race.
>
Only the correct combination between rim, spokes and tyre can give the
'fastest ' wheel. A larger section tyre will permit the use of a
lighter more flexible rim. A very narrow hard tyre has a short
contact patch. A heavier rider on hard,narrow tyres will distort the
arch of the rim so as to effectively reduce the wheel radius. The
smaller effective wheel radius increases rolling resistance. Tyre and
rim choice are interdependent unless your a fairy.
TJ
carl...@comcast.net
wrote:
>carl...@comcast.net wrote:
>> On Tue, 24 Feb 2009 12:00:55 -0800 (PST), Frank Krygowski
>> <frkr...@gmail.com> wrote:
[snip]
Dear Andrew,
Curse you! Are you in the pay of the used booksellers? Do you swap
them presta valve caps for literature?
Your cheerful soft-sell recommendation just made me look at the book
again, peek at www.bookfinder.com, and--
Bang! There goes another $0.05, plus $3.99 shipping for a used copy of
Dewdney's "200% of Nothing". That used-book site is a menace!
Meanwhile, a pleasant Australian University took my credit card,
charged me eight dollary-doos US, assured me that they'd email me
scans of the turn-of-the-century bicycle wheel test from an obscure
engineering journal--
And nothing has arrived!
I feel like Bart Simpson vs. Australia:
http://www.snpp.com/episodes/2F13.html
Cheers,
Carl Fogel
John Forrest Tomlinson
I think tire width and rim width affect aerodynamics, and the
appropriate tire width is somewhat related to rider weight.
But I find it hard to beleive rider weight is changing the shape of
the rim. Do you have any more info or evidence on that?
A Muzi
>> carl...@comcast.net wrote:
>>> Andrew Muzi mentioned "200% of Nothing" earlier.
>>>
>>> Inter-library loan came up with a copy.
>>>
>>> Here's an interesting comment on drug testing and false positives:
>>>
>>> "The general rule is simple: The lower the background level of the
>>> substance being tested for, the more likely that a test of a given
>>> accuracy will fail in the positive direction."
>>> --p.117
>>>
>>> In cruder terms, the harder it is to test (or measure) for something,
>>> the more likely it is that the test (or measurement) will produce a
>>> false positive.
>>>
>>> In other words, the smaller the theoretical improvement we try to
>>> measure, the more likely we are to come up with over-optimistic
>>> predictions.
>>>
>>> I seem to recall that in real bicycyling, predicted effects (good or
>>> bad) often turn out to be noticeably less than expected. Robert works
>>> hard to confirm his theoretical models, so I expect that they're
>>> better than some, but a lot of theories predicting specific effects
>>> turn out to be over-estimates.
> A Muzi <a...@yellowjersey.org> wrote:
>> I bought it when it was published and enjoyed it immensely. Glad you
>> liked it.
carl...@comcast.net wrote:
> Curse you! Are you in the pay of the used booksellers? Do you swap
> them presta valve caps for literature?
>
> Your cheerful soft-sell recommendation just made me look at the book
> again, peek at www.bookfinder.com, and--
>
> Bang! There goes another $0.05, plus $3.99 shipping for a used copy of
> Dewdney's "200% of Nothing". That used-book site is a menace!
>
> Meanwhile, a pleasant Australian University took my credit card,
> charged me eight dollary-doos US, assured me that they'd email me
> scans of the turn-of-the-century bicycle wheel test from an obscure
> engineering journal--
>
> And nothing has arrived!
>
> I feel like Bart Simpson vs. Australia:
> http://www.snpp.com/episodes/2F13.html
Although I admit ("My name is Andrew, and I'm a reader") to feeding the
used-book monkey on my back, I bought that title at full retail after
reading the review. Whatever I overpaid for it, I 'saved' on a beautiful
hardbound set of Boswell's Life of Johnson. After Vol I, it seems ten
minutes of action packed into many long hours...
Nick L Plate
wrote:
No hard evidence. It is usual when an arch is loaded at a point
contact to distort, increasing the radius of the curve at the point of
contact. With the ends of the arch solidly restrained, there will be
a reduction of radius either side of the load. Without abutments the
arch will fail. The abutments on a bike wheel are the spokes and it
is the fore and aft spokes which prevent the arch from spreading and
collapsing.
There is an optimum amount of restraint required which will minimise
rim distortion for a given rim with a point load of known force. When
a pneumatic tyre is fitted, the force is spread across the length of
the tyre contact patch so loading the arch more evenly. This spread
of load mean that the point contact is eliminated and the rim
distortion is reduced for the same load. The longer the tyre patch
the less distortion.
A rim with lower vertical stiffness may be used with a longer
contact patch for equal rim distortion. I hope to investigate
precisely how spoke restraint and tyre sizing effect rim ditortion and
overall wheel efficiency. I faintly recall some newsreel footage of a
track rider showing his exploded tyre, his wheel had buckled. I was
thinking that this was a case where the wheel was strong enough with
the tyre inflated, but immediately the tyre had collapsed, the wheel
followed. It's seems that the strength of the wheel was below the
riders weight for a point contact, but above his weight while the tyre
was inflated.
TJ
Michael Press
<fa370054-cfef-4a46...@u38g2000yqe.googlegroups.com>,
Frank Krygowski <frkr...@gmail.com> wrote:
> Just a side note: That's also how you do engineering problems; but
> engineers are trained not to oversimplify, because being simplistic
> won't get the job done. Physicists are sometimes not clear on that.
Physicists do a lot of approximation; they stack approximation
upon approximation. Seriously. Go to Landau and Lifschitz,
Theory of Elasticity and look at how much approximation
they do before they get to some hair differential equations
that need further approximation to solve. Finally their
results get to the engineers. Thing is, those physicists
are good at it. EM theory: antennae, interaction of matter
and EM, optics. More approximatons. Condensed matter, crytals.
How do we get any usable answeres out of quantum theory?
Show me where physicists go astray.
--
Michael Press
carl...@comcast.net
wrote:
>-snip-
Dear Andrew,
Well, no, "action" isn't what comes to mind with the later Augustan
authors.
"Terminator 4" probably won't include an exciting chase scene in which
a cyborg assassin pursues a sickly self-described harmless drudge who
spent ten years producing a dictionary that's more fun for the
scholarly than for normal carbon-based life-forms:
A DICTIONARY of the English Language: in which The WORDS are deduced
from their ORIGINALS, and ILLUSTRATED in their DIFFERENT
SIGNIFICATIONS by EXAMPLES from the best WRITERS. To which are
prefixed, A HISTORY of the LANGUAGE, and AN ENGLISH GRAMMAR.
By SAMUEL JOHNSON, A.M.
In TWO Volumes
(Weak plot, poor character development, many spelling mistakes--it's
hard to read all the way to the end just to see what happens.)
:-)
Cheers,
Carl Fogel
Frank Krygowski
> In article
> <fa370054-cfef-4a46-a832-22b229ecf...@u38g2000yqe.googlegroups.com>,
> Frank Krygowski <frkry...@gmail.com> wrote:
>
> > Just a side note: That's also how you do engineering problems; but
> > engineers are trained not to oversimplify, because being simplistic
> > won't get the job done. Physicists are sometimes not clear on that.
>
> Physicists do a lot of approximation; they stack approximation
> upon approximation. Seriously. Go to Landau and Lifschitz,
> Theory of Elasticity and look at how much approximation
> they do before they get to some hair differential equations
> that need further approximation to solve. Finally their
> results get to the engineers. Thing is, those physicists
> are good at it. EM theory: antennae, interaction of matter
> and EM, optics. More approximatons. Condensed matter, crytals.
> How do we get any usable answeres out of quantum theory?
> Show me where physicists go astray.
Already ansered, I think. For example, that physicist went astray
when he implied that producing holograms is not complicated - that
"it's just a Fourier transform." Dismissing all complication, all
practicalities, is going astray.
And we see that here when (for example) someone implies that a 3 gram
drag reduction in a wind tunnel really will lead to a two-second
improvement in a race time.
- Frank Krygowski
bjwe...@gmail.com
>
> Already ansered, I think. For example, that physicist went astray
> when he implied that producing holograms is not complicated - that
> "it's just a Fourier transform." Dismissing all complication, all
> practicalities, is going astray.
Don't confuse personal arrogance of some guy you
met with whether an analytic approach, as frequently
used in physics, is valid or not.
> And we see that here when (for example) someone implies that a 3 gram
> drag reduction in a wind tunnel really will lead to a two-second
> improvement in a race time.
How do you know it won't? You're awfully certain
about a lot of things. In the long run, reducing drag by
a small amount is going to make riders' average times
slower, faster, or no change. Which do you expect and
why? Sure, on any given Sunday, in any one race, it
will be impossible to measure whether a two second
difference actually happened. But over many trials,
do you really think it would average to zero? How
big does an effect have to be to not average to zero?
My time up local hills in the 8-10 minute range is
generally consistent to 5%, unless there's an insane
headwind or I'm sick or something. If I dropped 1-2%
of weight (off bike or body), it would take averaging
quite a few rides to detect the effect, but do you
think it would not happen?
Ben
Frank Krygowski
wrote:
> On Feb 24, 10:07 pm, Frank Krygowski <frkry...@gmail.com> wrote:
>
>
> > And we see that here when (for example) someone implies that a 3 gram
> > drag reduction in a wind tunnel really will lead to a two-second
> > improvement in a race time.
>
> How do you know it won't? You're awfully certain
> about a lot of things.
The more one studies, the more one learns. After a certain amount of
learning, one does become certain about a lot of things.
> In the long run, reducing drag by
> a small amount is going to make riders' average times
> slower, faster, or no change. Which do you expect and
> why?
In what situation? In a road race? A "small" amount of drag
reduction will cause no detectable improvement, and no _practical_
change. It will not increase your placement, which is the objective.
