How one vote can make a difference in Mitch's polls - and other Mitch steel poll thoughts
David H.--REMOVE "STOPSPAM" to reply
difference for coasters that not many people have ridden. Especially in
the comparisons with other coasters that haven't been ridden much,
A good example. Now that the steel results have been finalized, I checked
out the site again. Since I voted nearly at the last minute, my ballot
didn't get posted on the list until later, and I forgot to check it when it
did.
As a result, I didn't catch a bunch of name changes on the ballot over the
past couple of years that I haven't fixed. I mostly use my previous
ballot, and add in new rides, rearrange a few ones and update all of the
lines I missed. And I didn't update my ballot well last year. So there
were a bunch of changes I've missed, including a few that I've apparently
missed for several years.
Most of them don't make much of a difference to the results.
But one big one is one of my favorite coatsers: Pyrenees. Apparently, at
some point, they changed the name on the ballot to PyreNees, to denote the
squiggly line above the "n". Unfortunately, I never caught that. So, I
have it listed with a lower case "n", which caused that line not to get
counted by the automatic program which processes the ballots.
I'm a big B&M fan, especially of their inverted coasters. While Pyrenees
ranks at #11, I have it at #3. So, this could change some matchups.
particularly with other coasters with few riders.
Ironically, it mainly ends up changing the rankings of it and other B&M
inverteds -- but with one big exception!
If my vote for Pyrenees had been counted, it would have tied with Nemesis
(which it lost to 6-7). That could have given Nemesis one less win and one
more tie and Pyrenees one more tie and one less loss.
With Pyrenees, this would give it a 341-10-1 W/L/T score, which would put
it as a tie with Millennium Force (with a slightly higher overall score,
because there were fewer valid comparisons with lesser ridden coasters.)
That would have put the head to head battle as the tie-breaker. On the
spreadsheet, it says that MF lost to Pyrenees 4-7. My vote would have made
it an even more definitive 4-8.
So, one capitalized letter cost Pyrenees the chance to knock perennial
favorite MF out of the top 10!
And that would only make 2 out of the top 10 coasters on the planet in the
US. I find that interesting since the general opinion among
non-enthusiasts has always been that most of the best coasters in the world
are here in the US.
When I tell guests in my restaurant from other countries that I've been to
their country to ride their roller coasters, they're always surprised and
often ask why, since surely all of the best ones are here and their
county's coasters must have bored me.
It also shows the biggest strength of this poll: it's a lot fairer to the
coasters outside of the US that the popularity polls are. Think about that
result. 8 of the top 10 steel coasters in the world aren't in the US. No
other poll comes close to showing that.
In addition, Nemesis' loss to Pyrenees would have added one more tie to
Nemesis' score, which would have put it in a tie with Katun. Since Nemesis
lost to Katun 8-15, Katun would have switched spots with Nemesis in the
final rankings, moving up a notch to #7. Less symbolic than knocking
something out of the top 10, admittedly, but still nice for the park.
Well, there's nothing I can do about the rankings right now, other than to
fix my ballot so that next year's version is right.
But next time someone brags that Millennium Force is a top 10 coaster,
point out to them that's only because of a one letter glitch!
I also found it interesting to note that Nemesis lost significantly to both
of it's sister B&M inverteds, yet did much better than either of them in
the overall rankings. Nemesis had a LOT more riders than either Katun or
Pyrenees. I'd have thought that those who go so far out of their way to
ride them would be bigger B&M fans and would have ranked them more highly.
but apparently Nemesis is more of a general crowd pleaser. I'd be
interested to hear the opinions of people who've ridden 2 or three of
those.
Personally, while Nemesis may have had a bit more extreme positive G's, as
well as having the really cool theming and the very tight walls around it,
as an older B&M, it's more repetitive. Both Katun and Pyrenees have a
nicer set of elements overall. And I think that the section before the
trim brake on Pyrenees even outdoes Nemesis in the strong G's department.
And whatever happened to Montu? I actually put it above all of these, but
it's dropped to #18.
David H, davi...@STOPSPAMbellatlantic.net, Boston, MA
PLEASE remove "STOPSPAM" from my address when replying via e-mail.
"Patriotism means to stand by the country. It does not mean to stand by
the president or any other public official, save exactly to the degree
in which he himself stands by the country. It is patriotic to support
him insofar as he efficiently serves the country. It is unpatriotic not
to oppose him to the exact extent that by inefficiency or otherwise he
fails in his duty to stand by the country. In either event, it is
unpatriotic not to tell the truth, whether about the president or
anyone else."
-- Theodore Roosevelt
Wolf
"David H.--REMOVE "STOPSPAM" to reply" <davidhhh...@bellatlantic.net>
wrote in message news:2j1mk5hekl2vf5dj9...@4ax.com...
> It also shows the biggest strength of this poll: it's a lot fairer to the
> coasters outside of the US that the popularity polls are. Think about
> that
> result. 8 of the top 10 steel coasters in the world aren't in the US. No
> other poll comes close to showing that.
We also have to remember the other trend in Mitch's poll -- remote rides
tend to get a ratings bump so long as they have few enough riders.
Basically, there's a reverse Not Invented Here syndrome.
We saw it with Eagle's Fortress, TdZ, Balder, etc. Rides that dropped in the
polls after hitting near the top early or with few riders. Basically, these
rides get a temporary bump due to the monoculture that rode them. If a ride
had 5 people who all loved it (and only 5 riders), it's unstoppable. But
it's harder to get that uniform opinion with 50 riders.
--
|\-/|
<0 0>
=(o)=
-Wolf
Jeremy
all the mechanics that go into Mitch's poll, but isn't the opposite
true that one vote could make absolutely no difference if you've only
been to well-visited parks, even if you order your ranking really
radically? David, in your example, that vote would have made a huge
difference because you were one of the few people that had a chance to
ride both of those coasters, but if someone has been on less than
fifty coasters at the most popular parks in the states, then unless
there already just so happens to be a tie between two closely ranked
rides might they ever have the possibility of making a difference. If
every coaster they've been on would have more than a one vote
advantage over every other coaster (which I don't see as being
particularly unlikely as long as you're dealing only with rides that
all have 100-200 votes each) then their ballot is basically trash no
matter what they happen to rank first or last (i.e. finding a top
tener really overrated or loving a frequently hated ride, a situation
I frequently find myself in and no I'm not just referring to the
Beast).
By extension of this, let's assume that 100 people have all been on
coaster A, and 51% hated it and ranked it last, while 49% loved it and
ranked it first. Also assume that, on average, they have all been on
the same number of the same coasters (i.e. of that sample people that
have been on both coaster A and B also split along this 51%-49%
divide, and on average those that loved coaster A have been on the
same number of coaster as those that hated it). If I'm interpreting
the polling methods correctly, then wouldn't coaster A finish the poll
in the dead last spot, with the opinion of the 49% completely
disregarded because in every comparison coaster A lost by 49-51? I
don't know, but it does seem to me that it's really hard for 'niche'
coasters that have a strong local following to get much of any footing
in the poll whatsoever. There are rides that I know that, say, 3/4 of
the people that ride it aren't particularly impressed on their first
ride because they're just stopping through to credit it, and since
it's not by a brand name manufacturer they just sort of throw it
somewhere in the middle to middle-low end of their list. Meanwhile the
1/4 of the riders that are locals and really appreciate it all put it
in their top ten, but because on average they're expected to lose 4-1
to every coaster the majority opinion ranked above it, the final
ranking for that ride is exactly where the majority ranked it with
seemingly no impact by the minority voice at all (i.e. it the majority
ranks it in the 50's, minority ranks it in the top ten, it ends up in
the 50's in the final results). The only way to overcome this is by
what was pointed out above, by being an obscure foreign ride very few
other people have been on.
I don't know, to me it seems as though Mitch's poll isn't that great
at ranking the top coasters but instead finding what the majority
tastes happen to be like, and after finding that majority perspective
it then ranks the rides that best that best match those tastes. I
haven't done it this year but a couple years ago I reworked the steel
list by removing all the Intamin and B&M designs, and not only was it
not until the 30's that I even found one coaster that wasn't designed
by either (I think it was Phantoms Revenge) but the end results were a
bit perplexing, pretty nonsensical orderings, at least based on what I
thought the popular opinion of all of these 'alternative' coasters
were.
Also while I'm at it, can someone explain to me how the poll is not
biased to add extra weight to the opinions of more well-traveled
enthusiasts (especially those that have been to obscure foreign
parks). I'm not talking about the ability of a less-ridden coaster to
show up high in the rankings, I agree that that's a good feature to
have. But say if someone goes on 50 more coasters, then not only will
they be able to influence the rankings of those 50 that they
previously hadn't been on, but for each that they had been on
previously, that is now 50 more comparisons that they get to
contribute to for that ride, and therefore 50 more opportunities to
throw a tie-break (i.e. they cause five more losses against that
coaster which reduces it's percent of wins-to-loses and therefore
causes its score to drop, potentially below other coasters that this
person hasn't even been on). This influence on that coaster's rating
has nothing to do with their own opinion of the ride, which has
remained unchanged, but simply by virtue of having been on more
coasters, and therefore their opinion of the ride is given more weight
than someone else that hasn't.
Many thanks if anyone can clarify or correct any of my understandings
of Mitch's poll.
David H.--REMOVE "STOPSPAM" to reply
Well, Eagle's Fortress is still #2, but then again, it wasn't open this
year! So, all the the ECC people who went this year didn't get to ride it,
never mind anyone else who went there. Of course, I think that Eagle's
Fortress is more of a matter of one group during one visit REALLY loving
it. (Kind of the opposite of what happened with TdZ, as noted below.)
As for TdZ and Balder, I think that both of these coasters were more
victims of ride deterioration than anything else.
When Ace went to Europe 6 years ago, TdZ was horribly rough. And by rough,
I mean jackhammer bad, not rough and tumble good. Many ACErs only rode
once or twice, even though we had ERT on the ride and everyone had to wait
for the busses anyways. You know a ride is rough (or boring, which isn't
the case with TdZ) when ACErs who traveled overseas to ride would rather
sit on the sidelines and watch than actually ride the coaster! Luckily, I
liked what the coaster did enough to enjoy it despite the roughness, and as
such, rank it higher than most people have since then. But it definitely
would have ranked higher if it had been smoother. And I've heard that it
hasn't gotten any better.
As to Balder, I've heard from multiple sources that the park has modified
the ride, and that it's not as good as it used to be.
So, while your point probably has merit, these probably aren't the best
examples.
And I also think that your theory also usually applies to most new rides
anywhere.
Richard Bannister
<davidhhhSTOPS...@bellatlantic.net> wrote:
> Well, Eagle's Fortress is still #2, but then again, it wasn't open this
> year! So, all the the ECC people who went this year didn't get to ride it,
> never mind anyone else who went there. Of course, I think that Eagle's
> Fortress is more of a matter of one group during one visit REALLY loving
> it. (Kind of the opposite of what happened with TdZ, as noted below.)
I didn't vote this year, but I will say that Eagles Fortress is my
number one coaster (out of 1482, or 1565 if you count powered). It
really is/was that good. It's a shame that its position is clouded by
controversy over one particular club visit, much like T-Express in
last years wood poll.
(I don't actually rate T-Express that high myself, preferring Voyage,
but that's another story!)
Richard
BaSSiStiSt
wrote:
> I didn't vote this year, but I will say that Eagles Fortress is my
> number one coaster (out of 1482, or 1565 if you count powered).
You've been on 83 powered coasters????
Funtype
Jealous?
I smell a new PC cre-whore in the making.
--CM
BaSSiStiSt
> Jealous?
>
> I smell a new PC cre-whore in the making.
Nah, I don't count them. More power (yuck yuck yuck) for thems that
do. I'm just amazed by the total.
But then again, Richard's count is pretty staggering no matter how you
look at it.
David Sandborg
<c7cb3aa1-f0a9-4b15...@s31g2000yqs.googlegroups.com>,
Richard Bannister <richard....@gmail.com> wrote:
> On Jan 12, 8:56�am, "David H.--REMOVE \"STOPSPAM\" to reply"
> <davidhhhSTOPS...@bellatlantic.net> wrote:
> > Well, Eagle's Fortress is still #2, but then again, it wasn't open this
> > year! �So, all the the ECC people who went this year didn't get to ride it,
> > never mind anyone else who went there. �Of course, I think that Eagle's
> > Fortress is more of a matter of one group during one visit REALLY loving
> > it. �(Kind of the opposite of what happened with TdZ, as noted below.)
>
> I didn't vote this year, but I will say that Eagles Fortress is my
> number one coaster (out of 1482, or 1565 if you count powered). It
> really is/was that good.
I never quite found out...what is it that makes this coaster so good?
Maybe I suffer from a failure of imagination, but I find the suspended
genre limited enough that I simply can't picture even the ultimate
suspended coaster cracking my top ten. (In a similar way, Blackpool's
bobsled is pretty amazing for that kind of ride, but it certainly isn't
going to displace any of my really top steel coasters simply because as
good as that kind of coaster gets it just is incapable of doing the kind
of things that will boost it that high on my list.) Is Eagle's Fortress
just wilder than other suspended coasters or does it do something
genuinely different that no suspended coaster has done before? Put as
many quick direction change turns and fast helixes--which are the best
thing I can think of for a suspended coaster to do--and I still can't
picture it being so amazing as that.
--
Dave Sandborg
Remove Spam-away to respond via e-mail.
David H.--REMOVE "STOPSPAM" to reply
I've been wondering the same thing.
Especially when you consider that the swinging takes away most of the
lateral G's.