And even in a TT, the tiny "finger drag" advantage Chung pointed to
will not have a detectable benefit. IOW, after running a dozen time
trials, six with finger postion A and six with finger position B,
nobody will see the supposed two second difference in the times. Or
in the mean times. Or in the mean times with standard deviations
calculated. That's my bet, based on what I know about such things.
> Sure, on any given Sunday, in any one race, it
> will be impossible to measure whether a two second
> difference actually happened. But over many trials,
> do you really think it would average to zero? How
> big does an effect have to be to not average to zero?
To detect an aero benefit of a few percent Cd reduction (far larger
than what Chung was discussing) I think that it would take thousands
of otherwise identical races.
Have you ever delved into "propagation of error" in physical
measurements? Briefly, accuracy of a measuring device typically
involves many sources of potential error. For example, a scale may
employ linkages, a flexible element (like a spring or beam), and
display elements (either mechanical or electronic), each of which may
be a bit off from their design value.
To predict the precision and accuracy of a scale's design, one needs
to assess the expected error in each pertinent part. For example, the
stiffness of a beam (for scales employing strain gages on flexible
beams) depends on the width, length and depth of the beam, each of
which must have manufacturing tolerance. These (and many other
similar tolerances or variations) result in scales with a statistical
distribution of accuracies.
Propagation of error calculations assess the expected overall effect
of each tiny tolerance range; and importantly, some are more critical
than others. Employing better technology to make beam length 0.001"
more consistent is not nearly as effective as making beam depth 0.001"
more consistent. Briefly, the influence of depth is so great that it
would typically mask any tiny error in beam length, so fussing with
length isn't worth much. It takes only a little calculus to prove
this.
Still, we can build scales that are extremely accurate. They come off
the line with a spread of inaccuracies that are statistically
distributed, as a result of the combined statistical effects of each
potential inaccuracy, but we can correct for that. We do it largely
by calibrating after the fact, IOW by taking actual measurements.
In a bike race, we have a similar situation. A few grams, or even
ounces, of air drag isn't worth much, and its benefit is not as
definite as wind tunnel predictions claim, because there are so many
other things in a race that matter very much more.
>
> My time up local hills in the 8-10 minute range is
> generally consistent to 5%, unless there's an insane
> headwind or I'm sick or something. If I dropped 1-2%
> of weight (off bike or body), it would take averaging
> quite a few rides to detect the effect, but do you
> think it would not happen?
I think it would take hundreds of rides to see the effect of dropping
1% of your body weight on a ten minute climb. I think it would take
thousands of rides to see the effect of dropping 1% of your bike's
weight.
You do understand that what you'd be looking at would be two
overlapping normal curves, right? And that you'd be hoping to find a
small shift in the means of the data for each case, because there
would be so much overlap? And that there are statistical computations
necessary to assess a confidence interval, i.e. indicate the odds that
any changes weren't merely by chance? That's the situation you're up
against.
- Frank Krygowski
Robert Chung
> And we see that here when (for example) someone implies that a 3 gram
> drag reduction in a wind tunnel really will lead to a two-second
> improvement in a race time.
Frank, you ignorant slut, you keep bringing it up and I keep laughing at you
'cuz it shows how clueless you are. However, since we're winding down this
thread it's probably time I clear this up. If you go back to the post where
I linked that hand position test, you'll see that I wasn't claiming that it
was a decisive change. I was pointing to it as an example of how good racers
can't afford to dismiss any potential advantage as negligible -- that guy
actually tested it to *make sure.* The difference between you and those guys
is that they check their assumptions. You don't, which is why you've been
spanked so many times in this thread and Ben and I have been eating your
lunch. BTW, speaking of that, could you tell your momma that I'm getting
tired of turkey lunchmeat? Tossing in a bag of chips every once in a while
wouldn't hurt either.
Michael Press
<224132b3-39ad-42be...@v15g2000yqn.googlegroups.com>,
Frank Krygowski <frkr...@gmail.com> wrote:
> On Feb 24, 11:22 pm, Michael Press <rub...@pacbell.net> wrote:
> > In article
> > <fa370054-cfef-4a46-a832-22b229ecf...@u38g2000yqe.googlegroups.com>,
> > Frank Krygowski <frkry...@gmail.com> wrote:
> >
> > > Just a side note: That's also how you do engineering problems; but
> > > engineers are trained not to oversimplify, because being simplistic
> > > won't get the job done. Physicists are sometimes not clear on that.
> >
> > Physicists do a lot of approximation; they stack approximation
> > upon approximation. Seriously. Go to Landau and Lifschitz,
> > Theory of Elasticity and look at how much approximation
> > they do before they get to some hair differential equations
> > that need further approximation to solve. Finally their
> > results get to the engineers. Thing is, those physicists
> > are good at it. EM theory: antennae, interaction of matter
> > and EM, optics. More approximatons. Condensed matter, crytals.
> > How do we get any usable answeres out of quantum theory?
> > Show me where physicists go astray.
>
> Already ansered, I think. For example, that physicist went astray
> when he implied that producing holograms is not complicated - that
> "it's just a Fourier transform." Dismissing all complication, all
> practicalities, is going astray.
The remark is entirely out of context. I can imagine
several contexts, one of which is a clueless somebody made a
categorical statement exhibiting his utter ignorance
of the mathematics of holograms.
Why was that physicist there anyway? Was he a paid consultant?
>
> And we see that here when (for example) someone implies that a 3 gram
> drag reduction in a wind tunnel really will lead to a two-second
> improvement in a race time.
In article
<fa370054-cfef-4a46...@u38g2000yqe.googlegroups.com>,
Frank Krygowski <frkr...@gmail.com> wrote:
> Sure - right. But you don't even have to understand Fourier
> transforms to produce a good hologram.
But it helps. Saves a lot of time; unless they already rely
on work done by people who necessarily must understand
Fourier transforms. There is little enough to know about
FT's for all of engineering. It's like trigonometry. You
can learn enough for an engineering lifetime in a week.
The mathematics can occupy a someone for a good long time.
> To do them as we did, you do
> need to understand lasers, optics, process environmental control,
> electronic and computer control of complicated machines, geometry,
> machining, plating chemistry, etc. etc. (Not to mention economics,
> business, OSHA...) There's a LOT of stuff beyond the math theory. But
> that guy was interested only in the theory. He could never make a
> hologram.
Are you saying that research and development for this product
did not include a couple desk-high stacks of offprints from
physics journals?
--
Michael Press
Barry
>> generally consistent to 5%, unless there's an insane
>> headwind or I'm sick or something. If I dropped 1-2%
>> of weight (off bike or body), it would take averaging
>> quite a few rides to detect the effect, but do you
>> think it would not happen?
> I think it would take hundreds of rides to see the effect of dropping
> 1% of your body weight on a ten minute climb. I think it would take
> thousands of rides to see the effect of dropping 1% of your bike's
> weight.
> You do understand that what you'd be looking at would be two
> overlapping normal curves, right? And that you'd be hoping to find a
> small shift in the means of the data for each case, because there
> would be so much overlap? And that there are statistical computations
> necessary to assess a confidence interval, i.e. indicate the odds that
> any changes weren't merely by chance? That's the situation you're up
> against.
How small an advantage is worth worrying about? Suppose a rider is doing the
Mount Washington hill climb. Take combined body+bike weight to be 70kg, and
estimated time of 70 minutes for the climb. Quick rule of thumb says that
reducing weight by 1 kg (1/70) would save about one minute on the climb.
One minute will never be directly detectable. Winning times for the top
riders from year to year vary a lot more than that (weather is probably a big
factor), and since the road is only open to bikes a few days a year, it would
take years for a rider to generate usable statistics. So if the rider could
easily save one kg by removing his saddle bag, pump, second water bottle,
etc., should he even bother? If not, what's the smallest estimated time
savings that should matter to him? Should it be the several minutes (equating
to several kg) of variability that he would probably see if he could do the
climb many times? How about the unit of time measurement - one second
(equating to about 17g)? Or something in between?
Barry
Sandy
news:go4287$hu0$1...@news.motzarella.org...
--
Bonne route !
Sandy
Verneuil-sur-Seine FR
Frank Krygowski
> >
>
> How small an advantage is worth worrying about? Suppose a rider is doing the
> Mount Washington hill climb. Take combined body+bike weight to be 70kg, and
> estimated time of 70 minutes for the climb. Quick rule of thumb says that
> reducing weight by 1 kg (1/70) would save about one minute on the climb.
>
> One minute will never be directly detectable. Winning times for the top
> riders from year to year vary a lot more than that (weather is probably a big
> factor), and since the road is only open to bikes a few days a year, it would
> take years for a rider to generate usable statistics. So if the rider could
> easily save one kg by removing his saddle bag, pump, second water bottle,
> etc., should he even bother? If not, what's the smallest estimated time
> savings that should matter to him? Should it be the several minutes (equating
> to several kg) of variability that he would probably see if he could do the
> climb many times? How about the unit of time measurement - one second
> (equating to about 17g)? Or something in between?
>
designed to minimize the effects of random chance that I'm talking
about.
I don't deny that lightening a bike+rider will get the rider to the
top of a hill faster. That's a situation with even less randomness
than a flat-course time trial. Yet, as you say, there would still be
enough trial-to-trial variation to make the effect hard to notice.
Add in all the hundreds of random effects in a road race, and the
benefits of a slight improvement become practically undetectable.
Drill 200 1/8" holes in your chainrings if you want to. It was once
fashionable, and it probably won't hurt. But it did become
unfashionable, probably because it couldn't be shown to help.
- Frank Krygowski
carl...@comcast.net
Dear Frank,
In the 2005 TDF, Dave Canada used a stylish pair of drilled chain
rings, along with numerous other small but crucial improvements, like
the thumb-strap sleeves, face-shield, aero-helmet, presumably empty
but aero water-bottle, and mini front aero disk/hub:
http://www.cyclingnews.com/photos/2005/tour05/?id=tour051/JD05tdfstg1004
Alas, Canada finished 137th out of 189 riders, 2:52 behind the winning
time of 20:51 for the 19 km/11.6 mile TT.