David H.--REMOVE "STOPSPAM" to reply
wrote:
>Now correct me if I'm wrong because I don't think I fully understand
>all the mechanics that go into Mitch's poll, but isn't the opposite
>true that one vote could make absolutely no difference if you've only
>been to well-visited parks, even if you order your ranking really
>radically? David, in your example, that vote would have made a huge
>difference because you were one of the few people that had a chance to
>ride both of those coasters, but if someone has been on less than
>fifty coasters at the most popular parks in the states, then unless
>there already just so happens to be a tie between two closely ranked
>rides might they ever have the possibility of making a difference. If
>every coaster they've been on would have more than a one vote
>advantage over every other coaster (which I don't see as being
>particularly unlikely as long as you're dealing only with rides that
>all have 100-200 votes each) then their ballot is basically trash no
>matter what they happen to rank first or last (i.e. finding a top
>tener really overrated or loving a frequently hated ride, a situation
>I frequently find myself in and no I'm not just referring to the
>Beast).
I don't see how one vote SHOULD make a big difference in a coaster that's
been ridden by many people.
That's how almost any poll or election is run.
Your vote for your small town's clerk is going to have much more weight in
the results than will your vote for president.
If 300 people have ridden a coaster, and 200 of them put it in their top
10, and you happen to think that it's a bottom 10 kind of ride, why should
your opinion have any real significance in ranking that ride? Clearly,
your opinion does NOT reflect what most people think of the ride.
If there are a lot of riders of a coaster, then there are a lot more
opinions to consider, and more data points to be included. That makes it
easier to be more accurate in trying to figure out what most people would
think of the ride.
After all, if only 5 people have ridden a coaster, then it's far less
likely that those five people would accurately represent what the typical
enthusiast would think of a ride. However, if 300 people have ridden it,
then it's very likely that you'll get a pretty good estimate of what most
people will think of it. The more people who agree on any opinion, the
more likely it is to represent the most common opinion that most people
would have.
And ultimately, isn't that what any poll is trying to do? Find out what
most people think of something? When it comes to opinions, there are no
hard facts. That's what makes them opinions, rather than facts.
After all, who decides what a truly great ride is? You? Me? Someone who
runs a popular coaster site? All three are likely to have very different
opinions. What makes any one more valid than others? You can only assume
that if a lot of people agree on something, then that's what most people
will think about it.
Nitro is 230 feet tall. That's a fact. It's fairly indisputable (Well, I
guess we could argue about whether you should count from the ground level
or the lowest point or sea level, but you get my drift!)
How good of a ride is it? Well, it depends on who you ask. Many people I
know consider it a top 5 ride, or even a #1. Personally, I think it's fin,
but not terribly exciting. Who's right? We can only assume that if an
opinion is shared by the majority of people who have an opinion on the
subject, that it's the most popular opinion. And that's really all you can
look for in any poll of opinions.
Who's the BEST president in US history? It depends on who you ask!
Who's the most popular president in US history? Well, that can be found by
polling.
>By extension of this, let's assume that 100 people have all been on
>coaster A, and 51% hated it and ranked it last, while 49% loved it and
>ranked it first. Also assume that, on average, they have all been on
>the same number of the same coasters (i.e. of that sample people that
>have been on both coaster A and B also split along this 51%-49%
>divide, and on average those that loved coaster A have been on the
>same number of coaster as those that hated it). If I'm interpreting
>the polling methods correctly, then wouldn't coaster A finish the poll
>in the dead last spot, with the opinion of the 49% completely
>disregarded because in every comparison coaster A lost by 49-51?
But that's an extreme and theoretical example. Can you think of any real
life examples to which this applies?
It's impossible to come up with any polling method that's perfect. Every
method has its flaws. All you can do is TRY to come up with a method that
best considers as many factors as possible in trying to accurately gauge
what most people will think about something.
Mitch's method is not perfect. none is. We know that the fewer riders a
coaster has, the less likely its ranking is to being an accurate
representation of what most coaster enthusiasts might think about it. But
isn't that true of ANY poll? The more data points you have in ANY kind of
poll, the more accurate the poll, or that section of the poll -- will be.
>I
>don't know, but it does seem to me that it's really hard for 'niche'
>coasters that have a strong local following to get much of any footing
>in the poll whatsoever. There are rides that I know that, say, 3/4 of
>the people that ride it aren't particularly impressed on their first
>ride because they're just stopping through to credit it, and since
>it's not by a brand name manufacturer they just sort of throw it
>somewhere in the middle to middle-low end of their list. Meanwhile the
>1/4 of the riders that are locals and really appreciate it all put it
>in their top ten, but because on average they're expected to lose 4-1
>to every coaster the majority opinion ranked above it, the final
>ranking for that ride is exactly where the majority ranked it with
>seemingly no impact by the minority voice at all (i.e. it the majority
>ranks it in the 50's, minority ranks it in the top ten, it ends up in
>the 50's in the final results). The only way to overcome this is by
>what was pointed out above, by being an obscure foreign ride very few
>other people have been on.
If the vast majority of people who ride a coaster think that it's mediocre,
then it's fair to say that it's probably a mediocre ride. Or at least that
the vast majority of people who ride think that it's mediocre. That should
be reflected in any poll's results.
Ideally, every ride should be ridden dozens of times over multiple visits
to get a true reflection of it's overall and typical performance. That's
simply not possible. How many trips to Korea do you have the time and
money for? Can you spend months there so that you can give all rides a
chance to shine?
Do you really WANT to ride coasters again and again that didn't impress you
-- or worse, which HURT you? Maybe you have endless time and money for one
hobby, but few of us do.
Trust me, Gouderix is a ride I never need to ride again to give a fair
ranking to. Unlike most people, I actually gave it a second chance. I was
sorry I did!
>I don't know, to me it seems as though Mitch's poll isn't that great
>at ranking the top coasters but instead finding what the majority
>tastes happen to be like, and after finding that majority perspective
>it then ranks the rides that best that best match those tastes.
Isn't that what ANY opinion poll is?
If the majority of citizens of a state are conservative, then wouldn't a
conservative politician be very likely to be the most popular politician in
that state? Just because a liberal politician in that state happens to be
an outstanding representation of a liberal politician doesn't mean he
should get extra points in polls in that state that he hasn't actually
gotten from the polling.
Just imagine if Fox News decided that because the majority of people in
their office are conservative and don't like Obama that they should
subtract 10 points from any opinion poll about Obama to account for that.
Would that be fair?
>I
>haven't done it this year but a couple years ago I reworked the steel
>list by removing all the Intamin and B&M designs, and not only was it
>not until the 30's that I even found one coaster that wasn't designed
>by either (I think it was Phantoms Revenge) but the end results were a
>bit perplexing, pretty nonsensical orderings, at least based on what I
>thought the popular opinion of all of these 'alternative' coasters
>were.
If Intamin and B&M make coasters that most people love, shouldn't their
coasters be ranked highly?
I simply don't understand why you wouldn't think that that's fair. Or more
importantly, why you think that somehow biasing a poll AGAINST the most
popular coasters would give a more accurate ranking in showing what people
think.
>Also while I'm at it, can someone explain to me how the poll is not
>biased to add extra weight to the opinions of more well-traveled
>enthusiasts (especially those that have been to obscure foreign
>parks). I'm not talking about the ability of a less-ridden coaster to
>show up high in the rankings, I agree that that's a good feature to
>have.
Aren't they the same thing? In general, it is precisely the most traveled
enthusiasts who have ridden the less-ridden coasters.
The only less-traveled enthusiasts who are likely to have ridden the
less-ridden coasters are locals who can't accurately say how their local
coasters compare to the many other coasters in the world.
Say an enthusiast in Korea LOVES Eagle's Fortress and ranks it is #1
coaster. But he's never been outside Korea. Why should his opinion count
when trying to compare it to Bizarro or Millennium Force or even somewhat
local to him Pyrenees if he's never even ridden it?
>But say if someone goes on 50 more coasters, then not only will
>they be able to influence the rankings of those 50 that they
>previously hadn't been on, but for each that they had been on
>previously, that is now 50 more comparisons that they get to
>contribute to for that ride, and therefore 50 more opportunities to
>throw a tie-break (i.e. they cause five more losses against that
>coaster which reduces it's percent of wins-to-loses and therefore
>causes its score to drop, potentially below other coasters that this
>person hasn't even been on). This influence on that coaster's rating
>has nothing to do with their own opinion of the ride, which has
>remained unchanged, but simply by virtue of having been on more
>coasters, and therefore their opinion of the ride is given more weight
>than someone else that hasn't.
Because how can you get a better idea of how a coasters ranks against other
coasters other than by getting more data points? People who have ridden
more coasters provide more data points.
If you've never ridden a coaster, you can't have a valid opinion on a
coaster. Likewise, if you've never ridden a PAIR of coasters, you can't
have a valid opinion on how those two coasters compare.
If you've only ever ridden the coasters of Korea or Ohio, you can only give
an accurate ranking of how good the coasters of Korea or Ohio are IN
RELATION TO THE OTHER COASTERS in Korea and Ohio. You can't have ANY valid
opinion on how those coasters are compared to the rest of the coasters in
the world. Thus, the only valid opinions you can have -- and the only
valid data points you can add to the collection to be considered in the
results -- is in how those coasters rank within the coasters in that area.
That's completely fair.
It's not that the opinion of well-traveled enthusiasts are more important
that those of less-traveled enthusiasts. It's that they have more valid
data points to add to the pool. You may feel that those are effectively
the same thing, and there's some validity to that opinion. But it's good
polling method to try to seek out more data points. The more data points
that a poll has, the more accurate it is likely to be in judging what rides
are liked by most enthusiasts WHO HAVE RIDDEN THEM.
Knowing my tastes in coasters, I knew that I'd love Katun in Italy. I was
pretty sure I'd love it a whole lot more than Millennium Force. But I
haven't been able to ride it until this year. But until I actually rode
it, my opinion on it and them wasn't valid. Is that fair? Maybe I don't
like it, but it's certainly fair.
Now that I've ridden it, my opinion IS valid. That's fair. After all,
what if it wasn't as good as I'd thought it would be? (Or, as in the case
of it's sister in Japan, Pyrenees, it was better than I thought it would
be?)
Or more to the point, if all you've ridden is what are generally considered
to be the 10 worst coatsers in the world, and you rank an enormously
UNpopular coaster as your number 1, should your opinion in that poll be
weighted as highly as someone who has actually ridden that coaster as well
as most of the most popular ones in the world, and hated it?
Let me ask you this: if someone from France had only ridden the coasters of
France and declared Gouderix the best coaster in the world, would you
believe his opinion as much as you'd believe Richard Bannister who's ridden
nearly 1500 coasters and (I think) thinks that Gouderix is crap? Or more
importantly, if 20 enthusiasts who had each ridden 1000+ coasters nearly
all agreed that Gouderix was crap, would you still believe that one
enthusiast's opinion more than those of all of those well, traveled
enthusiasts?
Maybe you would, but I'd call you foolish if you did!
>Many thanks if anyone can clarify or correct any of my understandings
>of Mitch's poll.
I don't think that it's your understandings that are incorrect. But I
think that you want everything to be super fair to every coaster and every
enthusiast. You seem to want the poll to weight underdog coasters and
underdog enthusiasts higher than the actual data points that they have
contributed to the poll. And that is inherently UNfair.
Sure, it sucks that not everyone has the time, money or opportunity to ride
all of the coasters in the world. It also sucks that their are people
starving in the world. I can't do much about either situation, though I
know which one I find more of a tragedy.
But the basic truth is that people who have ridden more coasters have more
data points to add to any poll. And that should be reflected in the
results.
Popularity polls like Amusement Today's Golden Ticket Awards don't account
for that. Or more accurately, they try to account for it by stacking the
poll with more well traveled enthusiasts.
David H.--REMOVE "STOPSPAM" to reply
Personally, I don't think that it should be controversial. If most of the
people who have ridden a non-US coaster were on one particular trip, then
that's all the data you have to work with. And if most of them happened to
love it or hate it, then so be it. That shouldn't be controversial, in my
book.
But the relatively low number of riders/voters for the coaster should be
taken into account, as it should be with any coaster with a low ridership
in the poll.
As with any coaster, the more riders it has, the more accurate its ranking
is likely to be as a reflection of how the majority of coaster enthusiasts
would rank the coaster.
And please DO try to take the poll next year, even if you just give general
rankings to the lower ranked coasters. You really have a wealth of opinion
and experience to add to the poll. You'd add a ton of data points!
If you don't have the time to rank all of them, do what I do. Rank the top
ones. Then give everything else a broader ranking, like 100, 200, 300, 400
or 500. And maybe rank the absolute worst ones, too, if you'd like. Even
if you don't have time to get as specific as giving the other coasters even
a general ranking like I've suggested, just giving them a ranking (like
500, for example) creates valid data points for comparing your higher
ranked coasters to them.
Richard Bannister
wrote:
> I never quite found out...what is it that makes this coaster so good?
Hmm, it's a bit tricky to describe, but I'll give it a go.
As you can see from the photos the ride is on the side of a hill using
natural terrain to great advantage. The layout did not have any
braking other than the final run, meaning that it got faster and
faster all the way to the brakes. After about the first fifteen
seconds the train was already fast enough for the cars to (feel like)
they were swinging to ninety degrees or beyond - and they were still
getting faster and faster. Robb's POV video sort of shows what I mean.
It goes without saying that there's a certain amount of perception
here, and that the actual numbers probably don't line up with that -
but anyway, the experience just felt completely wild, utterly out of
control, and the most thrilling ride I've ever had on a steel coaster
- by far. I suspect, to be honest, that the sensations are/were rather
like what The Bat would have been. Unfortunately I'm a bit young to
have gotten to ride that :)
I guess it's also worth stating what thrills me in a coaster. For me
it's the sheer sensation of speed and out-of-controll-ness (!), which
is probably why I like Millennium Force despite it doing relatively
little (and why I'd rate that over Steel Dragon 2000, which feels just
the same as every smaller Morgan). In the wood world, I can't envisage
any ride beating out Voyage for me.
David H. wrote:
>And please DO try to take the poll next year, even if you just give general
>rankings to the lower ranked coasters. You really have a wealth of opinion
>and experience to add to the poll. You'd add a ton of data points!
True enough, but time is the fire in which we burn - and just you try
creating a ballot for that many rides :)
Richard
Jeremy
<davidhhhSTOPS...@bellatlantic.net> wrote:
> On Tue, 12 Jan 2010 00:13:54 -0800 (PST), Jeremy <jkthompso...@gmail.com>
>
> read more »
I think you missed a lot of my points, I'm not saying anything about
the way that the people voting in the poll should behave, I'm saying
that given the raw data the poll collects, it is not interpreted in an
altogether accurate way and it leaves big of oversights in what
popular opinion as reflected by the ballots actually is, it's just
that it's complicated enough that it's not immediately obvious where
some of those problems occur. The example that you dismissed as
extreme and theoretical, about what would happen if a majority ranks a
ride last and a minority ranks it first (even if it's only a
difference of 1% total population between the two groups), so long as
we're to accept that it could at least be possible, I think does show
a HUGE flaw with the poll. Even if we're never going to get a
situation that clearly defines this problem by itself in real life,
this trend of throwing away minority opinions on a ride (either good
or bad) and letting only the popular opinion determine its final
ranking is still emergent, even if there might be other factors in the
poll that obscure it somewhat so it's hard to pin down. Obviously
there's never going to be an actual divide like that under those
circumstances, but the poll's so complex it can be hard to clearly
identify a particular problem unless you use a hypothetical example
like that. I'm not sure if you understood the implications of that
example either; obviously a poll should rank the rides based on the
most popular, but what happens here is it takes the data (which in
this case was split 49% / 51% so we can more easily see what's
happening to it once it's interpreted than if it was more complex as
in real-life), finds what the majority opinion on the ride seems to
be, and then discard the minority opinion and have the ride
represented in the final rankings only by where the majority opinion
seems to rank it. With some rides this isn't too much of a problem,
but the bigger problem comes with more polarizing rides, this system
will not find the median between the two extremes and will instead
side with one group or the other. Although in actuality it probably
will find a bit of a median, but only because people won't have all
been on the same rides so each side can win comparisons from more
obscure coasters... again this is a problem because it shows that
simply going on other unrelated coasters can have a greater impact on
their weight in the original coaster's position than what they
actually think of the ride in question. This means that, let's say two
people vote think more favorably of a less-popular coaster, while most
other people would rank it halfway down their lists. Person A ranks
this ride around their 30th percentile of coasters ridden, while
person B ranks it in their number 1 spot. Who's going to move that
coaster's final ranking position up more? Normally we would say person
B, no questions asked... however, I forgot to mention that person A
also happened to take a tour of many obscure parks last year, ranking
them all around their halfway point. This resulted in them throwing a
couple of comparisons out of a tiebreak into the coaster's favor, and
therefore person A, who only ranked the coaster in question about in
their top third on their list, will bump it upwards much further in
the final rankings than person B, whose rank of number 1 didn't affect
any comparisons for coasters that are traditionally more popular and
ultimately had no influence on its final position.
David H.--REMOVE "STOPSPAM" to reply
wrote:
>On Jan 14, 7:26�am, "David H.--REMOVE \"STOPSPAM\" to reply"
Again, you're making a mountain out of a molehill here. No method of
polling or tabulating results can completely account for every minor
theoretical example.
And again, saying that a theoretical example shows a "HUGE FLAW" in the
polling and tabulating methodology is hardly fair. Unlike any other poll
on the subject, you can actually see every ballot AND every matchup on
Mitch's site. Show us a real-life example. Otherwise, your theory is just
that: a theory.
Finally, let's be honest here: does it REALLY matter if a mediocre coaster
that most people don't particularly care for ranks as #160 vs #110, where
it perhaps might belong? Does it matter to anyone, other than the most
die-hard, obsessed coaster geek?
Does anyone actually care where coasters below say the top 50 or so or the
bottom 20 or so actually rank? I don't expect to see many parks bragging
about having "The 117th best steel coaster on the planet! -- not counting
all of the Batman clones and ZacSpins separately!"
Jeremy
'issues' people take with other polls such as the GTAs are only
slightly larger but not by much. Mitch's poll is still probably the
best one around just because it provides semi-accurate data on less-
visited rides compared to well-visited ones, but personally that's the
only real defense I think I've ever heard for the poll and it seems
like most of these other issues are remaining unaddressed so right now
I only take the results slightly more seriously than the GTAs and am
certainly annoyed whenever results from the poll are thrown about as
the defacto "official" rankings of rides. Again, this poll performs
very poorly in terms of displaying a rank that utilizes everyone's
opinion, it just figures out what the majority tastes tend to be like
and then throws away the rest of the minority opinions so they have no
weight in the final outcomes whatsoever. As someone who actually tends
to be in the minority opinion for a lot of rides, this poll is
actually much less helpful than the GTAs because it only gives you a
list that the majority opinion would perfectly agree with because
there's no representation of minority tastes at all. For another
hypothetical example of where this can really throw rankings off, say
we have two coasters A and B, the majority ranks them both around the
top twenty, A slightly higher than B. However, everyone agrees that B
is a pretty good ride, while 1/3 of the voters really hate A. A will
still rank above B because that 1/3 minority won't be able to
influence the outcomes of any head-to-head comparisons which will
still win on average by 66%. The only way minorities are given any
sort of platform is when they go out and ride a bunch of obscure rides
without as many data points, then there's a better random chance that
they could somehow end up as the majority on a certain coaster's head-
to-head comparison and start influencing a ride's position that way,
but that's a completely arbitrary system of influence that has no
impact on the actual quality of the rides (this is partly why, when
you strip away the popular 'brand name' rides that make up the top 30
spots and look at the 'niche-appeal' rides like I would tend to do,
the results can appear rather sloppy.)
Actually I think the problem of only taking the results from one
viewpoint of a polarizing ride could be resolved if, instead of
marking each head-to-head comparison as either 'win' or 'lose' (and
then calculating the percent of total wins to losses from there), to
have each reflected as the actual percent of wins vs. losses that it
received. Naturally that would make calculations a bit more difficult
but at least I could vote for a personal favorite in first place and
feel that that would actually do something, because right now for
every coaster that traditionally would rank above it by the majority
standards I know that my ranking will have no affect on the poll
because it's impossible to influence any head-to-head outcomes. I'd be
curious to have that method tested out to see if there would be as big
of a shake-up in the rankings as I predict there could be. Even if it
doesn't have that huge of an effect it would still help the poll to be
more accurate without as great of a risk of certain strands of data
being completely overlooked while others are given weighted
preference. That still wouldn't completely solve the problem of some
ballots being given more weight in the final results than others (and
I don't mean that if you go on an obscure ride that gives you greater
influence on just that ride's ranking, but also adds more weight to
your opinion on popular rides as well), but I suppose some people
would argue that those ballots deserve to be given added weight.
David Sandborg
<59d516a3-ab00-4407...@35g2000yqa.googlegroups.com>,
Jeremy <jkthom...@gmail.com> wrote:
> If these issues only amount to a molehill then I would say the
> 'issues' people take with other polls such as the GTAs are only
> slightly larger but not by much. Mitch's poll is still probably the
> best one around just because it provides semi-accurate data on less-
> visited rides compared to well-visited ones, but personally that's the
> only real defense I think I've ever heard for the poll and it seems
> like most of these other issues are remaining unaddressed so right now
> I only take the results slightly more seriously than the GTAs and am
> certainly annoyed whenever results from the poll are thrown about as
> the defacto "official" rankings of rides.
If the only virtue of Mitch's poll was to allow less well-ridden rides
to come to the fore, that would still be a major victory because it's a
hugely pervasive problem with other major polls (such as Amusement
Today). The results speak for themselves, I think. We can argue over
the relative position of a few rides on Mitch's poll, but overall I
can't think of any other poll that comes close to capturing the general
feelings of well-seasoned enthusiasts.
Mitch's algorithm may not be perfect, but there are solid theoretical
reasons in polling theory why no algorithm can be. I think any adequate
algorithm would have to at least start from the point he does, using
mutual preferences between rides for each individual polled as the raw
data, rather than a simple "#1 = 10 points", "#2 = 9 points", etc.,
system. It's a far richer data set that makes Mitch's poll a much more
sensitive instrument. That it might take a more sophisticated algorithm
to iron out the "problems" you mention is not much of a detriment since
any such algorithm would be far harder to implement and the results far
harder to interpret.
I think it's also worth mentioning that the problems with "popularity
contest" polls were widely remarked upon prior to (and even since)
Mitch's poll came out. They were far from theoretical, whereas the
issues you are trying to raise are ones I don't think have ever come up
in practice at all, and there's no evidence that other polls are any
better at solving them. There are only two issues that I can think of
that have really come up with Mitch's poll in practice over the years.
The first is that a coaster with very few riders can gain extreme
prominence. I think that's going to be true of any poll that gives a
little-ridden ride any chance at all; it's more an issue of small sample
size. The other issue is whether collusion among ballots can alter the
results unfairly, but of course this is really an issue of the integrity
of the data. Any poll is going to come up with bad results if the input
data is flawed in this way.
Wolf
> bottom 20 or so actually rank? I don't expect to see many parks bragging
> about having "The 117th best steel coaster on the planet! -- not counting
> all of the Batman clones and ZacSpins separately!"
>
> David H, davi...@STOPSPAMbellatlantic.net, Boston, MA
Grizzly acknowleges being the worst woodie on the planet.
Rastus O'Ginga
wrote:
> In article
> <c7cb3aa1-f0a9-4b15-8f35-4d34449dc...@s31g2000yqs.googlegroups.com>,
To me, Eagle's Fortress really was that good. It's probably 2 or 3 in
my list behind MilF and a B&M hyper. EF had a lot of unique
atributes. First up was the obvious that it was totally engulfed in
trees. For much of the route, it really was like you were flying
through the trees. You couldn't really see the tracks, and in the
front seat, the view was phenomenal.
Now, I am NOT a violent ride lover. Voyage isn't in my top ten, and
SFNE's Superman is below MilF, so take that as you may. Plus, EF's
big sister across the park is probably just barely in my top ten, and
that's becuase of the first half only. Someone who wants to be abused
by rides won't love EF as much. However, for a suspended, EF was as
violent as it could be. The final couple of swings were completely
going from one extreme bank to the other in a blink of an eye. Like
I've said before, it was sort of like the turn after BBW's second drop
for the whole ride. And I didn't even get a late night ride because I
was friggin exhausted. Those crazy Koreans and their no air-
conditioning culture.
-RO
I'm shocked that Richard has it as number one. I do think that it
gets a little bit of help from being so out of the way, and a coaster
that no one, besides Koreans, get to that easily. However, Richard
has ridden coasters that no one else even knew existed, so that sure
doesn't influence him.
Richard Bannister
> I'm shocked that Richard has it as number one. I do think that it
> gets a little bit of help from being so out of the way, and a coaster
> that no one, besides Koreans, get to that easily. However, Richard
> has ridden coasters that no one else even knew existed, so that sure
> doesn't influence him.
All I can say is that no other coaster has *ever* left me with quite
such a "holy shit!" feeling. In a good way.
Richard
BaSSiStiSt
wrote:
> All I can say is that no other coaster has *ever* left me with quite
> such a "holy shit!" feeling. In a good way.
>
> Richard
I could tell from the video it woulda been at least a top ten
contender for me. I love suspendeds and this one looked like it was
easily the cream of the crop. I'm bummed it appears I just missed
it...we were planning to go last summer.
David Sandborg
<4c0ee08a-82dd-45a5...@q4g2000yqm.googlegroups.com>,
BaSSiStiSt <bassis...@hotmail.com> wrote:
Thanks to Rastus and Richard for their input. I did watch the video and
it certainly seems that it is an action-packed ride. I like the
description of it being like that one BBW turn throughout the ride.
I guess I still have a hard time imagining that a suspended could crack
my top 10. Still seems like it's a limited and flawed genre. (Which
reminds me, does it ever have that hesitation/stutter that I feel on a
bunch of the Arrow suspendeds?) Unfortunately it looks like I'll never
have a chance to find out.
Absimilliard
wrote:
> In article
> <4c0ee08a-82dd-45a5-a3d5-726e08b62...@q4g2000yqm.googlegroups.com>,
I'm the guy who rated it number 1 on the Steel Coaster Poll... I did
ride it during the first TPR and had no expectations whatsoever about
the ride. I had never seen an onride video and had no clue at all what
the layout did.
I got on and after the first ride, I was shocked! Shocked a ride could
be so smooth, yet so powerful, out of control and it even had airtime!
The surrendings are amazing and the layout just amazing. At some
points, your body feels like it just did a complete barrel roll.
I did return to South Korea on my own in 2009 and unfortunately, it
was already closed at that point. I did ride T Express and placed it
as my number 1 woodie, even through it had nothing going for it! My
trip wasn't going so well with the friend I was with, it took forever
to get to the park, there was a one train, 120 minutes wait when we
got there. Yet, even with all those negatives on my mind, it
delivered. Best wooden coaster I have ever been on. I am looking
forward to going back in better circumstances so I can appreciate it
even more.
Theme Park Review
<davidhhhSTOPS...@bellatlantic.net> wrote:
> As to Balder, I've heard from multiple sources that the park has modified
> the ride, and that it's not as good as it used to be.
Not true! It's BETTER! Our rides were so INSANE this past June that
many of us ranked it higher than even T Express. I was shocked at how
good it ran and I certainly didn't think it would beat out T Express
for me!
--Robb Alvey
www.themeparkreview.com
David H.--REMOVE "STOPSPAM" to reply
Interesting to hear that. Just about everyone on the ECC Italy trip and
the ACE Spain trip who'd ridden it more recently had said that it had gone
way downhill.
Maybe that was more from rides in 2008?
Then again, we all have different tastes in coasters, too.