Maybe his failure to wear wind-booties held him back.
Removing that small chainring instead of drilling it fulll of holes
would have saved more weight and improved the aerodynamics.
Cheers,
Carl Fogel
jobst....@stanfordalumni.org
http://home.comcast.net/~carlfogel/download/CF_ala_1970__0_.jpg
http://home.comcast.net/~carlfogel/download/CF_ala_1970__1_.jpg
http://home.comcast.net/~carlfogel/download/CF_ala_1970__2_.jpg
http://home.comcast.net/~carlfogel/download/CF_ala_1970__3_.jpg
In the days of drillium, CF had not yet been used for bicycles and
drilling holes in components was the way to differentiate one's
bicycle from the norm. This rider (Bill Robertson) was a good racer
and had people believing holes in Campagnolo parts helped him win.
Jobst Brandt
Nick L Plate
Look how low his hands are. Extra energy required to maintain
position plus a restricted thoracic chamber.
TJ
Nick L Plate
John Forrest Tomlinson
<frkr...@gmail.com> wrote:
>In what situation? In a road race? A "small" amount of drag
>reduction will cause no detectable improvement, and no _practical_
>change. It will not increase your placement, which is the objective.
Will not? Never?
John Forrest Tomlinson
<frkr...@gmail.com> wrote:
>Like Robert, you're giving a time trial example,
No, the Mt. Washington Hill Climb is a mass start race. Pretty well
known in the US BTW.
Frank Krygowski
>
> Dear Frank,
>
> In the 2005 TDF, Dave Canada used a stylish pair of drilled chain
> rings, along with numerous other small but crucial improvements, like
> the thumb-strap sleeves, face-shield, aero-helmet, presumably empty
> but aero water-bottle, and mini front aero disk/hub:
>
> http://www.cyclingnews.com/photos/2005/tour05/?id=tour051/JD05tdfstg1004
>
> Alas, Canada finished 137th out of 189 riders, 2:52 behind the winning
> time of 20:51 for the 19 km/11.6 mile TT.
>
> Maybe his failure to wear wind-booties held him back.
>
> Removing that small chainring instead of drilling it fulll of holes
> would have saved more weight and improved the aerodynamics.
I think it was the shoe straps. Look how long! He's already lost the
race.
- Frank Krygowski
James Thomson
> Removing that small chainring instead of drilling it fulll of holes
> would have saved more weight and improved the aerodynamics.
Tangential point, but in this case the drilling is to allow the orientation
of the oval chainring to be varied, rather than to save weight:
http://www.rotorcranksusa.com/images/130lg1.jpg
James Thomson
bjwe...@gmail.com
Chung,
Do you have a spare Dunning-Kruger meter I could
borrow? I think mine's broken. The needle is pegged
and it keeps flashing "OVERLOAD."
Franklin Institute,
You're making my head hurt. Is this farrago
of assumptions, bets, and lack of quantification
an example of how you teach?
I am familiar with error propagation. I do experimental
design in a line of work where we cannot modify or have
physical access to the objects under study, so our
results generally require statistical comparisons and
control samples.
You claim that small improvements have negligible
effect because there are so many other factors.
Let's consider an all-else-is-equal case and assume
that the other factors are random. In fact, let's
just think about my example of racing up a 10-minute hill,
with 5% variation in time, and a 1% difference between
riders.
I estimated that my time up such a hill varies by no
more than 5% from one given Sunday to another, but
let's be generous and say that my time has an RMS
of 5% and is normally (Gaussian) distributed. That
means that on 32% of the Sundays, I will be slower
than 10:30 or faster than 9:30. That's actually
quite a bit of variation, probably more than I have
when riding alone. But what the hell, let's assume
funny things happen in races to increase the
variation.
Now suppose I'm Fat Me and I'm racing Skinny Me
up the hill. Skinny Me is just like me but is 1%
faster because he lost approx. 2 lb off rider+bike,
or read the chart and used better tires, or trained
himself to be 1% better (which is a fair amount of
work). On the average he's 6 seconds faster.
How significant is that?
I summarize your assertion as (1) Fat Me and Skinny
Me have overlapping normal distributions of times,
so it would take hundreds of tries to measure a
statistically significant difference in their
mean times, therefore (2) Skinny's advantage
over Fat is negligible.
It turns out that (1) is correct but the conclusion
(2) is quite wrong. The reason is that you don't
have to win a race by 3 sigma, you just have to
be faster than the other guy. You need to make
the 3 sigma measurement if you want to know
_how much_ faster you were.
First, quantifying the difference in mean times.
The standard error in the mean of N trials
is RMS/sqrt(N). RMS= 0.05, we are trying
to measure a difference of 0.01, so to make a
3 sigma measurement we need error=0.0033.
The errors on the mean times of Fat and Skinny
add in quadrature so the error on one has to
be 0.0033/sqrt(2). So we find that we need
459 races to make a 3 sigma measurement
of the difference in times.
It's clearly a tiny effect ... but not so fast!
In terms of racing, we care not about the mean
times, but about how often Skinny beat Fat
and vice versa. This is a tedious calculation
with error functions, so I wrote a little program
to generate random "races" with normal
distributions and Skinny 1% faster. (It's
a trivial program if you can generate Gaussian
random distributions.)
Guess what - it turns out that Skinny wins 55.6%
of the trials! His probability of winning improved
by 1.1x (10%) even though he only made a 1%
improvement in ability, well below the RMS.
The 1%, even though it is within the RMS,
shifts the peak of the bell curve enough that
it makes a difference in a fair fraction of the
trials. Sure, you would have to run a lot of
races to measure that Skinny's winning
percentage was 55% precisely. But Skinny
doesn't need to estimate it to 3 sigma. Faster,
it turns out, is in fact better. The potential
improvement when races are close - and they
generally are close for the top people, as opposed
to backmarkers like me - is substantial.
You're getting dropped, Frank. If you aren't ever
going to quantify your assertions, you're going
to keep coming in behind.
Ben
Barry
> times, but about how often Skinny beat Fat
> and vice versa. This is a tedious calculation
> with error functions, so I wrote a little program
> to generate random "races" with normal
> distributions and Skinny 1% faster. (It's
> a trivial program if you can generate Gaussian
> random distributions.)
> Guess what - it turns out that Skinny wins 55.6%
> of the trials! His probability of winning improved
> by 1.1x (10%) even though he only made a 1%
> improvement in ability, well below the RMS.
It's easy to solve this analytically. Since the times for Fat and Skinny are
both normally distributed, the difference between the two is also normal,
with:
Mean = difference of the two means = 1% * 10 minutes = 6 seconds.
Sigma = sqrt of the sum of the squares of the two sigmas. Since the two
sigmas are equal, this gives sigma*sqrt(2) = 5% * 10 minutes * sqrt(2) = 42.4
seconds.
The proportion of the time Skinny will beat Fat is equal to the value of the
cumulative normal distribution for z = Mean/Sigma = 1/(5*sqrt(2)) = sqrt(2)/10
= 0.14. This is about 0.556 = 55.6%, which agrees with your simulation
result.
Barry
incredulous 2
1. Yes, the average rider who has given himself an advantage he lacked
yesterday will perform better today. But, the average rider is so
unlikely, with that small advantage, to significantly advance his /
her standing relative to winning. He /she is, after, all, just
average.
2. Taking the thread subject seriously, in tire design and performance
all else is not equal. It surprises me that race-oriented posters here
don't comment knowledgeably on offsets. Most races, including time
trials do not call for all-out speed, but energy conservation and
appropriate rest. Tires and operating conditions that beat the rider
up may be less desireable than those that are faster. There are
differences in adhesion etc.
3. In competitive swimming in which the equipment is reduced to body
and suit (costume to Brits?) hundreds of thousands of age-group
swimmers are inclined to spend large sums on the latest of short-lived
hi-tech suits. The same sums might be spent on filming and analysis of
their stroke instead. For the overwhelming majority, whatever the
small difference in place --with certainty -- of the suit, the suit
will not advance them from mediocrity to excellence. Which has been, I
think, one of FK's points.
Harry Travis
Frank Krygowski
> You're making my head hurt. Is this farrago
> of assumptions, bets, and lack of quantification
> an example of how you teach?
Ben, I'm not getting paid for this. I do _lots_ of numerical examples
when I teach. It shouldn't be necessary to open a book for this
conversation. The concepts should be obvious, especially if you've
got the background you claim.
This is not hard to visualize, even without the math. In a sense,
you're overlapping two bell curves, and looking at how often random
points on one are to the right of random points on another. But see
below.
>
> The 1%, even though it is within the RMS,
> shifts the peak of the bell curve enough that
> it makes a difference in a fair fraction of the
> trials. Sure, you would have to run a lot of
> races to measure that Skinny's winning
> percentage was 55% precisely. But Skinny
> doesn't need to estimate it to 3 sigma. Faster,
> it turns out, is in fact better. The potential
> improvement when races are close - and they
> generally are close for the top people, as opposed
> to backmarkers like me - is substantial.
>
> You're getting dropped, Frank. If you aren't ever
> going to quantify your assertions, you're going
> to keep coming in behind.
>
> Ben
Ben, I think you're underestimating the real variability. A road race
(what I've been trying mightily to discuss) is not "fat me vs. skinny
me." A road race is "fat me vs. skinny me vs. skinny him vs. mid-
weight that guy vs. muscular but heavy guy vs. light guy with aerobic
capacity of a horse vs. young kid with killer watt/kg but no
experience vs. old cunning dude who's still in damned good shape" and
on and on. And each of those guys has some slight difference of
opinion on equipment, which is what we're really focusing on. And
each will roll over smoother or rougher patches of asphalt. Each will
take a different line through a crowded turn, each will be behind
someone bigger or smaller, each will react differently to a breakaway,
each will be lucky or unlucky in positioning for a sprint.