David H.--REMOVE "STOPSPAM" to reply
wrote:
>If these issues only amount to a molehill then I would say the
Again, you seem to have an issue with this poll that NO ONE else have ever
actually argued. You seem to feel that the poll should be designed to make
YOU feel good, rather than to accurately reflect which coasters are
actually considered the best among enthusiasts as a whole.
What I find ironic is that you're almost completely dismissing the poll's
biggest strength: the fact that NO other poll gives any less-ridden coaster
ANY chance to rank even remotely near the top.
Some REAL WORLD -- not hypothetical -- examples:
-- T-Express in South Korea: #3 on Mitch's poll (#1 last year), #46 on the
GTA's
-- Balder in Sweden: #7 on Mitch's poll, #21 on the GTA's
-- Fireball in China: #9 on Mitch's poll, not ranked on the GTA's at all!
-- Aska in Japan (my personal #4!): #10 on Mitch's poll, tied for #47 on
the GTA's -- and I personally supplied 1/6 of the points it got! Without
my ballot, it wouldn't even have ranked in the top 50 that they list
-- Troy in the Netherlands: 13 on Mitch's poll, 35 on the GTA's
-- El Toro in Germany, a new coaster as well -- #20 on Mitch's poll, not
ranked on the GTA's
That's 6 out of the top 20 and 4 of the top 10 ranked wooden coasters on
Mitch's poll that barely ranked at all on the GTA's. How can that be
considered only slightly more accurate?!?
And on the steel poll, it's more of the same, but far worse.
-- Eagle's Fortress: #2 on Mitch's poll, didn't make the GTA's at all
-- Expedition GeForce: #3 on Mitch's poll and top 3 since it opened, #6 in
the GTA's (as more people have finally gotten to it after all these years)
-- Goliath at Walibi World in Germany -- #5 on Mitch's poll, #40 in the
GTA's (Hell, it only came in 4th among all Goliath's on the GTA's!)
-- Piraten in Denmark: #6 on Mitch's poll, not ranked on the GTA's
-- Nemesis in the UK: #7 on Mitch's poll, #20 on the GTA's
-- Katun in Italy (my #4 steel!): #8 on Mitch's poll, #50 on the GTA's
(again, I gave this nearly 1/6 of all the points it got!)
--
And FAR more importantly, like you say if a minority of riders who have NOT
ridden a lot of coasters happen to like a coaster and that coaster happens
to be at a park that the vast majority of the people in the poll have been
to, then it will automatically rank near the top with most other polling
methods, even though the majority of riders do NOT think it should be
there. On the popularity polls, it will automatically rank highly just
because lots and lots of people have ridden it who liked it. It doesn't
matter if the vast majority didn't consider it a top ride. It will still
have more points overall that FAR better coasters.
Take The Beast for the perfect example. It's a perfectly fine coaster.
Lots of fun. Fast and long and wild through the woods. But there's
nothing particularly outstanding about it. And it certainly isn't anything
near a top 10 coaster. I rank it #47 on my list. That's slightly higher
than the #51 it got on Mitch's poll this year, although I haven't ridden 11
of the coasters that rank higher than it on that poll. But that's pretty
standard for how The Beast ranks on the poll. For the last FOURTEEN YEARS,
it has ranked between #43 and #68. So, it's certainly not a fluke.
But the Amusement Today poll -- which you claim is just "slightly" less
accurate, has it ranked at #7. Now, this isn't a hypothetical example.
This is a real world one. And it's not a difference between ranking at #79
and #110, which is what you seem so concerned about. It's a difference
between ranking in the TOP TEN of all coasters in the world in the GTA's
and not even making the top 50 consistently for ten of the last 12 years in
Mitch's poll.
And there is only one reason for this. The Beast has been ridden by more
enthusiasts than almost any other woodie on the planet. The only coasters
with more riders on Mitch's polls are the two at Cedar Point.
Or how about Millennium Force. It's at Cedar Point. Other than on Mitch's
poll -- which suddenly got a TON of people from the UK voting this year --
it's going to have the most enthusiast riders of any park in the US, except
maybe the Disney parks. So, take the most popular ride at the park, and
you've got a guaranteed top coaster on any popularity based poll.
Now, sure, those minority of riders GET THEIR VOICES HEARD, which is your
big concern. But considering that the vast majority of people who have
actually ridden it do NOT thin it's a top ride, how is ranking it HIGHER
than the vast majority of people think it deserves going to make for a more
accurate poll? Maybe if your poll is for the most POPULAR coasters, sure.
But not for the BEST coasters, which is what both polls claim to be
ranking.
It's #2 on AT's GTA's. Yet, it's only #10 on Mitch's poll. And it
wouldn't have even ranked in the top 10 if the lack of the capitol N in
"PyreNees" on my ballot hadn't disqualified my vote for it!
Or how about Magnum? #9 on the GTA's and #69 on Mitch's poll. Sure, I
love it. And I know that a lot of the old-school enthusiast types who tend
to be on the AT survey love it. But it's certainly no longer considered a
top 10 ride. Or even a top 4 ride AT THE SAME PARK!
How about a quick peek at GTA's top 10 steelies vs their spots on Mitch's
poll:
Coaster GTA Mitch
Bizarro SFNE 1 1 (yea! At least these are the same!)
Millennium Force 2 10 !
Nitro 3 19 !!!
Goliath SFoG 4 15 !
Apollo's Chariot 5 16 !
Expedition GeForce 6 2 !
Diamondback 7 4
Phantom's Revenge 8 25 !
Magnum XL - 200 9 69 !!! LOL!
Top Thrill Dragster 10 31 !
How can you say that it's almost as accurate when the top ten steel coaster
lists don't even RESEMBLE each other?!?
And are we surprised that 5 of the top 10 coasters on the popularity poll
are from the heavily traveled coaster area of Cedar Point/King's
Island/Kennywood, which are often visited on the same coaster trip?
Now none of this is meant to actually disparage the GTA's. They serve
their purpose well. They give the industry -- who are their primary
readers -- a sense of what is considered the most popular and best coatsers
and parks in the world. And for the most park, that's exactly what that
poll gives them. Surely The Beast is a tremendously popular coaster in the
coaster world, even if it isn't among the very BEST coasters.
And, to be honest, a top 10 coaster is NOT what the parks covet most in
these awards. Other than the #1 coaster in the world bragging rights
(which Bizarro has posted right in the station), the parks are far more
excited by being listed as the having the best hospitality or food or
shows. Because small parks generally know that they can't afford to build
a #1 coaster, especially in the steel category. But any park (in the US,
at least) can earn a cleanest or friendliest park award if they try hard
enough (and work to get enough enthusiasts at the park!)
What it sounds like to me is that what you're complaining about is that YOU
don't get any big say in Mitch's poll. Well, there are 600 people voting
in it. Of course, you don't get much say!
And you seem overly concerned about how mediocre coasters will rank.
Honestly, who cares? Wouldn't you much rather see a poll that has a pretty
accurate top 10 and top 40 listings, rather than a slightly more accurate
middle 50?
I can almost guarantee you that the people at the parks themselves when the
GTA's are announced, spend about a minute scanning the list to see if their
coaster made the list if it wasn't in the top 10. They were mildly glad to
see that their coaster was recognized at all. Then they moved on and
probably didn't even read anything on the list outside the top 5. And
those are the paying readers of Amusement Today.
What's interesting here is that you continue to speak in generalities and
hypotheticals, rather than in real world examples. Can you name some
examples of coasters that you don't think are fairly ranked on Mitch's
poll? How do they rank on the GTA's? What evidence can you cite to show
why they deserve to be ranked more highly than the nearly 1.5 MILLION data
points in Mitch's poll puts them? Can you back that up by explaining why
it should be ranked higher, even though the majority of enthusiasts
disagree with you? And for evidence, I mean more than "Well, I like it, so
it should rank higher."
Or how about this? Mitch has so kindly posted ALL of the ballots from this
year (and every year) on his site. He's even posted a spreadsheet with the
results of ALL of the head to head matchups -- all 135,000+ of them!
If you think that your method would yield significantly better results, why
not do the calculations yourself? Or write a program that will do it for
you, if you'd rather not deal with nearly 1.5 million data points or
135,000 matchups by hand.
If course you'll also have to deal with how you'd end up raking the top
coasters, since they're likely to have infinitesimal differences in
percentage differences in your rankings.
If your results are significantly different from Mitch's then I'm sure he'd
be interested in maybe applying your methodology to his polls in the
future. But by significantly different, I'm talking about rankings in the
top 20 of more than one or two notches. Again, I don't think that anyone
else cares if a coaster ranks at 142 or 167. Certainly not enough to
completely change the methodology and write a new program to help that poor
coaster pick up those 26 notches it needs to make it into the top 150!
And if he's not interested in changing the polls he's been running for
fifteen years, then at least you'd have the program up and running and
ready to start your own poll.
And finally, could you please explain exactly why you think that the
minority opinion should be accounted for in polls? I just don't get this
idea. A poll is supposed to try to gauge what MOST people think of it. You
don't see McDonalds advertising that they have the #1 restaurant chain in
the world (except for that one guy in the Bronx, who prefers his local
deli!) When elections are held, politicians don't call in those who voted
against them and try to find out why they didn't like them. And polls
should not be trying to find out what a MINORITY of people think. They're
supposed to -- by definition -- find out what the MAJORITY think. That's
the whole point! If you want to know what individual people think, go to
their blogs or facebook pages!
The job of a poll is to show us what are most people consider to be the
best coasters. If you want to know what one guy in Indiana thinks about
it, ASK HIM.
What it sounds to me like you're saying is this: "Most people seem to
disagree with me. Thus if *I* don't get to personally sway the results of
a poll with 600+ participants, the poll 'performs very poorly'." That's
simply selfish.
Perhaps you hate it when people use Mitch's polls to state what are
generally considered to be the best coasters in the world. But when people
do so, they're generally right. And they absolutely are considered to be
the de facto rankings of coasters, even if you personally disagree with
them.
And just a hint here: NONE of us completely agree with the rankings. That's
how majority opinions of a list of 300+ items works. We ALL have different
tastes. And we ALL sometimes agree with the majority opinion.
That's life.
My favorite steel coaster comes in at #18 on Mitch's poll, and #11 on the
GTA's. That doesn't make either poll flawed. It just means that I have
different tastes in coasters than the majority of enthusiasts.
And that's perfectly ok. That's fewer people in line at the next ERT
session! I can live with the fact that it's not "considered" the best
steel coaster, even if I think it is.
David Sandborg
"David H.--REMOVE \"STOPSPAM\" to reply"
> Take The Beast for the perfect example. It's a perfectly fine coaster.
> Lots of fun. Fast and long and wild through the woods. But there's
> nothing particularly outstanding about it. And it certainly isn't anything
> near a top 10 coaster. I rank it #47 on my list. That's slightly higher
> than the #51 it got on Mitch's poll this year, although I haven't ridden 11
> of the coasters that rank higher than it on that poll. But that's pretty
> standard for how The Beast ranks on the poll. For the last FOURTEEN YEARS,
> it has ranked between #43 and #68. So, it's certainly not a fluke.
Well, you're not going to convince him with this argument since he wrote
an essay defending the Beast!
Wolf
> love it. And I know that a lot of the old-school enthusiast types who
> tend
> to be on the AT survey love it. But it's certainly no longer considered a
> top 10 ride. Or even a top 4 ride AT THE SAME PARK!
This is probably more of a generation gap in the voters than an indication
of the sample weighting, which was your point.
And what rides do you put ahead of Magnum at CP? I could see MF and an
argument for Raptor, but what else? And I still have Magnum in my top-10
overall, and I've ridden a sizeable chunk of both the wood and steel
top-20s.
> How can you say that it's almost as accurate when the top ten steel
> coaster
> lists don't even RESEMBLE each other?!?
Here you're just presupposing that Mitch's poll is accurate.
> And you seem overly concerned about how mediocre coasters will rank.
> Honestly, who cares? Wouldn't you much rather see a poll that has a
> pretty
> accurate top 10 and top 40 listings, rather than a slightly more accurate
> middle 50?
Why would it be mutually exclusive? Accurate is accurate.
> If you think that your method would yield significantly better results,
> why
> not do the calculations yourself? Or write a program that will do it for
> you, if you'd rather not deal with nearly 1.5 million data points or
> 135,000 matchups by hand.
Mitch's poll fundamentally uses a sound algorithm. I'm curious what the GTA
sample would generate using Mitch's tabulation mechanism. The issue Mitch's
is prone, too, though, is the monoculture visiting an out of the way park
and generating a unanimous positive or negative ranking. That tends to
bypass the primary strength of the mutual-comparison system, which assumes a
random population of visitors. If that population isn't actually random, a
lot of the validity breaks down.
For instance, for the steel poll, it would be trivially easy to make John
Ivers' Blue Flash the #1 steel coaster.
This is why we have a minimum sample size requirement, and why it keeps
getting adjusted upward. It would be interesting to plot ranking volatility
as a function of voters per ride.
Josh Wozny
book. I'd definitely put that ahead of it. I'm also a fan of MF and
Raptor, but my favorite ride at the point is Maverick, hands down.
Josh
David Sandborg
<b3edeca9-1b0a-4ca8...@j14g2000yqm.googlegroups.com>,
Josh Wozny <josh....@gmail.com> wrote:
> I think Magnum's pretty fun, but Maverick is by FAR superior, in my
> book. I'd definitely put that ahead of it. I'm also a fan of MF and
> Raptor, but my favorite ride at the point is Maverick, hands down.
Funny. I like Maverick in concept, and definitely like that they backed
off from the "build a record-breaker every time" model to "build
something unique and fun". But in execution I can't quite fall behind
it. It's a familiar refrain...the restraints make it much less fun for
me than it should be.
As for Magnum, I think it will always have a special place for me. It
was the king of Cedar Point at the time I first went there as an
enthusiast fresh into the hobby. It blew me away then, and to this day
I still get a feeling of both fun and nostalgia riding it that I get
from few other coasters. If I'd come into the hobby 10 years later I
might well have different feelings about it.
Jeremy
say that it's only slightly more accurate than the GTAs, although I
will say, for all that added accuracy, I personally only find Mitch's
poll to be slightly more meaningful, and only because I can see where
obscure rides place that the GTA never accounts for. The reason I
haven't listed any specific examples is because, 1): this problem only
needs to be shown true in theory to show that it is still a problem,
real-life examples are irrelevant to the proof of the reality of these
problems, and 2): I don't have the time to completely duplicate
Mitch's methods and write a program for me to run all that data just
to prove a point in a usenet discussion. If you insist on real-world
examples, I do agree that the very high rankings of both the Beast and
Magnum on the GTA's (which are both my personal #1 rated wood and
steel) do represent a failing of that poll because in those ranking
are much higher than what the general opinion of those rides seem to
be, and if next year they showed up in the top ten in Mitch's poll I
would probably be one of the first to cry foul play because I know
that's not representative of the actual voter's preferences. However,
based on my own experience of talking with other people and trying to
get a feel for the sentiments other people have towards those rides
(and others), they seem to rank much lower relative to other rides
than they deserve to in Mitch's poll (the opposite is true of many
very highly ranking rides which surprised me because I was fairly
certain I was not the only one with anything negative to say about
them and I never got the sense that the rest of the people that did
like it liked it enough to overcome the minority that didn't and in
the end still rank in the top five... I just happened to be thinking
specifically of Diamondback as I wrote that but we could easily
substitute other examples if we wanted). Maybe the majority opinion
really does dislike those rides just that much that they're able to
drag the rankings down that far to compensate for the small but strong
voices of people that still love those rides, I don't know because
I've got better things to do than read through every single submitted
ballot and see what the results should actually be (that's Mitch's
job). But it did cause me to investigate the polling methods enough to
see if there could be any potential problems with the poll, and that's
what I'm trying to describe, not my own personal opinions about how
the results ought to be arranged, which I do recognize would be a very
easy trap you'd like me to step in so that you can then easily dismiss
the rest of my argument, but I'm not going to so I will now stop
talking about my personal opinions of actual rides and continue to use
hypothetical examples that illustrate general problems with the poll
as a whole, as I had already been doing.
My problem with the poll mostly stems from two issues:
A): The fact that each head-to-head data point is recorded only as
either a win or loss, (of which the final results are calculated by
the total percentage of wins-to-losses for each ride). This means that
up to 49% of the data collected in the poll (and *always* the minority
opinion) is completely discarded and has absolutely no influence
whatsoever in the final results. In an ideal poll, the majority
opinion is given the most weight, but only to the degree of which they
are the majority. Meanwhile the minority is allowed to triangulate
that result only to the degree of the size of that minority (i.e. is
it just one person saying the top ranked ride actually sucks, or is it
1/3 of all riders saying it does?) or the strength of their voice
(i.e. does that 1/3 minority opinion only differ by a few ranking
points or is it ranking a ride #1 that normally comes halfway down the
results). In Mitch's poll, the majority opinion (even if it's only by
51% and barring all the outside 'noise' of different people having
been on different rides, which we can ignore for the sake of this
argument since the law of probabilities should hold that both the
minority and majority opinions have, on average, been on the same
coasters), is first determined, and then that opinion is given 100%
weight. This could easily be fixed, instead of recording each head-to-
head result as either a 1 or 0, record it as the actual percentage of
people that preferred one ride over the other. For example, record a .
66 under the Win column, and a .34 under the Loss column. You see how
that's a pretty dramatic difference and is not me simply splitting
hairs over insignificant details that have no affect over the final
results, correct? I'll repeat it again just to emphasize my point: 49%
of the data collected by Mitch's poll is discarded, and 100% of the
minority opinion on each head-to-head comparison is always part of
that missing data. I don't know what your definition of 'accurate' is,
but that seems like a pretty huge oversight to me and severely calls
into question any pretensions of accuracy associated with the poll.
B): Ballots do not have equal weight in the poll, for the most part
it's fairly random chance on whether or not your ballot will make any
difference in the final results, although your chances of affecting
the outcome of a popular ride improves if you've been on many obscure
coasters. The randomness of the influence could partly be solved by
fixing the way head-to-head comparisons are handled as I noted above
(because right now a ballot's influence in the final results only
depends on how many head-to-head comparisons you're able to tip to the
other side of the fence, and someone who's been on a lot of obscure
coasters has the potential to change a lot more data points to a
popular ride). For an example of this, let's pretend there are thirty
people out there that happen to like CGA's Grizzly. Of that thirty,
twenty think it's the best thing since sliced bread and rank it as
their number one coaster, while ten only think it's okay and place it
about halfway up their list. Obviously the Grizzly's not going to be
moving up too far, if at all in the final results, although it maybe
does have a better chance to finally get bumped up out of the dead
last position. However, who's going to make the bigger difference in
it's final rank? In an accurate poll, the opinion of the twenty who
placed it at number one should help to bump it up maybe an extra spot
or two, but in Mitch's poll, as we've already established, they do
nothing and their votes are discarded because the twenty have only
been on popular rides and as such are unable to break any head-to-head
comparisons. Grizzly remains dead last (as perhaps it should, but this
isn't a question of where rides deserve to be ranked but only of if
the results accurately reflect all of the data). So then, what about
the ten who only placed it in a lukewarm position halfway up their
list? If the twenty that voted it number 1 couldn't shake the Grizzly
from it's lowly position then surely these ten who didn't even like it
that much won't, correct? Wrong, because these ten have also recently
completed a trip to several completely unknown parks, and were able to
turn seven or eight head-to-head comparisons with some meaningless
obscure rides in the Grizzly's favor that it previously would have
lost to (say, it had originally lost 2-8, now it wins 12-8), therefore
giving the Grizzly seven wins in the final percent of wins-to-losses
and as a result the ride is now ranks slightly higher in the final
results. This has absolutely nothing to do with the quality (or lack
thereof) of the Grizzly, even as reflected by the submitted ballots;
the best way for someone to affect the ranking of a well-visited ride
isn't by rating it extremely high or low, it's by going out and riding
a bunch of obscure rides that have nothing to do with the ride in
question at all. Personally, if I'm going to spend all the time it
takes to fill out one of those ballots, I'd like to know I'd be making
a difference in the poll based on which rides I happen to rank first
or last, and it's not much solace knowing that my opinion on these
popular rides only matters once I've been crediting a bunch of obscure
rides no one has been on or cares about.
As for the rest of your arguments, as Wolf already pointed out, your
comparisons of Mitch's poll with the GTAs to show how 'wrong' the GTAs
are is based on the assumption that Mitch's poll is completely
accurate, which is precisely what is being called into question here,
so those points are invalid. You also suggest that my points aren't
relevant because no one else has ever had any problems with Mitch's
poll; again, this is a pretty obvious fallacy that's completely beside
the point of whether or not the poll actually is accurate or not. You
claim my issue with the poll is that I personally don't like the
results and want a system that better favors my preferences, and yet
criticize me for remaining entirely theoretical and not mentioning any
specific examples of rides I'd like to see rank higher or lower. The
reason I've stuck to only describing theory is because I don't know or
care if the actual results are 'correct' or not, only that there is
theoretical potential that they could be up to 49% inaccurate based on
all on the data received from the ballots collected (besides, you
yourself then go on to describe where a few rides 'deserve' to be
ranked based on your own impression of them and cite the fact that
Mitch's poll generally supports those opinions of yours as a indicator
of it's validity). You say I'm upset that I don't get a big enough say
in the final results when there are some 600 ballots. I say that I
expect my opinion to have about 1/600 of an affect on the results, but
as I've already demonstrated I'm fairly confident I don't even get
*that* much of an influence, while other people voting likely get
three to four times that based on random factors such as "how many
obscure credits did you collect last season?" You say only the top
twenty or so positions are what really matters, but nowhere did I say
that any alternate methods would compromise the accuracy of those
'important' positions (in fact they would likely only improve them),
and besides, that's simply begging the question of how do we know that
those top twenty are accurate in the first place, especially if there
are potentially rides beneath them that deserve to be rated much
higher based on the ballots submitted? You continue to cite Mitch's
poll's superior accuracy based on the shortcomings of popularity
polls, but that's also completely beside the point, whether or not
those other polls have awful polling methods does not make Mitch's
poll in and of itself any more accurate, which is what we are
discussing. And finally, you say, "could you please explain exactly
why you think that the minority opinion should be accounted for in
polls? I just don't get this idea." I've already elaborated a bit
more on what is meant when I say how minority opinions should be
represented in a poll in the point A above. This is a pretty
fundamental concept to every sort of democratic polling scenario in
which the outcomes are not binary but can be spread across a spectrum,
such as a roller coaster poll. The final results should triangulate
perfectly to represent the mean average of all of the submitted
opinions. You seem to be suggesting that all minority opinions are, by
definition, bad data, and should be removed wholesale from the poll so
as to not contaminate the purity of the majority opinion, which is
then given the absolute decision in the final rankings whether or not
they are the majority by 51% or 99%. (P.S. your example of asking one
guy in Indiana what he thinks of a coaster as a way to discredit my
points about minority opinions not being fairly represented was a
decent example of a straw man argument.) Actually, after restating all
of these reasons I'm tempted to return to my original proclamation
that Mitch's poll is only slightly more accurate than the GTA's, but
I'm not opening that can of worms again, and will simply leave it at
this: as great of a poll as it may be, I hope I have shown (albeit
without rigorous mathematical proofs, but at least pointed in the
right general direction) that there are still some large inaccuracies
left unaccounted for based on the way it tabulates the raw data that
prevent it from being the "be-all and end-all" of coaster polls many
people purport it to be.
ansley
overlooking. YES, someone who's ridden 600 coasters, say vs. 60, will
and SHOULD have more influence in the poll. Why? Because the avid
coaster rider who has ridden many coasters has far more data points
and therefore have more influence. This makes perfect logical sense
when ranking all coasters. If you've only ridden 60 coasters you will
only have 60 lines and far fewer comparions so your data will not show
up on the data sheet in very many places.
If you're only doing top 10 coasters in a "popularity" poll then the
person with 60 coasters is unfairly voting for 1/6 of his coasters,
where in theory the person who's ridden 600, his number 60 coaster
would equate to number 6 on the other person's poll. All the great
coasters outside of his top ten list have no impact at all!
David H.--REMOVE "STOPSPAM" to reply
And the Beast is a fine ride and certainly has lots of people who love it,
included more than a few die-hard enthusiasts.
However, it's not generally considered a top 10 coaster by the majority of
enthusiasts.
Hence my point exactly.
And that's probably what he wants. When a minority of enthusiasts like a
coaster, he thinks it should allow it to override the majority opinion.
And no decent poll would do that.
David H.--REMOVE "STOPSPAM" to reply
wrote:
>Alright, I'll concede that it might have been a bit of a hyperbole to
>say that it's only slightly more accurate than the GTAs, although I
>will say, for all that added accuracy, I personally only find Mitch's
>poll to be slightly more meaningful, and only because I can see where
>obscure rides place that the GTA never accounts for. The reason I
>haven't listed any specific examples is because, 1): this problem only
>needs to be shown true in theory to show that it is still a problem,
>real-life examples are irrelevant to the proof of the reality of these
>problems,
No. Sorry. Absolutely not. NO.
In theory, I COULD be hit by lighting today.
In the real words, it's highly unlikely.
As such, I'm not staying in all day every day for the rest of my life.
Give me some real world examples of how likely it is that I can do so, and
I'll consider it.
You can theorize all you want about how the poll MIGHT be inaccurate. But
until you can show examples of where it IS inaccurate, then it's simply a
theory.
>and 2): I don't have the time to completely duplicate
>Mitch's methods and write a program for me to run all that data just
>to prove a point in a usenet discussion.
Yet, you seem to expect HIM to do so.
That's hardly fair now, is it?
>If you insist on real-world
>examples, I do agree that the very high rankings of both the Beast and
>Magnum on the GTA's (which are both my personal #1 rated wood and
>steel) do represent a failing of that poll because in those ranking
>are much higher than what the general opinion of those rides seem to
>be, and if next year they showed up in the top ten in Mitch's poll I
>would probably be one of the first to cry foul play because I know
>that's not representative of the actual voter's preferences. However,
>based on my own experience of talking with other people and trying to
>get a feel for the sentiments other people have towards those rides
>(and others), they seem to rank much lower relative to other rides
>than they deserve to in Mitch's poll
If the vast majority of people who ride a coaster don't think it's
particularly great, then I would think that it should rank rather low. Just
because I happen to sometimes disagree with that majority opinion doesn't
change it.
And, for the record, I'd actually agree with you more than disagree with
you on many of these specific examples.
A few off the top of my head, where I vastly agree with the polls:
-- Magnum. It's a favorite of mine, easily my second favorite coaster at
CP (behind only Raptor), and my #33 steel out of many hundreds. (One of
the downsides of being a well-traveled enthusiast is that as you add those
rare great foreign coasters, other very good old time favorites inevitably
get knocked down.) That's nearly 36 notches above where Mitch ranks it.
But I know how it's fallen from favor in many coaster circles. Again, I
don't expect my personal tastes to have much influence on the poll.
-- Montu. It's my personal favorite steel, though I know for sure it's not
everyone's But #18? And I know that non-launched loopers are now out of
favor among many enthusiasts. But ten notches below Katun, which is
similar, but I think inferior? I don't get it, personally. but, hey, if
that's what people think, then so be it.
-- Pretty much all of the B&M hypers. Sure, they're lots of fun, but not
nearly intense enough to be anywhere near my top 10... or my top 20... or
my top 30.. or even my top 40. In fact, other than Apollo, which for some
reason I've always loved, and I rank at 41, non of the other B&M hypers
even show up on my list until the 60's. But I know that I'm the minority
opinion here. They're huge crowd pleasers, both among the masses and among
enthusiasts. And no matter how many coasters I've personally ridden, my
opinion should NOT be able to drag them down in the polls.