Briefly, I think that your finishing position in hypothetically
identical races would have standard deviations far higher than the 5%
what you assume for your hill climb time. (And ISTM that it's
actually finishing position, not time, that's a concern for most road
racers. Stage races are a partial, not complete, exception.) For
example, in a 20 racer field, is it realistic to say that a rider who
places tenth, on average, will essentially _never_ be worse than 13th,
and _never_ better than 7th (i.e. always within 3 times 'sigma = 5%'
of his mean) ? I don't believe that's true. In fact, I think if it
were true, most amateurs would give up racing entirely. They race for
the hope that they'll place way better next Sunday. Too much
consistency would destroy the sport.
I'll also note that you've avoided the propagation of error analogy,
which is at the heart of the question. What percentage of the
variability in finishing position is assignable to weight? The 5%
sigma vs. 1% weight you've proposed is really your way of estimating
the relative effect of all the other variables, compared to taking 2
pounds off the bike+rider. Again, I think that 5% is insufficient;
there are so many other influences. (And you've chosen the one type
of race that emphasizes weight differences and minimizes the many
other factors.) (BTW, I've never done a competitive mass-start hill
climb, but climbing stages of the Tour de France seem to have
competitors all over the place at the finish, i.e. not very consistent
at all.)
Another point: You've correctly focused on bike+rider weight, since
that's what gravity attracts. But most of the "nothing is negligible"
crowd has been talking about equipment. When I've talked about
alternatives to equipment that make more difference, I mentioned
training and diet. In a sense, your example argues against me by
using part of what I was advocating. Unless, that is, you believe a
racer can reasonably find two pounds to remove from his bike. I think
that would be rare, unless he just buys a new bike; and I think we'd
agree that's no small change - by either possible meaning of the
phrase!
Finally, regarding calculations and quantification: I think the huge
randomness of road racing is way beyond what we can calculate or model
easily. I know that even modeling ordinary flow of roadway traffic is
very difficult, and is quite imperfect despite decades of dedicated
research and computer programming. That's why I suggested dice - lots
and lots of dice - to try to understand the situation.
Or testing. Can we get two evenly matched racers to blindly,
randomly, trade a few ounces of hidden weight back and forth for a
season, and analyze their finishing positions relative to each other
at season's end? Wouldn't that tell us something? It won't be hard
to do the test. All we need is two volunteer racers and one helper.
- Frank Krygowski
John Forrest Tomlinson
<travis...@gmail.com> wrote:
>In competitive swimming in which the equipment is reduced to body
>and suit (costume to Brits?) hundreds of thousands of age-group
>swimmers are inclined to spend large sums on the latest of short-lived
>hi-tech suits. The same sums might be spent on filming and analysis of
>their stroke instead. For the overwhelming majority, whatever the
>small difference in place --with certainty -- of the suit, the suit
>will not advance them from mediocrity to excellence. Which has been, I
>think, one of FK's points.
Are 700x25s really that much cheaper than 700x23s? I did not know
that.
John Forrest Tomlinson
<frkr...@gmail.com> wrote:
>When I've talked about
>alternatives to equipment that make
>more difference, I mentioned
>training and diet.
You mean the choice of a better tire is distracting me from improving
training or diet? Really?
Nick L Plate
How to win a road race.
Block in the horse, when the old pro jumps, let the horse go and jump
on his back. Keep doing similar, then outsprint them at the finish.
Too late for me now.
TJ
Sandy
news:85040221-5a01-4343...@v15g2000yqn.googlegroups.com...
Well, Harry, you don't have any idea how races unfold, do you? Have you
ever heard the desperate cry of "Break-Break!!!" ? That's when the places
get sorted out and the negligible advantages play out. Training, equipment,
state of energy. So racing, not riding, requires several optimum
performances at different times, and most of all at the ride towards and
across the finish line. It is just the average lame excuse for someone who
fails to take first place to give eloquent expression to the detailed
reasons his effort failed, but at those critical moments of exceptional
performance, both body and equipment speak even more eloquently and
succinctly. Ben has done what he can to quantify the elimination process,
yet you fail to see that you spelled out exactly what you need to look at -
EVERYTHING. To paraphrase another long-battling contributor: if you decide
not to see if some potential bit of gimmickry or mode can help you get
across the line first, you just made your odds of being there negligible.
carl...@comcast.net
<yosn...@hotmail.com> wrote:
Dear James,
Interesting--thanks!
If I follow the idea, the numerous holes allow you to mount the oval
rings at almost any angle relative to the pedals?
I wonder which way they were mounted on Canada's bike?
http://www.cyclingnews.com/photos/2005/tour05/?id=tour051/JD05tdfstg1004
The egg-shape isn't pronounced enough for me to tell from the photo.
Is the pointier, bulgier end pointing toward his feet? Or at 90
degrees to the crank? Somewhere in-between?
(I know that both ways have been advocated, but can't remember which
way is preferred.)
Two ominous possibilities arise.
First, drilled oval rings might be _heavier_ than normal chain rings.
Second, Canada might have mounted the eggs the wrong way.
Cheers,
Carl Fogel
Michael Press
<89bca400-f4c7-496c...@z9g2000yqi.googlegroups.com>,
Frank Krygowski <frkr...@gmail.com> wrote:
> But most of the "nothing is negligible"
> crowd has been talking about equipment.
Remember when you were characterized as an anti-helmet zealot?
--
Michael Press
bjwe...@gmail.com
Elegant - thank you. I was hung up on integrating
over the two probability distributions and didn't think
of the path of calculating the single distribution of
time gaps, which is much easier to cumulate, as you
point out.
Ben
bjwe...@gmail.com
Franklin Covey Time Organizer,
When you noted (correctly) that the 1% advantage
with 5% rms would take hundreds of trials to determine
a statistically significant difference, you didn't object
to the specifics of that example.
When I pointed out that it's rank-ordering that's
important and just the 1% advantage does in fact
make a substantial difference in finish order in this
simplified example, you didn't like the numbers anymore.
I didn't avoid the propagation-of-error. I modeled effects
using observations: my time up climbs is consistent
to within 5%. You present no alternative number or model.
You also say:
> (BTW, I've never done a competitive mass-start hill
> climb, but climbing stages of the Tour de France seem to have
> competitors all over the place at the finish, i.e. not very consistent
> at all.)
It's true about the gaps, but ignores that (1) many of the riders
who are not in GC contention are not pedaling full-gas, (2) there
are large gaps but consistent performances, in that the same
few riders are usually in contention for a win, (3) different
riders have different strengths in the Tour - some are slower
on hills but stronger on flat TTs, and so on. Small gaps
are important in races.
Sure, there are lots of random uncontrollable variables
in road racing. Faced with such a problem, we can
pretend ignorance, or we can construct a model based
on physics that we do understand using input from
experiments that can be controlled (for example,
tire Crr tests or aero drag measurements and predictive
models of actual race times). Several times in this
thread, I've constructed an example, you've claimed
there's a negligible difference, I've shown the difference
is potentially significant (for ex the Pescadero Road
Race where riders frequently finished within <10 sec
of the winner), and then you've claimed there are
too many things we don't know. It's getting tiresome.
It seems that you don't think 1% is ever going to be
a significant difference. How big a difference would you
accept as significant? 5%? 10%? Can you offer any model
or design an experiment and show how large a difference
could be measured? Don't forget the error budget.
While you're thinking about percentages, keep in mind
that the difference in watts/kg between an out of
shape ex racer approaching middle age like me, or as
the South Park kids would say, a fat fuck, versus a
quality if not top pro, is less than a factor of 2. Maybe
approaching 50-60%. Their recovery is better,
because youth is wasted on the young, and they can
race for 8 hours which amateurs can't. But this is
in fact a game of percentages.
Ben
Barry
> over the two probability distributions and didn't think
> of the path of calculating the single distribution of
> time gaps, which is much easier to cumulate, as you
> point out.
To prove that the sum of two normal distributions is also normal does require
doing that integration, which is tricky. The proof I found involves two
changes of variables.
Barry
Frank Krygowski
wrote:
>
>
> Franklin Covey Time Organizer,
?? Do you think silly salutations really add quality to your argument?
>
> When you noted (correctly) that the 1% advantage
> with 5% rms would take hundreds of trials to determine
> a statistically significant difference, you didn't object
> to the specifics of that example.
>
> When I pointed out that it's rank-ordering that's
> important and just the 1% advantage does in fact
> make a substantial difference in finish order in this
> simplified example, you didn't like the numbers anymore.
On the contrary, I embraced the idea. I asked it be applied more
realistically, to the problem I've been discussing, a road race. I
asked that you apply the +/- three sigma estimates to the rank order
of a 20 person road race. Your premises gave an unreasonable result -
that a rider would never finish three places different from his
average. Obviously, this means your 5% sigma is unrealistic.
>
> I didn't avoid the propagation-of-error.
As I've always seen it, the first benefit of propagation of error
calculations is to help decide what influence is most important, and
what influences are comparatively negligible. I don't see you've
addressed that at all.
> > (BTW, I've never done a competitive mass-start hill
> > climb, but climbing stages of the Tour de France seem to have
> > competitors all over the place at the finish, i.e. not very consistent
> > at all.)
>
> It's true about the gaps, but ignores that (1) many of the riders
> who are not in GC contention are not pedaling full-gas, (2) there
> are large gaps but consistent performances, in that the same
> few riders are usually in contention for a win, (3) different
> riders have different strengths in the Tour - some are slower
> on hills but stronger on flat TTs, and so on. Small gaps
> are important in races.
ISTM you're saying "There are all sorts of reasons for the
inconsistencies," which is what I'm saying, too. That doesn't make
the inconsistencies vanish!
And regarding "the same few riders usually in contention for a win,"
is that really true? How often has a tour contender blown up on a
tough climb? It's hardly rare.