-- Phoenix, surely a great ride, and a classic. But top 10 for 8 years in
a row? Hardly. At least in my book. (I rank it at #19, mainly based on
the years when it's been performing at its best.)
-- Tremors, my #2 woodie only ranks at #15. I can't understand that at
all.
-- Legend, poor Legend, the Jan Brady of Holiday World. When it's running
well and tracked well, it's one of my favorite coasters, #3 even. But it
gets no love among enthusiasts.
>(the opposite is true of many
>very highly ranking rides which surprised me because I was fairly
>certain I was not the only one with anything negative to say about
>them and I never got the sense that the rest of the people that did
>like it liked it enough to overcome the minority that didn't and in
>the end still rank in the top five... I just happened to be thinking
>specifically of Diamondback as I wrote that but we could easily
>substitute other examples if we wanted).
Diamondback is one of those coasters that 80- 90% of people seem to
absolutely love. Accept that. I haven't ridden it, but knowing my tastes
in coasters, I imagine that I won't be one of them. I'm sure it will be in
my top 60 or so, which still puts it in my top 10%, which is pretty damn
good. But it's extremely unlikely to come anywhere nearer tot the top of
my personal list any time soon.
>Maybe the majority opinion
>really does dislike those rides just that much that they're able to
>drag the rankings down that far to compensate for the small but strong
>voices of people that still love those rides,
There's no maybe about it.
And again, why should a small. but vocal minority have any major influence
on a poll with 600+ people taking it?
I just don't understand your point. You haven't adequately elucidated WHY
that small minority should have undue influence on the poll.
I mean, they already have some influence. There are lots of enthusiasts
who would rank those coasters much lower than they do rank them. But
there's enough people who DO like them enough to rank them higher that you
don't see them at the bottom of the poll.
I mean, Beast again is the perfect example. On Mitch's poll, it ranks at
51. That's in the top 28% of all coasters. That's pretty damn good.
I actually took a few minutes to look at the first 200 or so ballots which
included the Beast. What's absolutely remarkable about it is how there is
absolutely NO overall consistency in its ranking. It's really all over the
board. Probably the most polarizing coaster in existence. But one thing I
noticed was that most of those who ranked it in the top 10 and especially
the top 5 were people who'd only ridden 20-40 coasters. In fact, I only
saw a SMALL handful of people who ranked it in the top 10 who'd even ridden
50 or more coasters. (And that still puts it at around the 20th
percentile, which is pretty close to where it scores in Mitch's poll.) And
even among those who ranked it 3 or 4 or 5, the vast majority have only
ridden 20-30 coasters max. Again, that STILL puts it in around the 20th
percentile. And there were a whole lot of people who ranked it in the
bottom half of their rankings, and quite a few who ranked it in the bottom
25% or so, or even the bottom 10%.
But if there's any consensus that I could see just by looking at where most
people seemed to rank Beast on their ballots is that the vast majority seem
to think that it's a very good -- but not great -- coaster. While the
overall ballots seem to place Beast all across the board, most of them seem
to rank it somewhere at somewhere around the 20-35 percentile of all of the
coatsers that they ranked.
And that's what placing at #51 would show. Show me ANY other coaster poll
that actually represents a showing like that. Any. You can't. Because
enough of the people who've ridden 20-30 coasters put it in their top 10
(i.e their 40-50th percentile) that it ranks up lots of points simply by
being at a park and a state that most enthusiasts have been to and being an
above average ride.
>I don't know because
>I've got better things to do than read through every single submitted
>ballot and see what the results should actually be (that's Mitch's
>job).
No, it's not his job. He wrote a computer program that does it.
Again, you're missing the point that there are over a million data points
here. No person could accurately analyze the entire poll.
>But it did cause me to investigate the polling methods enough to
>see if there could be any potential problems with the poll, and that's
>what I'm trying to describe, not my own personal opinions about how
>the results ought to be arranged, which I do recognize would be a very
>easy trap you'd like me to step in so that you can then easily dismiss
>the rest of my argument, but I'm not going to so I will now stop
>talking about my personal opinions of actual rides and continue to use
>hypothetical examples that illustrate general problems with the poll
>as a whole, as I had already been doing.
Hypothetical examples can only point to hypothetical problems. If your
only problem with a poll is hypothetical, than I have to say that's pretty
good!
>My problem with the poll mostly stems from two issues:
>
>A): The fact that each head-to-head data point is recorded only as
>either a win or loss, (of which the final results are calculated by
>the total percentage of wins-to-losses for each ride). This means that
>up to 49% of the data collected in the poll (and *always* the minority
>opinion) is completely discarded and has absolutely no influence
>whatsoever in the final results.
UP TO 49% of the data. but in the real world, it's actually FAR, FAR less.
The vast majority of matchups are NOT anywhere near that close. You can
see that simply by looking at the spreadsheet. But even then, you don't
have to look at the whole spreadsheet. You can just look at the top 12
listed on the results page.
When it comes to coasters that end up ranking very close, you'll usually
see a close matchup. But in anything more than 5-10 notches apart, it's
not even remotely close. That says that the VAST, VAST majority of the
data is NOT being thrown out. El Toro at #2 beat Thunderhead at #12 102
to 24.
But then look at Thunderhead's other matchups. It doesn't even come close
to anything in the top 7. Then, for the coasters ranked in the same range,
from 8 to 16, it comes pretty close. So, yes, for those few matchups, a
lot of data might be being "discarded", as you say. But then once you drop
below that lever, the matchups again aren't even remotely close. For the
rest of the top 100, it beats all of them by at least 3-1 margins, if not
much more. Then once you get outside the top 100, the matchups aren't even
worth looking at because there are so very few people who voted against it.
For instance, in the matchup with Racer at a still respectable 115 rank, it
crushes it 148-2.
So, while theoretically "up to 49% of the data" may be being discarded, in
reality, it's more like 2%.
But again, if it's the minority opinion, who cares? I mean, really? Again,
does it really bother you that much that Gwazi ranks at 116 instead of
maybe the 90 it deserves? Honestly?
>In an ideal poll,
There is no ideal poll. Because no poll in existence can account for the
varieties of human thought. Period. All you can do is try to minimize the
randomness as much as possible.
The fact that the exact same poll can be taken at different times and show
quite different results is testament to that fact. And that's true of
every poll in existence. It's also why polls aren't used to decide
elections.
I'm not saying that your idea may not lead to a slightly more accurate
poll. Give it a try, if you'd like.
What I'm saying is that it's likely to be only slightly. Because
ultimately, in the real world, rather than in the hypothetical one, Mitch's
poll is going to be about as accurate as you can get.
>B): Ballots do not have equal weight in the poll, for the most part
>it's fairly random chance on whether or not your ballot will make any
>difference in the final results, although your chances of affecting
>the outcome of a popular ride improves if you've been on many obscure
>coasters.
Why is sit so important for you to feel like you have a personal influence
on the poll?
Rational human beings understand that when they take a poll with over 600
other people taking it, they're likely to have about a .167% influence on
that poll. That's not a hell of a lot.
Admittely, in this poll, a less-traveled coaster enthusist will have
significantly less than a .167% influence on the poll. Sorry, but that's
fair, since they don't have as many coasters to compare. And if they're
contributing fewer data points to the poll, then they are giving us less
information to use in the poll. How is that not fair? And more important,
how is it not a valid polling method?
But who really cares if they have less influence when the influence they
would have if it was completely fair is so tiny.
I mean, do you feel more important than me if you have .32% influence on
the a roller coaster poll, and I only have a .015% influence on it?
Honestly?
It's kind of like spending years fighting for the rights of black lesbian
disabled transgendered "women" named Rose Kline who live in Toledo and work
as plumbers for a company owned by someone who was in Republican
commercials. Sure, it might be the right thing to do. She certainly
deserves the same rights as everyone else. But who cares, besides Rose?
And how much effort is it really worth?
Why are we wasting time pretending the results of hypothetical people
taking a hypothetical poll. We have an ACTUAL poll with ACTUAL ballots
posted right on the site.
Again, hypothetical poll results are utterly irrelevant.
>As for the rest of your arguments, as Wolf already pointed out, your
>comparisons of Mitch's poll with the GTAs to show how 'wrong' the GTAs
>are is based on the assumption that Mitch's poll is completely
>accurate, which is precisely what is being called into question here,
>so those points are invalid. You also suggest that my points aren't
>relevant because no one else has ever had any problems with Mitch's
>poll; again, this is a pretty obvious fallacy that's completely beside
>the point of whether or not the poll actually is accurate or not. You
>claim my issue with the poll is that I personally don't like the
>results and want a system that better favors my preferences, and yet
>criticize me for remaining entirely theoretical and not mentioning any
>specific examples of rides I'd like to see rank higher or lower.
I didn't say you should point out rides you'd LIKE to see ranked higher or
lower.
I said rides that you think deserve to be ranked higher or lower, based on
what ACTUAL enthusiasts think about them and showed on the ACTUAL ballots
that are ACTUALLY posted on Mitch's own site. You've got all the data you
need there to make your argument. You're choosing to ignore it because
it's too much work. Yet, somehow you expect Mitch to change his entire
polling algorithm based on theoretical examples. That's both lazy and
unfair to him.
>The
>reason I've stuck to only describing theory is because I don't know or
>care if the actual results are 'correct' or not, only that there is
>theoretical potential that they could be up to 49% inaccurate based on
>all on the data received from the ballots collected (besides, you
>yourself then go on to describe where a few rides 'deserve' to be
>ranked based on your own impression of them and cite the fact that
>Mitch's poll generally supports those opinions of yours as a indicator
>of it's validity).
There's a lot of theoretical potential that the world will go up in flames
tomorrow, but until someone shows me some real-world evidence, I'm not
cashing in my 401k yet!
A theoretically inaccurate poll does not make an ACTUAL inaccurate poll.
I can argue with you all day that theoretically you are a woman. That
doesn't make you a woman.
>You say I'm upset that I don't get a big enough say
>in the final results when there are some 600 ballots. I say that I
>expect my opinion to have about 1/600 of an affect on the results, but
>as I've already demonstrated I'm fairly confident I don't even get
>*that* much of an influence, while other people voting likely get
>three to four times that based on random factors such as "how many
>obscure credits did you collect last season?"
I say that you want a hell of a lot more than 1/600 of an affect on the
results. Because if that's all the affect you had, it would mabe change
the ranking of one or two coasters by one or two notches. You seem to
think that a few people who agree with you should be able to move coasters
many notches. Which is NOT how 1/600 of an affect on the ballot would
ACTUALLY impact on the ballot.
And you're complaining that you only get a 1/1200 impact on the poll
because you haevn't ridden enough coasters. Most people would say that
that's all you deserve if you haven't ridden enough coasters to accurately
compare them.
And I say that who the hell cares if they have .0005% more impact on
anything in the world? It's irrelevant.
>You say only the top
>twenty or so positions are what really matters, but nowhere did I say
>that any alternate methods would compromise the accuracy of those
>'important' positions (in fact they would likely only improve them),
>and besides, that's simply begging the question of how do we know that
>those top twenty are accurate in the first place, especially if there
>are potentially rides beneath them that deserve to be rated much
>higher based on the ballots submitted?
I said that if you don't like his algorithm, then CREATE YOUR OWN DMAN
POLL.
It's really that simple.
M<itch saw flaws with the way then current polls were run. He decided to
do something about it, rather then just sit on the internet and bitch about
it. He put a lot of work into it. If you don't like what he's doing, then
DO SOMETHING ABOUT IT, besides bitching on the internet. If you're so sure
that your personal ideas are so good, then PROVE IT.
I say that all of your issues combined would have a TINY impact on the
results. And it would NOT be worth the efforts to completely rewrite it
just to move Beast up a notch or three because it would make you feel
better. Because no one cares how mid-range coasters actually rank.
If YOU think that it's so important that the mid-range coatsers rake more
"accurately", then design a coaster poll that does so. Then convince 600+
people to take the poll.
Until you do so, stop complaining about someone who HAS TAKEN THE TIME TO
DO SO.
Calling his poll inaccurate because it could theoretically be inaccurate
without bothering to take the time to come up with even ONE example of
where it's ACTUALLY inaccurate is both lazy and disrespectful for the work
he put into it.
And it's completely unfair to claim that it's not actually accurate without
actually showing any actual examples of where it actually is.
Actually. ;-)
>You continue to cite Mitch's
>poll's superior accuracy based on the shortcomings of popularity
>polls, but that's also completely beside the point, whether or not
>those other polls have awful polling methods does not make Mitch's
>poll in and of itself any more accurate, which is what we are
>discussing.
That's admittedly valid. It's either accurate or it's not. But my point
is that it's more accurate than ANY other coaster poll out there.
And that's pretty damn good.
If you can do better, do so. Honestly, I think it would be fascinating to
see you succeed.
>And finally, you say, "could you please explain exactly
>why you think that the minority opinion should be accounted for in
>polls? I just don't get this idea." I've already elaborated a bit
>more on what is meant when I say how minority opinions should be
>represented in a poll in the point A above.
But they already are. A hell of a lot of people HATE Beast and rank it in
the bottom 10 percentile. But in Mitch's poll, it does NOT rank there. It
ranks in the 27th percentile, which is a pretty good accounting of where
most enthusiasts seem to rank it.
The same seems to hold true for just about every other coatsers I've looked
at on the poll.