>
> Sure, there are lots of random uncontrollable variables
> in road racing. Faced with such a problem, we can
> pretend ignorance, or we can construct a model based
> on physics that we do understand using input from
> experiments that can be controlled (for example,
> tire Crr tests or aero drag measurements and predictive
> models of actual race times).
And by analogy, there are lots of uncontrolled variables in weather,
but we can construct a model based on physics we understand and
predict the weather next week. Or, that's what some people thought in
the 1970s, when computing power started to take off. But it doesn't
work, despite decades of very expensive and dedicated work driven by
immense potential financial benefits. The randomness and chaos
overpowers deterministic physics. We've finally gotten to where
tomorrow's weather is a fairly good bet, but next week? Forget it!
> It's getting tiresome.
The thread is getting tiresome for both sides.
> It seems that you don't think 1% is ever going to be
> a significant difference. How big a difference would you
> accept as significant? 5%? 10%?
<sigh> Go back to the dice example. It's a probabilistic question,
with too many random variables to enable a realistic analysis.
The proper sort of answer would be "a 1% weight advantage will prove
significant (i.e. change a person's final placement) X% of the time."
But I don't believe anyone has enough understanding, or enough data,
to place a reasonable estimate on X.
What's needed (as I've said) is actual test. The easiest of the three
main equipment factors to test is weight. So again: We need two
evenly matched racers and one helper. The helper constructs two
outwardly identical weights, differing by - whatever, 1% of the bike
weight? - and randomly hides them (say, below the bottom bracket) on
each bike before each race. Obviously, he can't tell the racers who's
got the extra weight. He keeps track of the relative finishing order
of the two riders for each race during the season. And then he starts
looking for significance in the results.
Testing is the only way. The problem is no more solvable by clean
calculations than is the position of a chaotic double pendulum at a
specified time.
Now, can someone do that test and report?
- Frank Krygowski
bjwe...@gmail.com
> On Feb 27, 4:43 am, "b...@mambo.ucolick.org" <bjwei...@gmail.com>
> wrote:
> > Franklin Covey Time Organizer,
>
> ?? Do you think silly salutations really add quality to your argument?
Frank Blank,
No, but they amuse me, which you are ceasing to.
> On the contrary, I embraced the idea. I asked it be applied more
> realistically, to the problem I've been discussing, a road race. I
> asked that you apply the +/- three sigma estimates to the rank order
> of a 20 person road race. Your premises gave an unreasonable result -
> that a rider would never finish three places different from his
> average. Obviously, this means your 5% sigma is unrealistic.
You conflated 5% in time and in rank order again.
At the top levels of a race (which is the only place that
worrying about small differences matters) riders are
pretty consistent. I used to help with a series of local
MTB and CX races and over 4-5 races it would usually
be the same 3-5 people duking it out in the top few
places of a given category.
> > It seems that you don't think 1% is ever going to be
> > a significant difference. How big a difference would you
> > accept as significant? 5%? 10%?
>
> <sigh> Go back to the dice example. It's a probabilistic question,
> with too many random variables to enable a realistic analysis.
>
> The proper sort of answer would be "a 1% weight advantage will prove
> significant (i.e. change a person's final placement) X% of the time."
> But I don't believe anyone has enough understanding, or enough data,
> to place a reasonable estimate on X.
You are avoiding the question. I don't want to know
how big an effect you think a 1% change in ability has.
I want to know how big a change in ability you think it
would take to make a detectable difference. Because
you keep saying that it would be better to train harder
instead of worrying about tires, I assumed that you had
some idea that training harder improves race performance.
On the other hand, perhaps you believe that there are
so many confounding variables that we can't ever
be sure that training harder would improve performance
when averaged over a number of races.
That would be somewhat dispiriting.
>
> What's needed (as I've said) is actual test. The easiest of the three
> main equipment factors to test is weight. So again: We need two
> evenly matched racers and one helper. The helper constructs two
> outwardly identical weights, differing by - whatever, 1% of the bike
> weight? - and randomly hides them (say, below the bottom bracket) on
> each bike before each race. Obviously, he can't tell the racers who's
> got the extra weight. He keeps track of the relative finishing order
> of the two riders for each race during the season. And then he starts
> looking for significance in the results.
>
> Testing is the only way. The problem is no more solvable by clean
> calculations than is the position of a chaotic double pendulum at a
> specified time.
>
> Now, can someone do that test and report?
I asked for an error budget on any suggested experiment.
Let us assume for the sake of argument that a 10% improvement
in performance - racer A beats B 55% of the time, rather than
50% - would be worthwhile. I think many racers would consider
such an improvement significant. It would certainly be
considered important in Vegas. In your suggested experiment,
how many races would the two have to do in a season to
detect this 10% improvement (at say 95% confidence)?
Ben
Robert Chung
>>> It seems that you don't think 1% is ever going to be
>>> a significant difference. How big a difference would you
>>> accept as significant? 5%? 10%?
>>
>> <sigh> Go back to the dice example. It's a probabilistic question,
>> with too many random variables to enable a realistic analysis.
>>
>> The proper sort of answer would be "a 1% weight advantage will prove
>> significant (i.e. change a person's final placement) X% of the time."
>> But I don't believe anyone has enough understanding, or enough data,
>> to place a reasonable estimate on X.
>
> You are avoiding the question. I don't want to know
> how big an effect you think a 1% change in ability has.
> I want to know how big a change in ability you think it
> would take to make a detectable difference. Because
> you keep saying that it would be better to train harder
> instead of worrying about tires, I assumed that you had
> some idea that training harder improves race performance.
> On the other hand, perhaps you believe that there are
> so many confounding variables that we can't ever
> be sure that training harder would improve performance
> when averaged over a number of races.
> That would be somewhat dispiriting.
Dumbass,
Except for fat masters like you and me who get, um, fat in the off season,
semi-serious racers don't let themselves go so they won't see more than
maybe a 10% difference between off season and in season power. That 10%
difference from training? Frank isn't convinced that's not negligible.
Michael Press
"Barry" <Ba...@invalid.invalid> wrote:
The sum is the convolution of the two pdf's, and that is
the Fourier transform of the product of the Fourier transforms.
Gaussian pdf of mean 0, variance ss:
f(x) = 1/(s.sqrt{2.pi}) exp{-x^2/ss}
The pdf for the sum of two Gaussian pdf's of mean 0, variance ss:
g(x) = (f * f)(x)
Denote by F[f] the Fourier transform of f.
F[f](z) = 1/(s.sqrt{2.pi}) sqrt{ss.pi} exp{-(pi.s}^2 z^2}
= 1/sqrt{2} exp{-(pi.s}^2 z^2}
g(x) = F[(1/2 exp{-2(pi.s}^2 z^2}]
= 1/2 sqrt{pi/(2(pi.s}^2)} exp{- pi^2.x^2/(2(pi.s}^2)}
= 1/(2.s.sqrt{1/(2.pi}} exp{- x^2/(2(s}^2)}
and this is a Gaussian pdf of mean 0, variance 2.ss
I used F[exp{-a.x^2}](z) = sqrt{pi/a} exp{(pi.z)^2/a}.
The extension to non-zero means and unequal variances is intuitively obvious.
--
Michael Press
bjwe...@gmail.com
wrote:
To be clear, I should have called this a "10% improvement
in results" where by results I mean rank-order of race finishes.
I want to keep distinct the percentage improvements in
ability (such as watts/kg, or speed on climbs), finishing
time, and rank ordered results, since they are frequently
not linearly related.
Ben
Frank Krygowski
wrote:
> On Feb 27, 10:47 am, Frank Krygowski <frkry...@gmail.com> wrote:
>
> > On Feb 27, 4:43 am, "b...@mambo.ucolick.org" <bjwei...@gmail.com>
> > wrote:
>
> > On the contrary, I embraced the idea. I asked it be applied more
> > realistically, to the problem I've been discussing, a road race. I
> > asked that you apply the +/- three sigma estimates to the rank order
> > of a 20 person road race. Your premises gave an unreasonable result -
> > that a rider would never finish three places different from his
> > average. Obviously, this means your 5% sigma is unrealistic.
>
> You conflated 5% in time and in rank order again.
Somehow, "returning to the subject" is being disparaged.
When I first described tiny bike improvement as being negligible,
JFT's (and others') objections were that any tiny bike improvement
_might_ make a critical difference in the finish of a race - that they
might enable one to outsprint an opponent, or perhaps win overall. At
least, that's how I understood the argument.
That's why I've mightily tried to stick to road races and crit races.
(I've inserted recreational rides from time to time only because the
OP really didn't specify "racing" at all.) My major point has been
that the myriad of factors in road racing, especially the myriad of
random factors, make tiny bike changes undetectable.
Chung repeatedly tried to shift the focus to time trials, and maybe
triathlon bike legs. You're trying to shift to - what? Two-person
match racing up a hill? Interesting as a side thread perhaps, but
different. Those are less random. You've got a better chance at
detecting a small bike improvement's benefit. But you still won't be
able to detect a tiny bike improvement.
And as I've explained, it's impossible to quantify "small" and "tiny"
beyond mere guesswork, because we have no hope of quantifying the
influences of the hundreds of other factors.
If you can produce a realistic model that predicts race placement
based on percentage of race miles a racer passes over rougher, vs.
smoother patches of pavement; on percentage of race miles he's
drafting a bigger, rather than smaller opponent; percentage he's
riding toward back-center of a wide peloton, rather than being out on
the windy edge; percentage of breaks he's not blocked from jumping
onto; percentage of corners he can get an ideal line, and so on, then
perhaps we can postulate two ideal, identically matched riders and
model a few ounces extra on one bike. To me, it sounds like a PhD
thesis, and I doubt it'll be an accurate model anyway. So good luck.
In a situation like this, the only practical way of quantifying is by
repetitious experiment. It's going to take hidden weights applied at
random and an entire season's racing to even gather the data. (And,
of course, I expect a null result.)
Now why are no racers volunteering to do that test?
- Frank Krygowski
jwbinpdx
Another way of looking at this is that in any road race, there is a
time trial. It might be when the pack splits and you are stuck behind
some dill weed who gaps you off. You have to jump around and time
trial for a minute or five to get back on -- or much longer if you are
unfortunately dumped on a climb. Or, in the finishing sprint, you are
time trialing for the last 300 meters alone because someone baubles
the final turn and the pack slows, and there you are, fortuitously on
the outside front and destined to be a hero if you can hold it
together for 60 second. Look at me! I'm going to win the 650B sew up
and bag of pretzels! . . . Do'h! (two people go by in the last two
feet -- you only get the burrito meal deal certificate). Long hill
climbs in road races are basically time trials, or they can be if the
pack splits up enough. Our spring races in Oregon are usually run in
rain storms and high winds. The finish sometimes look like the end of
the Boston Marathon -- lots of individuals straggling over the finish
line. You never know when a road race is going to be a time trial, or
for whom. -- Jay Beattie.
bjwe...@gmail.com
>
> If you can produce a realistic model that predicts race placement
> based on percentage of race miles a racer passes over rougher, vs.
> smoother patches of pavement; on percentage of race miles he's
> drafting a bigger, rather than smaller opponent; percentage he's
> riding toward back-center of a wide peloton, rather than being out on
> the windy edge; percentage of breaks he's not blocked from jumping
> onto; percentage of corners he can get an ideal line, and so on, then
> perhaps we can postulate two ideal, identically matched riders and
> model a few ounces extra on one bike. To me, it sounds like a PhD
> thesis, and I doubt it'll be an accurate model anyway. So good luck.
>
> In a situation like this, the only practical way of quantifying is by
> repetitious experiment. It's going to take hidden weights applied at
> random and an entire season's racing to even gather the data. (And,
> of course, I expect a null result.)
>
> Now why are no racers volunteering to do that test?
Franklin Pierce,
I asked you to do an error analysis for this
proposed repetitious experiment of yours. Do it.
When you write a proposal to do an experiment,
you have to show that the experiment is
sensitive enough to detect a effect at an
interesting level. Otherwise it won't get funded,
approved by an insitutional review board, or
otherwise paid attention to.
I postulated that an improvement of say 10% in
rank-order of results is something most racers
would consider significant. That is, if A and B
are otherwise equal, but A makes one change
and then has a 55% chance of finishing ahead
of B in any given race, I think A would care.
(This would also apply to your proposed experiment
where we blind it by swapping the advantage
between A and B randomly to eliminate other
effects, placebos, etc.)
My question to you is, how many races do A and
B have to do in your experiment before you can
be confident of detecting this effect?
Ben
Frank Krygowski
wrote:
> On Feb 28, 9:50 am, Frank Krygowski <frkry...@gmail.com> wrote:
>
> > In a situation like this, the only practical way of quantifying is by
> > repetitious experiment. It's going to take hidden weights applied at
> > random and an entire season's racing to even gather the data. (And,
> > of course, I expect a null result.)
>
> > Now why are no racers volunteering to do that test?
>
> proposed repetitious experiment of yours. Do it.
? Is that a command? Or are you offering me serious money?
Whatever it is, I won't be doing it, and nobody else will be doing it,
until you or someone else provides the road race statistical model I
mentioned. Here it is again:
> > If you can produce a realistic model that predicts race placement
> > based on percentage of race miles a racer passes over rougher, vs.
> > smoother patches of pavement; on percentage of race miles he's
> > drafting a bigger, rather than smaller opponent; percentage he's
> > riding toward back-center of a wide peloton, rather than being out on
> > the windy edge; percentage of breaks he's not blocked from jumping
> > onto; percentage of corners he can get an ideal line, and so on, then
> > perhaps we can postulate two ideal, identically matched riders and
> > model a few ounces extra on one bike. To me, it sounds like a PhD
> > thesis, and I doubt it'll be an accurate model anyway. So good luck.
> When you write a proposal to do an experiment,
> you have to show that the experiment is
> sensitive enough to detect a effect at an
> interesting level.
:-) Even if we're trying to detect whether the effect is detectable?
What was the definition of "tautology" again?
> Otherwise it won't get funded,
> approved by an insitutional review board, or
> otherwise paid attention to.
Oh come on, Ben. You don't need funding. The experiment I described
wouldn't cost $5. Isn't it worth it, to prove youself correct?
You use two different hidden weights that are visually identical
during a season of racing. You zip tie the weights under the bottom
brackets of two closely matched racers, shuffling the weights each
time, and you keep track of their finishing order.
But yes, I predict people won't pay attention to the results, for two
reasons. The results will be "no detectable difference," which few
find interesting. And the "nothing is negligible" crowd will say
"Well, _some_ of the time the guy with the lighter hidden weight beat
the other guy, and it COULD have been because of the lighter weight,
so _maybe_ it helps sometimes, and that's good enough for us."
That's the sort of thinking that they've been using, anyway. I doubt
that test results would have any affect on such "logic."
- Frank Krygowski
Michael Press
<46b5e895-6ee0-42c9...@c36g2000yqn.googlegroups.com>,
Frank Krygowski <frkr...@gmail.com> wrote:
> On Feb 27, 8:16 pm, "b...@mambo.ucolick.org" <bjwei...@gmail.com>
> wrote:
> > On Feb 27, 10:47 am, Frank Krygowski <frkry...@gmail.com> wrote:
> >
> > > On Feb 27, 4:43 am, "b...@mambo.ucolick.org" <bjwei...@gmail.com>
> > > wrote:
> >
> > > On the contrary, I embraced the idea. I asked it be applied more
> > > realistically, to the problem I've been discussing, a road race. I
> > > asked that you apply the +/- three sigma estimates to the rank order
> > > of a 20 person road race. Your premises gave an unreasonable result -
> > > that a rider would never finish three places different from his
> > > average. Obviously, this means your 5% sigma is unrealistic.
> >
> > You conflated 5% in time and in rank order again.
>
> Somehow, "returning to the subject" is being disparaged.
>
> When I first described tiny bike improvement as being negligible,
> JFT's (and others') objections were that any tiny bike improvement
> _might_ make a critical difference in the finish of a race - that they
> might enable one to outsprint an opponent, or perhaps win overall. At
> least, that's how I understood the argument.
The argument is that very small advantages abide. They
are always there, saving energy. A small difference in
efficiency that is the result of a sum of small advantages
can be enough at a critical point in the race to enable
a racer to match an acceleration without going anaerobic.
Going anaerobic is a big bet. Racers prefer to call raises
with a large stack of chips; that is match acceleration,
bridge to a breakaway, or fake a break without going
anaerobic. In racer's parlance: only so many matches to burn.
For a mathematical insight solve the problem of the gambler's ruin.
Two people, A and B, play a game in which the probability
that A wins is p, the probability that B wins is q, and
the probability of a draw is r. At the beginning, A has
m dollars and B has n dollars. At the end of each game
the winner takes a dollar from the loser. If A and B
agree to play until one of them loses all his/her money,
what is the probability of A winning all the money?
Small differences in a player's bank make large differences
in the outcome.
--
Michael Press
John Forrest Tomlinson
wrote:
>The argument is that very small advantages abide. They
>are always there, saving energy. A small difference in
>efficiency that is the result of a sum of small advantages
>can be enough at a critical point in the race to enable
>a racer to match an acceleration without going anaerobic.
>Going anaerobic is a big bet. Racers prefer to call raises
>with a large stack of chips; that is match acceleration,
>bridge to a breakaway, or fake a break without going
>anaerobic. In racer's parlance: only so many matches to burn.
But your forget Frank's point that most changes racers make probably
have disadvantages that outweigh the advantages. So the rider is
probably going to go slower or crash or have the bike break. Yes, I
know its' confusing, so here it is, more clearly, from Frank:
"If something increases speed, but by so small a margin that it is
practically undetectable, it should be considered negligible. Its
detriments will probably overbalance the minuscule benefit."
Tim McNamara
"Robert Chung" <anonymou...@invalid.address> wrote:
A 10% difference in physical ability resulting from training would seem
likely to be very non-negligible, in whatever capacity (sprinting,
climbing or endurance). A 1% difference might not be negligible between
competitors who are very closely matched. That's in ideal
circumstances, though; in real life there are many random factors that
determine the outcome of competition- if not, there'd be no point in
actually running the race. Joop Zoetemelk's or Maurizio Fondriest's
world championships are examples of this.
Tim McNamara
<51d914f1-5ca2-414d...@p6g2000pre.googlegroups.com>,
"b...@mambo.ucolick.org" <bjwe...@gmail.com> wrote:
> On Feb 27, 6:16 pm, "b...@mambo.ucolick.org" <bjwei...@gmail.com>
> wrote:
> > On Feb 27, 10:47 am, Frank Krygowski <frkry...@gmail.com> wrote:
> >
> > > Now, can someone do that test and report?
> >
> > I asked for an error budget on any suggested experiment. Let us
> > assume for the sake of argument that a 10% improvement in
> > performance - racer A beats B 55% of the time, rather than 50% -
> > would be worthwhile. I think many racers would consider such an
> > improvement significant. It would certainly be considered
> > important in Vegas. In your suggested experiment, how many races
> > would the two have to do in a season to detect this 10% improvement
> > (at say 95% confidence)?
>
> To be clear, I should have called this a "10% improvement in results"
> where by results I mean rank-order of race finishes.
You mean, placing 9th instead of 10th, 18th instead of 20th, 36th
instead of 49th, and so on?
> I want to keep distinct the percentage improvements in ability (such
> as watts/kg, or speed on climbs), finishing time, and rank ordered
> results, since they are frequently not linearly related.
"Not linearly related" = "influenced by random events," which has been
one of Frank's main points all along.
incredulous 2
> "incredulous 2" <travis.ha...@gmail.com> wrote in message
Actually, Sandy, if you confuse the product of tiny advantages
multiplied by rare opportunities for them to make a difference, you
will exhaust your resources to little effect. Even the same person
competing against his double will find only rare opportunities for the
small differences to matter, especially in the heterogeneous
environment of the peloton.