>This is a pretty
>fundamental concept to every sort of democratic polling scenario in
>which the outcomes are not binary but can be spread across a spectrum,
>such as a roller coaster poll. The final results should triangulate
>perfectly to represent the mean average of all of the submitted
>opinions. You seem to be suggesting that all minority opinions are, by
>definition, bad data, and should be removed wholesale from the poll so
>as to not contaminate the purity of the majority opinion, which is
>then given the absolute decision in the final rankings whether or not
>they are the majority by 51% or 99%. (P.S. your example of asking one
>guy in Indiana what he thinks of a coaster as a way to discredit my
>points about minority opinions not being fairly represented was a
>decent example of a straw man argument.) Actually, after restating all
>of these reasons I'm tempted to return to my original proclamation
>that Mitch's poll is only slightly more accurate than the GTA's, but
>I'm not opening that can of worms again, and will simply leave it at
>this: as great of a poll as it may be, I hope I have shown (albeit
>without rigorous mathematical proofs, but at least pointed in the
>right general direction) that there are still some large inaccuracies
>left unaccounted for based on the way it tabulates the raw data that
>prevent it from being the "be-all and end-all" of coaster polls many
>people purport it to be.
No, you've only shown that there COULD THEORETICALLY be some inaccuracies.
That's nowhere near the same thing as showing that "there ARE still some
large inaccuracies left unaccounted for" [emphasis mine].
"There are some POSSIBLE inaccuracies left accounted for" is a much more
accurate statement.
These are VERY different statements.
In order to prove something is true, you need to actually prove it IS true,
not that it theoretically COULD BE true. That's why newspapers can't print
theoretical "facts".
And that's why without ANY examples, every argument you make is only
theoretical. And irrelevant.
The thing that you're missing in ALL of this explaining is the fact that
it's MITCH's poll. He spent a lot of time and energy writing the program
and algorithm.
I'm arguing that it's the best poll out there and is quite accurate. You're
arguing that it's not the best theoretical poll out there.
In theory, we could also write a poll that reads everyone's minds and
completely accounts for every possible factor. But until someone actually
does so, it's irrelevant.
If you'd like to see a poll that you think takes far more factors into
effect, THEN DO SO. Criticizing someone who has done so is not
productive.
You complain that everyone takes Mitch's polls' results as the de facto
coaster rankings. That's because it's the single most accurate coaster
poll that exists in the real world.
Until you or someone else comes up with a better one (and I'm sure that
it's possible), that remains true. And until someone does so, Mitch's
polls' results will REMAIN ***THE*** de facto coaster rankings.
It's just that simple.
David H.--REMOVE "STOPSPAM" to reply
>> Or how about Magnum? #9 on the GTA's and #69 on Mitch's poll. Sure, I
>> love it. And I know that a lot of the old-school enthusiast types who
>> tend
>> to be on the AT survey love it. But it's certainly no longer considered a
>> top 10 ride. Or even a top 4 ride AT THE SAME PARK!
>
>This is probably more of a generation gap in the voters than an indication
>of the sample weighting, which was your point.
Actually, I think it's more an effect of the fact that Magnum has been
eclipsed by many more similar rides in the years since it's been built.
(And also many dissimilar coasters, which would still knock it down in the
rankings.)
After all, if you're a big fan of Hypercoasters, there are hell of a lot
more of them to choose from than there used to be.
>And what rides do you put ahead of Magnum at CP? I could see MF and an
>argument for Raptor, but what else? And I still have Magnum in my top-10
>overall, and I've ridden a sizeable chunk of both the wood and steel
>top-20s.
I was saying that in the general opinions, it's fallen. Even at the Point.
Dragster, MF, Maverick and Raptor all are pretty clearly more popular among
both enthusiasts and the GP.
Just look at the lines at the park during coaster events. Magnum late
night rides used to be legendary, leading to HUGE lines at events. Now,
not so much.
Personally, I'm with you. I haven't ridden Maverick yet (probably this
year, before HWN), but I agree that the only coaster I've ridden at the
park I'd put above Magnum is Raptor.
But as much as I love both coasters, which were certainly once in my top
10, they've fallen as I've ridden more and more coasters, particularly so
many newer ones.
>> How can you say that it's almost as accurate when the top ten steel
>> coaster
>> lists don't even RESEMBLE each other?!?
>
>Here you're just presupposing that Mitch's poll is accurate.
>
>> And you seem overly concerned about how mediocre coasters will rank.
>> Honestly, who cares? Wouldn't you much rather see a poll that has a
>> pretty
>> accurate top 10 and top 40 listings, rather than a slightly more accurate
>> middle 50?
>
>Why would it be mutually exclusive? Accurate is accurate.
Well, for one thing, very few people bother to rank the coasters outside
their top 10 or 20 or so (or at least their top 50 or so, especially on the
steel poll), simply grouping the rest to save time. As a result, you can't
get a truly accurate result on how coasters that people can't be bothered
to rank accurately should be ranked. There isn't much meaningful data
there.
And that's fair. Not only do I not have the time or energy to figure out
whether I think that Dahlonega Mine Train is better than Gold Rusher, I
don't care.
And that's the same reason I don't care about the results past 50 or so.
And let's be honest. Just about no one else does either.
That was my point.
This actually makes me wonder if the poll wouldn't be more accurate by NOT
listing all of the low-ranking -- or maybe even medium-ranking coasters. It
would also avoid the issue of less-traveled coasters having a lower
percentage because they don't have enough valid comparisons with low-ranked
coasters.
And it would probably bring a lot more people to fill out the poll,
especially the steel one. Let's face it, the poll ballot is pretty
daunting for both die-hard enthusiasts and newbies alike. The only reason
most of us can be bothered to rank more than 20-30 coasters is because we
can cut and paste last year's ballot!
If it brings people like Richard to take the poll, then that would be
particularly valuable to the date set.
Just something to think about.
>> If you think that your method would yield significantly better results,
>> why
>> not do the calculations yourself? Or write a program that will do it for
>> you, if you'd rather not deal with nearly 1.5 million data points or
>> 135,000 matchups by hand.
>
>Mitch's poll fundamentally uses a sound algorithm. I'm curious what the GTA
>sample would generate using Mitch's tabulation mechanism. The issue Mitch's
>is prone, too, though, is the monoculture visiting an out of the way park
>and generating a unanimous positive or negative ranking. That tends to
>bypass the primary strength of the mutual-comparison system, which assumes a
>random population of visitors. If that population isn't actually random, a
>lot of the validity breaks down.
True. but I can't honestly think of any way to avoid this. After all,
there aren't a lot of Koreans reading this newsgroup, for example. Or even
all of the internet coasters forums combined. And certainly not a
statistically significant sample.
Therefore, group tours of faraway countries will still end up providing
most of the people who have ridden the coasters in those countries.
And that can end up skewing the results.
For instance, it will be interesting to see how Katun scores next year. The
ECC -- which seems to have more B&M looper fans than some groups and forums
-- has done two trips to Italy that I know of (one of which I've been on.)
And let's face it, anyone going on a coaster trip JUST to Italy and the
surrounding areas is probably a big B&M looper fan, since that's really the
only notable coaster in the country! That's probably why Katun scores more
highly that other great B&M's particularly Montu, which has had a LOT more
riders in the polls.
It will be interesting to see what happens next year, when TPR goes to
Mirabilandia, as part of a large trip all over Italy. Not only do I see
much less of a bias towards B&M loopers among TPR members, but Mirabilandia
is not the most notable park on the trip, since it's going a great distance
to many great parks. Thus, I think that fewer of the people on that trip
will be as pre-disposed to love Katun as those who've already ridden it
previously in past polls.
But I don't really see any way to account for that in any poll.
Unless, of course, someone wants to create a fund to send lots of coaster
enthusiasts around the world! ;-)
>For instance, for the steel poll, it would be trivially easy to make John
>Ivers' Blue Flash the #1 steel coaster.
Not now that it's off the ballot! ;-)
>This is why we have a minimum sample size requirement, and why it keeps
>getting adjusted upward. It would be interesting to plot ranking volatility
>as a function of voters per ride.
True.
I think that ultimately that until a coaster has at least 50 and preferably
100 riders, you're likely to see a lot of variance in its rankings.
What's ultimately most interesting is to look at the 16 year results table
of the woodie poll. There's tremendous volatility in the less-ridden
coasters.
But when you look at most of the ones with the most riders, preferably over
100, and certainly over 200, you'll find remarkable consistency in the
rankings, with the inevitable declines as more new coatsers are built and
as older coasters need maintenance and retracking. This is particularly
true if you just look at the last 10 years or so, as much larger numbers of
people have taken the poll.
Pick just about any coaster with over 200 riders, and you'll see a very
strong consistency, when you allow for the natural slide as new coasters
are added.
And you certainly won't see the level of inconsistency of a coaster with 10
riders like La Fiera in Mexico, which in the past 5 years has jumped from
31 to 106 to 89 to 50 to 39! Compare that to say Colossus at SFMM, which
has ranked between 140 and 145 for the past five years. Or Racer at
Kennywood, which has ranked between 73 and 79 for the past 5 years. And
the same holds true for most coasters with over 200 riders.
What that tells me is that the algorithm is sound. It just needs more
riders and more data points.
Steelforce
<davidhhhSTOPS...@bellatlantic.net> wrote:
> PLEASE remove "STOPSPAM" from my address when replying via e-mail.
>
> "Patriotism means to stand by the country. It does not mean to stand by
> �the president or any other public official, save exactly to the degree
> �in which he himself stands by the country. It is patriotic to support
> �him insofar as he efficiently serves the country. It is unpatriotic not
> �to oppose him to the exact extent that by inefficiency or otherwise he
> �fails in his duty to stand by the country. In either event, it is
> �unpatriotic not to tell the truth, whether about the president or
> �anyone else."
> � � � � � � � � � � � � � � �-- Theodore Roosevelt
Dave, you really need to take more time to explain your opinions! ;^)
BaSSiStiSt
> Dave, you really need to take more time to explain your opinions! ;^)
The phrase "Cheeseburgian rant" was flying through my mind for some
reason too!
:-)
As the kids say these days, "tl,dr".
David Sandborg
"David H.--REMOVE \"STOPSPAM\" to reply"
> I said that if you don't like his algorithm, then CREATE YOUR OWN DMAN
> POLL.
Come on. Jeremy's quite within his rights to raise questions about
Mitch's polls, and I don't see why he has to create a new poll of his
own in order to do so. He's raising discussion points, which is quite a
legitimate way to criticize.
You continually raise the issue that the problems are theoretical, not
actual; I don't think you understand the value of theoretical
argumentation. The data set that Mitch is working from is large and
messy, and interpretation of it can be difficult (as the sub-argument
about whether Mitch's poll is fundamentally reliable or just more
reliable than others demonstrates). A theoretical setup can provide a
crystal-clear set of (hypothetical) data, the interpretation of which is
not in question. Then you can see where the results lead. Anyway,
today's theoretical problems could become tomorrow's actual ones. That
a scenario hasn't happened yet doesn't mean it couldn't ever happen.
In general, I just find the tone of your response overblown. Jeremy's
not making a personal attack of any sort on Mitch, yet you're responding
as if it were. I'd suggest dialing it back a bit.
Jeremy
your perspective better but there's still some points of mine I don't
think you're interpreting correctly so I'll try better to be a bit
clearer.
First of all *please* don't use the whole "Mitch worked very hard on
it and you've not put in anywhere near as much effort as he did so
therefore you shouldn't be criticizing his work." That's true of the
vast majority of things anyone will ever criticize, and if there was a
universal caveat that one first must work as hard at the problem as
the original creator before they are allowed to make criticisms of
their work for the sake of politeness, then humanity would probably
still be in the stone-age, that's how slow it would impede our ability
to improve upon things. If I were making personal judgments against
Mitch then you might be right in calling me out on that, however I'm
pretty sure I've done nothing but objectively describe what I perceive
as inaccuracies with the polling method, and (correct me if I'm wrong
here) any honest attempts to criticize anyone's work for the sake of
trying to shed light on previously unrealized truths (no matter what
their status is or if it ends up that they're completely wrong) are
never disrespectful to anyone. I honestly love reading your replies to
find ways to strengthen or concede points in my argument, but when you
write things like "Until you do so, stop complaining about someone who
HAS TAKEN THE TIME TO DO SO", that really leaves a bad taste in my
mouth and doesn't win your position any points either. Besides I've
already spent several hours writing out arguments on a topic that
doesn't even affect me that much explaining why I think Mitch should
change one line of code to help make his poll more accurate which he
could probably do in five minutes since he already has the programs
and everything written while I don't. If that's still not being fair
to Mitch's work then I don't know what is.
While we're on this subject, you don't need to bring my personal
motivations for criticizing this poll into this discussion. Yes, I
could very well be a selfish, egomaniacal, f'ed-up human being that
probably isn't worthy of the air I breathe. I perhaps do have a
personal vendetta to watch Mitch's poll fail horribly and then be
given absolute dictatorial power over every roller coaster opinion
poll to ever be released in the future. Regardless of what I have done
or what my motivations may or may not be, this has absolutely no
relevance to the factual argument at hand, which is the accuracy of
Mitch's poll. And yes, I am in the minority on a lot of the rankings
and would like to see the results be improved to better match my
standards. Did you expect that someone who for the most part agrees
with the poll would be the one to raise an objection? You'll be able
to easily shoot down a lot of potential arguments that way because
it's generally people in the minority that raise objections to the
mainstream status quo (and I'm not even that much in the minority, I'm
just frustrated that many mediocre B&M and Intamin designs are able to
far outrank great designs by other, lesser appreciated manufacturers).
If you look beneath the surface of nearly any argument you'll be able
to find some sort personal motivation that initially spurred it on,
that doesn't affect the potential validity of that argument at all.
Just because I might happen to think that MY ballot isn't counted
fairly, or the fact that I don't personally agree with some of the
results doesn't actually influence the argument I use to show that
there are potentially inaccuracies with the poll, it just makes it
clear why I happened to be arguing from that perspective in the first
place. [FYI none of the above is part of my actual argument on the
issues nor does it affect my reading of what your main points are, I'm
just getting this stuff out of the way this discussion can be more
streamlined to focus on the key issues at hand without these
extraneous side-arguments.]