Lovely expression though. You've a future selling insurance against
rare but not unthinkable risks. For performance enhancing drugs and
potions,too, with the last line. Just lovely.
Harry Travis
bjwe...@gmail.com
> "b...@mambo.ucolick.org" <bjwei...@gmail.com> wrote:
>
> > To be clear, I should have called this a "10% improvement in results"
> > where by results I mean rank-order of race finishes.
>
> You mean, placing 9th instead of 10th, 18th instead of 20th, 36th
> instead of 49th, and so on?
No. I'm talking about probability of finishing higher.
In this case, to be simple, I'm only comparing
two racers. If they are equally matched, A beats B
50% of the time. When I say A improves by 10% in
results, I mean his chance of beating B improves
by 1.1x, so he beats B 55% of the time (1.1 x 50%).
Whether this means he improves his rank order
of finish by one place, two places, or whatever
would clearly depend on how many other people
there are in the race and how close they are in
ability. For people at the back of the pack, none
of this matters. But suppose we are talking about
the few fast local guys who are duking it out for
the top three places in races in a series. I'm sure
you've seen this in, say, a local cyclocross series -
in a given category there will often be a few closely
matched people who are in close contention for
the individual races and the overall. For these people,
improving their chances of victory over even odds
would be important.
>
> > I want to keep distinct the percentage improvements in ability (such
> > as watts/kg, or speed on climbs), finishing time, and rank ordered
> > results, since they are frequently not linearly related.
>
> "Not linearly related" = "influenced by random events," which has been
> one of Frank's main points all along.
No no no!!! That's not correct! You are conflating
two ideas here.
1. The first idea is that quantities may be related by
a non-linear relation. For example, speed on a climb
is nearly linear in watts/kg ("nearly" because there's
a small amount of air resistance too). But speed on
the flats goes as watts^1/3, because the power to
overcome air drag goes as speed^3. That's non-linear.
A more subtle non-linear effect is that because riders
are often closely matched in performance, a modest
improvement can make a large difference in results.
A guy (or gal) with 30% more power than me is going
to crush me in every race, not beat me 1.3x as much
as random (i.e. beat me 65% of the time).
2. The second idea that you refer to is that race
results can show variations. We wouldn't know how
much of this is stochastic (truly uncorrelated variations)
and how much is systematic (due to unmodeled
effects, e.g., certain riders do better on hillier courses).
But that's too deep for our problem.
Frank says more than that there is randomness.
He says that the randomness is so large that
it makes any small improvements negligible.
What I'm trying to drag him into seeing is that:
1. this would also imply that small improvements
in ability due to training are negligible
2. the size of what he considers negligible
(an effect that it takes lots and lots of races to
demonstrate at 3 sigma) is actually a
percentage effect that many people would
like to have, for example beating your buddy
55% of the time instead of 50%.
If Frank does the error analysis that I keep asking
for, for his proposed experiment of switching an effect
(tires or whatever) between riders A and B, for
an effect that gives a winning percentage of say
55% or 60%, he'll see how many races it would
take to establish the effect at 3 sigma or 95%
confidence or whatever statistical level one likes.
And that will give some idea of whether this
experiment would be likely to be useful.
Ben
Michael Press
I respect Frank for knowing how 'minuscule' is spelled.
--
Michael Press
carl...@comcast.net
<bjwe...@gmail.com> wrote:
[snip]
>A more subtle non-linear effect is that because riders
>are often closely matched in performance, a modest
>improvement can make a large difference in results.
>A guy (or gal) with 30% more power than me is going
>to crush me in every race, not beat me 1.3x as much
>as random (i.e. beat me 65% of the time).
[snip]
Dear Ben,
A less subtle effect is that power alone may not mean much.
Will a heavier rider with 30% more power crush a lighter rider
climbing?
That is, will a large, powerful rider like a Boonen crush a smaller,
less powerful rider like a Pantani up the Alp d'Huez?
I'm only guessing about their weight and power, but I hope that
they'll do to illustrate the point.
Cheers,
Carl Fogel
Sandy
Sweet sarcasm, but you still don't know shit about racing. Not to mention,
your prose is actually an incorrect statement of the underlying facts. Go
ahead, amuse us some more. Not that I'll be reading you again.
Frank Krygowski
wrote:
> On Feb 28, 7:03 pm, Tim McNamara <tim...@bitstream.net> wrote:
>
> > "b...@mambo.ucolick.org" <bjwei...@gmail.com> wrote:
>
> > > To be clear, I should have called this a "10% improvement in results"
> > > where by results I mean rank-order of race finishes.
>
> > You mean, placing 9th instead of 10th, 18th instead of 20th, 36th
> > instead of 49th, and so on?
>
> No. I'm talking about probability of finishing higher.
> In this case, to be simple, I'm only comparing
> two racers.
IOW, you're removing racing's always-present and critical randomness
and chaos in order to generate a problem you can solve to your
liking.
If they are equally matched, A beats B
> 50% of the time. When I say A improves by 10% in
> results, I mean his chance of beating B improves
> by 1.1x, so he beats B 55% of the time (1.1 x 50%).
>
> Whether this means he improves his rank order
> of finish by one place, two places, or whatever
> would clearly depend on how many other people
> there are in the race and how close they are in
> ability....
... and VERY much else!
> > > I want to keep distinct the percentage improvements in ability (such
> > > as watts/kg, or speed on climbs), finishing time, and rank ordered
> > > results, since they are frequently not linearly related.
>
> > "Not linearly related" = "influenced by random events," which has been
> > one of Frank's main points all along.
>
> No no no!!! That's not correct! You are conflating
> two ideas here.
>
> 1. The first idea is that quantities may be related by
> a non-linear relation. For example, speed on a climb
> is nearly linear in watts/kg ("nearly" because there's
> a small amount of air resistance too). But speed on
> the flats goes as watts^1/3, because the power to
> overcome air drag goes as speed^3. That's non-linear.
It's non-linear in a way that works against your argument. IOW, a 5%
reduction in Cd, if possible, will yield _less_ than a 5% increase in
speed. (And I think even 5% Cd reduction is very unlikely without
tremendous expense.)
> A more subtle non-linear effect is that because riders
> are often closely matched in performance, a modest
> improvement can make a large difference in results.
> A guy (or gal) with 30% more power than me is going
> to crush me in every race, not beat me 1.3x as much
> as random (i.e. beat me 65% of the time).
30% power difference is not "modest." You're talking about two riders
who are in different leagues - or more likely, two different racing
categories. Anyone telling the slower rider that he may make up the
difference via equipment is selling crack.
> 2. The second idea that you refer to is that race
> results can show variations. We wouldn't know how
> much of this is stochastic (truly uncorrelated variations)
> and how much is systematic (due to unmodeled
> effects, e.g., certain riders do better on hillier courses).
> But that's too deep for our problem.
What you're hinting at is what I've been saying all along - although I
think that most of us do know that a very large amount of the
variations are random. And many of the remaining variations are
better attacked as strategy problems, rather than equipment problems.
> Frank says more than that there is randomness.
> He says that the randomness is so large that
> it makes any small improvements negligible.
> What I'm trying to drag him into seeing is that:
> 1. this would also imply that small improvements
> in ability due to training are negligible
I'll buy that. For a certain value of "small," of course.
> 2. the size of what he considers negligible
> (an effect that it takes lots and lots of races to
> demonstrate at 3 sigma) is actually a
> percentage effect that many people would
> like to have, for example beating your buddy
> 55% of the time instead of 50%.
I know of no evidence that the small equipment changes touted here
will produce even that small effect. And I propose that if my hidden
weight experiment is run for a season, you won't observe such an
effect.
- Frank Krygowski
Dan O
> On Feb 28, 10:18 pm, "b...@mambo.ucolick.org" <bjwei...@gmail.com>
> wrote:
>
> > On Feb 28, 7:03 pm, Tim McNamara <tim...@bitstream.net> wrote:
>
> > > "b...@mambo.ucolick.org" <bjwei...@gmail.com> wrote:
>
> > > > To be clear, I should have called this a "10% improvement in results"
> > > > where by results I mean rank-order of race finishes.
>
> > > You mean, placing 9th instead of 10th, 18th instead of 20th, 36th
> > > instead of 49th, and so on?
>
> > No. I'm talking about probability of finishing higher.
> > In this case, to be simple, I'm only comparing
> > two racers.
>
> IOW, you're removing racing's always-present and critical randomness
> and chaos in order to generate a problem you can solve to your
> liking.
>
In randomness and chaos, things still average out. A small advantage
that is regularly "lost in the noise" is still always there making a
difference nonetheless, and one of these days it could be the make or
break difference.
It could even wind up a factor in everything falling into place for a
rider to get in the groove (or on a bubble or whatever) for an
extended stretch.
I don't think some of you guys know what I mean when I say "on a
bubble". It's a very highly ethereal condition where you are simply
very hard to beat - even by competitors with apparently demonstrable,
quantifiable, technical advantages. It's not just mental. The
complete set of circumstances that produces this state - the recipe -
cannot be defined, AFAIK. Otherwise people would just whip up
everything on the list, and there would be no such thing.
But by all means take it or leave it for your own self. I just think
it's foolish to assume that - just because some particular thing seems
unimportant to you - it can't possibly be what lets me beat you.
Tim McNamara
<589a4f0e-cca9-4ee6...@r29g2000vbp.googlegroups.com>,
"b...@mambo.ucolick.org" <bjwe...@gmail.com> wrote:
> On Feb 28, 7:03 pm, Tim McNamara <tim...@bitstream.net> wrote:
> > "b...@mambo.ucolick.org" <bjwei...@gmail.com> wrote:
> >
> > > To be clear, I should have called this a "10% improvement in
> > > results" where by results I mean rank-order of race finishes.
> >
> > You mean, placing 9th instead of 10th, 18th instead of 20th, 36th
> > instead of 49th, and so on?
>
> No. I'm talking about probability of finishing higher. In this case,
> to be simple, I'm only comparing two racers. If they are equally
> matched, A beats B 50% of the time. When I say A improves by 10% in
> results, I mean his chance of beating B improves by 1.1x, so he beats
> B 55% of the time (1.1 x 50%).
That's clearer, thanks.
> Whether this means he improves his rank order of finish by one place,
> two places, or whatever would clearly depend on how many other people
> there are in the race and how close they are in ability. For people
> at the back of the pack, none of this matters. But suppose we are
> talking about the few fast local guys who are duking it out for the
> top three places in races in a series. I'm sure you've seen this in,
> say, a local cyclocross series - in a given category there will often
> be a few closely matched people who are in close contention for the
> individual races and the overall. For these people, improving their
> chances of victory over even odds would be important.
>
> > > I want to keep distinct the percentage improvements in ability
> > > (such as watts/kg, or speed on climbs), finishing time, and rank
> > > ordered results, since they are frequently not linearly related.
> >
> > "Not linearly related" = "influenced by random events," which has
> > been one of Frank's main points all along.
>
> No no no!!! That's not correct! You are conflating two ideas here.
>
> 1. The first idea is that quantities may be related by a non-linear
> relation. For example, speed on a climb is nearly linear in watts/kg
> ("nearly" because there's a small amount of air resistance too). But
> speed on the flats goes as watts^1/3, because the power to overcome
> air drag goes as speed^3. That's non-linear.
That's a consistent ratio and predictable, which is still a linear
concept.
> A more subtle non-linear effect is that because riders are often
> closely matched in performance, a modest improvement can make a large
> difference in results. A guy (or gal) with 30% more power than me is
> going to crush me in every race, not beat me 1.3x as much as random
> (i.e. beat me 65% of the time).
And yet the weaker rider does win sometimes and does not lose every
time, because racing is not deterministic. If it were, there would be
no reason to race. Just measure the rider's power, weight, rolling
resistance and aerodynamic drag and then plug the numbers into
analyticcycling.com and learn who won.
And yes, the more closely matched riders are in ability the more impact
small factors are going to have in the outcome. So will the impact of
"randomosity" on the outcome- pebbles on the road, slower riders getting
in your line, reaching down for your water bottle at the moment the
decisive attack is launched, etc. etc.
> 2. The second idea that you refer to is that race results can show
> variations. We wouldn't know how much of this is stochastic (truly
> uncorrelated variations) and how much is systematic (due to unmodeled
> effects, e.g., certain riders do better on hillier courses). But
> that's too deep for our problem.
It's the core of the problem. Do the small gains from, say, using tires
with a Crr that is .001 lower overcome the normal random events that
influence outcome?
> Frank says more than that there is randomness. He says that the
> randomness is so large that it makes any small improvements
> negligible. What I'm trying to drag him into seeing is that: 1. this
> would also imply that small improvements in ability due to training
> are negligible 2. the size of what he considers negligible (an
> effect that it takes lots and lots of races to demonstrate at 3
> sigma) is actually a percentage effect that many people would like to
> have, for example beating your buddy 55% of the time instead of 50%.
Small improvements from any source that do not break the threshold of
significance are going to be negligible in terms of reliably affecting
the outcome.
> If Frank does the error analysis that I keep asking for, for his
> proposed experiment of switching an effect (tires or whatever)
> between riders A and B, for an effect that gives a winning percentage
> of say 55% or 60%, he'll see how many races it would take to
> establish the effect at 3 sigma or 95% confidence or whatever
> statistical level one likes. And that will give some idea of whether
> this experiment would be likely to be useful.
This seems like supply side economics- the answers look nice so they
must be true. Are you proposing that changing tires can give a "winning
percentage" of 55-60%? What if your competition changed tires, too?
Tim McNamara
<05787acd-9cb8-4ab6...@e36g2000prg.googlegroups.com>,
Dan O <danov...@gmail.com> wrote:
> In randomness and chaos, things still average out. A small advantage
> that is regularly "lost in the noise" is still always there making a
> difference nonetheless, and one of these days it could be the make or
> break difference.
Could be. Might not be. Do you want your success or failure hanging on
such vagaries?
> It could even wind up a factor in everything falling into place for a
> rider to get in the groove (or on a bubble or whatever) for an
> extended stretch.
>
> I don't think some of you guys know what I mean when I say "on a
> bubble". It's a very highly ethereal condition where you are simply
> very hard to beat - even by competitors with apparently demonstrable,
> quantifiable, technical advantages. It's not just mental. The
> complete set of circumstances that produces this state - the recipe -
> cannot be defined, AFAIK. Otherwise people would just whip up
> everything on the list, and there would be no such thing.
I know the feeling you mean. It *is* mostly psychological, IMHO, with a
dash of biological (better sleep, better food choices, etc.). The idea
that a Crr .001 lower gives this feeling is just too "Princess and the
Pea" especially since we very quickly accommodate to those changes and
they become invisible.
> But by all means take it or leave it for your own self. I just think
> it's foolish to assume that - just because some particular thing
> seems unimportant to you - it can't possibly be what lets me beat
> you.
The problem is that you're trying to get a negative proved, which is
logically impossible. The onus is on proving the positive factor:
*this* is what won the race."
Interestingly enough, pro cyclists almost never credit or blame their
equipment in a victory- unless something actually broke or otherwise
failed (puncture, snapped steerer, etc.). "I won today because of my
tires" is not something I've ever heard on the podium, but lots of
amateurs blame their failures on their equipment. As the saying goes,
it's a poor workman who blames his tools.
Dan O
> In article
> <05787acd-9cb8-4ab6-9e7e-eb0aa5de5...@e36g2000prg.googlegroups.com>,
> Dan O <danover...@gmail.com> wrote:
>
> > In randomness and chaos, things still average out. A small advantage
> > that is regularly "lost in the noise" is still always there making a
> > difference nonetheless, and one of these days it could be the make or
> > break difference.
>
> Could be. Might not be. Do you want your success or failure hanging on
> such vagaries?
>
There are not guarantees in life :-) FWIW, I do pay attention to the
demonstrable factors, too. But once you've put together what seems
the optimum package, then what? Do you call it good and sit on your
hands? Or do you diddle around at the edges with some ideas that
aren't necessarily proven.
> > It could even wind up a factor in everything falling into place for a
> > rider to get in the groove (or on a bubble or whatever) for an
> > extended stretch.
>
> > I don't think some of you guys know what I mean when I say "on a
> > bubble". It's a very highly ethereal condition where you are simply
> > very hard to beat - even by competitors with apparently demonstrable,
> > quantifiable, technical advantages. It's not just mental. The
> > complete set of circumstances that produces this state - the recipe -
> > cannot be defined, AFAIK. Otherwise people would just whip up
> > everything on the list, and there would be no such thing.
>
> I know the feeling you mean. It *is* mostly psychological, IMHO, with a
> dash of biological (better sleep, better food choices, etc.). The idea
> that a Crr .001 lower gives this feeling is just too "Princess and the
> Pea" especially since we very quickly accommodate to those changes and
> they become invisible.
>
I have already acknowledged that mentality is probably the most
significant factor. Here I'm just saying that the little things that
lend to mentality aren't all imaginary.
When I say "on a bubble", it's not a feeling that I'm describing
(though state of mind is usually a key ingredient). It's an actual
state of affairs where you are actually very hard to beat (sometimes
practically unbeatable).
> > But by all means take it or leave it for your own self. I just think
> > it's foolish to assume that - just because some particular thing
> > seems unimportant to you - it can't possibly be what lets me beat
> > you.
>
> The problem is that you're trying to get a negative proved, which is
> logically impossible. The onus is on proving the positive factor:
> *this* is what won the race."
>
I'm not trying to prove anything. I'm just disagreeing with those who
would flatly say, "that won't help", unless they can see a controlled
measurement of its significance.
I do appreciate physics and the value of understanding things. I'm
sorry if my words here are working at odds with science.
> Interestingly enough, pro cyclists almost never credit or blame their
> equipment in a victory- unless something actually broke or otherwise
> failed (puncture, snapped steerer, etc.). "I won today because of my
> tires" is not something I've ever heard on the podium...
I have certainly credited my tires with helping me win races, but
probably never passed up an opportunity to credit my own skill,
too ;-)
> ... , but lots of
> amateurs blame their failures on their equipment. As the saying goes,
> it's a poor workman who blames his tools.
Absolutely - especially since, IMO, the greatest racers are rider,
builder, and mechanic.
Bret
This is true. You can't win a mass start race without sticking your
nose in the wind for some amount of time. It might be seconds or the
it might be a long solo effort. When that time comes it is like a time
trial and small aerodynamic advantages are important.
Bret
bjwe...@gmail.com
No. I've generalized the problem to any kind of improvement,
including training. I'm also talking about percentage success
rather than physical models of tires or whatever because they
make the probability calculations simple.
I'm doing this because Frank wants A/B testing
of improvements. I'm trying to get him to do the calculation
of error analysis, because it will show that A/B testing over
the course of a typical amateur race season (even 40 races,
which is a lot) is not sensitive enough to detect an improvement
from 50/50 results to the 55/45% or 60/40% level at, say,
95% confidence. There just aren't enough races in a season
to get the statistics needed.
The conclusion we should draw from that is not that
a 55/45% or 60/40% results improvement is worthless.
If I told you that you were evenly matched with me, but if
you did an hour extra of intervals every week, you could
beat me 60% of the time, would you do the intervals?
Most riders would find that a good motivation. Frank's
proposed experiment wouldn't be sensitive enough to show
whether the improvement happened. We could draw
from this that training more is a negligible advantage,
but that would be foolish. It would be better to conclude
that the experiment isn't good enough to detect an
effect, even though the effect is one most riders would
be glad to have.
Ben