Now that I've said all of the above, please remember I've mentioned
that right now I don't know or care if any of my suggestions for
improved methods actually will affect the results in ways that I would
agree with them more or would give me more personal influence, I just
want to argue for more accurate results in general regardless of if I
even end up disagreeing with those results even more than the current
method. Accuracy is accuracy, that's all I want to see improved. Sure,
I do think that my suggestions might end up reflecting my opinions
better, but not only is that simply because I do think that some of my
opinions on where certain rides are over or under performing in the
poll does represent actual inaccuracies based on the votes that were
cast, but my opinion on these coasters is not really relevant to the
actual topic of accuracy, so hopefully I can stop mentioning that from
now on.
Okay, so about that actual discussion...
"No. Sorry. Absolutely not. NO. In theory, I COULD be hit by lighting
today..." (ps sorry I didn't set this reply up for proper quoting,
I'll try to do that next time). That's a good try at an analogy but
it's actually pretty off mark, at least if it's left just as it is.
For it to have worked you'd have to use an "in theory..." example that
has the same likelihood of occurring as my "in theory...", otherwise
it doesn't work. You would know that in theory your chances of getting
struck by lightning are less than winning the lotto (or so they say),
which is why you ignore that theory and go outside anyway. My theory
does not have the same odds of making a difference so the analogy
doesn't hold up. What if I were to say, "in theory, if you were to
wave a metal rod around at the top of the tallest hill in the county
in the middle of electric storm, your chances of getting hit are about
50/50?" Does that mean that you're still going to ignore the theory
and bring your golf clubs to do some practice driving at the top of
the hill next time the weatherman predicts thunderstorms? Hopefully
the theory is enough to convince you that that's not a very good idea
and you don't need to experiment with it yourself before making up
your own mind about the theory's relevance (actually that's not the
best counter-example either now that I think about it because you
could cite other cases where people have been struck by lightning in
that circumstance as the reason for not doing it; imagine an entirely
theoretical situation where the hypothesis claims that the likelihood
of it actually happening in practice is somewhat high, at least
relative to your original lightning example, and you'll have a better
analogy there). I've already mentioned that the poll, in theory, is up
to 49% inaccurate, that's a pretty high percentage of potential
inaccuracy, which even if the actual inaccuracy of the results is much
lower than that (which, as you pointed out and I was already well
aware of, it *will* be), still by itself as just a theory warrants a
lot more than an immediate dismissal if no hard evidence of it
occurring in practice has been provided yet. Plus as Dave mentioned,
using actual results can be way too messy with too many other
variables contaminating possible interpretations, while using a
hypothetical example can clearly illustrate where in the method
something is going wrong to cause results from a data set that are
less than ideal.
I think part of the problem might also be that we're arguing from two
different perspectives. You're only interested in this problem to the
extent that it will make a meaningful difference in real world
applications (which you are skeptical that they will; I'll get to that
in a moment), while for me, at least at this point in the argument,
I'm interested in demonstrating that these problems are theoretically
possible before moving on to judging how big of an impact they have on
the actual results. Furthermore, inaccuracy in theory is always the
most important thing to consider when determining how inaccurate
something is, not how accurate it turns out in practice. Any poll, no
matter what the methods, even just a randomly generated list, could
produce results that are more "accurate" to what the 'ideal' results
could be (you suggested mind-reading, so we'll use that as the
benchmark of the 'perfect' results). That doesn't mean that we should
respect the results of that poll at all since the true measuring
device of how good a poll is is by analyzing how well the mechanics of
it churn through and interpret the data. I think that's why most
people (including yourself) respect Mitch's poll, because the method
of how the results are in theory determined is better than any other
poll, not because we like the actual results the best (although maybe
that too, I'm not too sure about myself but that's beside the point).
My argument therefore is that, in theory, the mechanizations of the
poll are much more flawed than nearly anyone else will give it credit
for, and that's what ultimately really matters. (And to reiterate this
point once again because I still wasn't sure if you got it or not, I
do think that Mitch's poll is the most accurate out there relative to
other polls, it's the absolute level of accuracy that concerns me).
You also use the fact that the poll is consistent in ranking some
rides as an indication of it's accuracy. Throw that argument right out
the window, consistency in no way predicts accuracy, it just proves
that however bad or good the methods you use, they produce the same
result every time. The reverse is true, an inconsistent poll probably
does predict poor accuracy somewhere, but never the other way around.
Now, you do seem to agree that my suggested change in the method could
make it slightly more accurate, but you claim that the effect it would
have is pretty insignificant that it doesn't really matter enough to
write half an essay about. Maybe that's true, I suppose it is possible
that the results could be less than two percent and there would be
hardly any major shake-ups in the poll. But the problem is we don't
know. Going through the actual ballots as you did, I might believe
that there's not going to be as big of a change, but part of the
problem is the final results don't tell you at all whether that level
of accuracy is 51% or 99%. 49% of the head-to-head comparison data
goes discarded, but we do know that less than that amount (usually
much less) is made up of minority votes. But statistically speaking,
we can't label the data that was discarded but still happened to agree
with the majority that was recorded as a 100% as still being true and
represented in the poll. Data that confirms the results but was
discarded is just as big of an oversight as data that contradicts it
and was discarded, the only difference being one could affect the
turnout of the final results while the other doesn't. You're probably
shaking your head as you read this since this is still all
hypothetical that even I'll admit probably doesn't affect the results
as much as I may have anticipated when I first started this debate...
HOWEVER, if we're to accept this hypothetical at all, it still proves
that Mitch's poll is on a technical level only 51% accurate, with the
remaining 49% accuracy it might have (assuming the results actually
are 100% accurate to the 'ideal' results) is just being guesstimated
on the assumption that they're probably close enough to what the
actual results would be. Therefore it would still be technically
correct to say that the *methods* only produce an accuracy of 51%,
even if the results themselves are more accurate than that, we can't
claim that they are because they were never part of the data set that
produced the final results. Again, that might not bother you but it
bothers me, and I do hope I'm not the only one.
By the way, one thing that bothers me, you say that because Mitch's
poll is the most accurate out of any other poll out there so far that
makes it the de facto for coaster rankings. Probably our definitions
differ a bit, but I wouldn't call a poll as being 'de facto' just
because it happens to be the most accurate. If every poll were known
to have huge inaccuracies in them then I doubt anyone would want to
call any of those de facto. I guess to me, declaring a coaster as
being the de facto #1 in the world describes an absolute sense of
accuracy rather than simply being relative to all other polls. i.e. we
would say, "to the best of our knowledge this ride is maybe #1", but
not cite it as #1 with any other opinions being dismissed as factually
wrong (and PLEASE realize I'm talking just about polls as factually
representing the data to determine #1 and nothing more fundamental or
abstract!)
Next, about my laziness to not produce any concrete examples of it or
to start my own poll... first of all, as you mention there are
thousands of data points, and in order to find even on individual
ranking not only means that I'd have to read every ballot but also
chart each head-to-head comparison, and I don't have Mitch's
algorithms or programs or anything, I'd either have to completely
duplicate all the work Mitch has put forth to the poll and then add
the extra effort of adding my own parts to it, or do it by hand which
would probably take me the better part of the year to complete. No
thank you. That is simply an impossible order to request from me given
the resources I have to work with. I'm sorry if that would be a key
piece of evidence for my argument that I can't produce, but not only
can I not produce it, but I still don't think it's nearly as important
to the debate as you make it out. Having those concrete examples of
how it would actually affect the poll is great to have to back up the
theory, but not only do they not technically affect the theory, but in
this case, since we're just dealing with number and statistical
analysis and not a science that requires actual empirical evidence to
be collected, it's not even at all necessary to prove the point I'm
after.
Furthermore, as I've said, most of my issue with the poll stems from
what I imagine would be a singular line of code, which is where it
records the results of each head-to-head comparison as either a 1 or 0
instead of the actual percentage. That's not even changing any of the
methodology of the poll, just the way data is recorded in an empty
box. You say that's negligible to the final results, I say it's a
simple oversight that actually does produce fundamental inaccuracies
in the poll and the potential for much different results (it might not
make as big of a difference with generally low ranked coasters
compared to generally high ones, but those few spots where it
generally is tight between two rides are generally the most important
in determining the final rankings, and this goes all the way up to the
top ten). I've gone about as far I want to or even think I need to
with the theory of why this is a problem, now the only thing left to
do is to actually test it out as you've been suggesting and see if the
results vary greatly or not as I think they might. As I've said, I'm
not going to perform that test because the poll is set up in such a
way that you have to work through each and every data point to even
get through one ride, meaning I'd have to go through the same work
process to duplicate the same algorithms and computer programs that
Mitch already has, just so I can adjust the one line. Or Mitch, who
already has the program and obviously worked very hard to make it from
scratch, can take five minutes to find that single spot and test it
out himself if he's interested.
There's a couple other things I'll just repeat quickly because I'm
worn out and don't feel like discussing this much more, but again you
seem to be completely missing my point about minority opinions. In
order to be considered accurate the results need to be triangulated by
every opinion out there rather than just assumed when it reaches a
simple majority from one viewpoint, which Mitch's poll doesn't do at
all because of the same problem with recording each head-to-head
comparison as either a 1 or 0 and not a percentage. If there are
equally size minorities that rank a mid-ranking ride both equally high
and equally low, then the majority opinion will remain unaffected, but
otherwise the actual rankings are always going to be off. As it is
right now, you're basically saying that person A's vote will be given
it's full share of weight in the poll while person B's ballot will be
discarded because person A happens to agree with the majority while
person B doesn't, therefore making person B's opinion inferior to
person A's just by that criteria alone. Also, about the different
weighting of different ballots, I think based on your reply you didn't
get what I'm saying at all. If two people, one who has been on 50
coasters and the other on 500, both get to vote for their top ten as
in the GTAs, that will be 'fair' but not equitable, for the obvious
reason you mentioned. If those same two people get to vote in a way
that they affect each an every coaster they've personally been on
equally, then that would not be fair, because the second person would
get 10x the opportunities to influence different rides, but it would
be completely equitable. Generally speaking, that is what we are
striving for, because it produces the best results. However, what I am
claiming happens in Mitch's poll, because for each new coaster a
person has been on that creates another head-to-head comparison for
each of the coasters they've already been on. And because the poll is
determined by the number of head-to-head wins, that not only gives the
second person the 10x greater influence they would naturally have
simply by being on more coasters, but it would also give them 10x more
weight for each coaster they've already been on as measured by each
head-to-head comparison they affect for a certain ride, giving their
ballot 10x10=100 times more weight than the first person. This is
neither fair nor equitable. With the binary head-to-head comparison
system that favors less-ridden, obscure rides with a greater
probability that they can flip the results from the win to lose side
of the fence or vice versa (affecting not just the obscure rides
themselves but also the popular rides by changing more head-to-head
outcomes; i,e. you rank MF last and that happens to break the ties
with several obscure rides only a few other people have been on,
giving MF several more losses and therefore affecting its final
ranking much more than if you had not been on those obscure rides in
the first place), that actually means that Mr. 500 credit count will
be even more advantaged than that because it's more likely that a lot
of those 500 rides are the obscure ones that he can influence, thereby
also transferring greater influence on the popular rides, and in the
end the 10x difference between the two voter's coaster counts can add
up to probably over 500x, maybe even 1000x the weight given to the
more experienced coaster rider, assuming they get to influence the win/
loss outcome of a lot of obscure rides while the first person can't
influence anything because they've only been on popular coasters.
Fixing the one line as I've mentioned would be sure to keep it such
that the increase in weight each vote is given is at a fixed 100x for
each 10x the number of coasters each voter has been on. Just to
clarify, the reason this is a problem isn't because I want a greater
say in the poll, it's a problem because it threatens the accuracy of
the poll once different votes are allowed different weight in the
final results based on circumstantial factors that has nothing to do
with the actual quality of the rides, besides, equality of opinion is
something that's generally assumed in polls on at least an individual
ride, whether they've also been on 1000 or 10 other coasters beside
it, but it's not all that present here using Mitch's methods. I'm
still not 100% about all of this, will need to look at it more
closely, so my stance on this issue may change.
F**k it, I need to stop, I just moved from the US to Rome not even two
weeks ago and I don't even want to count how many hours I've devoted
to this subject when I should have been out exploring the culture
instead (not that that's anyone's fault but my own). For the record,
after reading your last post I have stepped down a bit on my position
that the actual results themselves are very flawed, although if it's
not already clear I'm continuing to stand strong to the belief that
the methods that achieve those results are much farther from ideal
than nearly anyone else will acknowledge, and I definitely don't want
to leave these ideas undefended since I don't think anyone else will
take my spot if I decide to give up. Probably what I should just do
now however is send Mitch an email outlining the main points/concerns
I still have with the poll that we've worked out now through this
debate and see if he thinks they're worth looking into further for
next year's poll.
Dana Schwartz
<jkthom...@gmail.com> wrote:
>F**k it, I need to stop, I just moved from the US to Rome not even two
>weeks ago and I don't even want to count how many hours I've devoted
>to this subject when I should have been out exploring the culture
>instead (not that that's anyone's fault but my own).
Well, the good news (I think) is that you're not an insomniac if you
posted this "Rome time". I wondered how anyone could type that much
at this time of day anywhere in the US (your time stamp was 5:45 AM
EST so anywhere west of that was in the wee hours here).
Then again, that was quite a few words most any time of the day, in
any time zone! :-)
Go outside and see Rome, Jeremy!
Dana
David Sandborg
<9b9f357a-e280-4bee...@g1g2000yqi.googlegroups.com>,
Jeremy <jkthom...@gmail.com> wrote:
> Next, about my laziness to not produce any concrete examples of it or
> to start my own poll... first of all, as you mention there are
> thousands of data points, and in order to find even on individual
> ranking not only means that I'd have to read every ballot but also
> chart each head-to-head comparison, and I don't have Mitch's
> algorithms or programs or anything, I'd either have to completely
> duplicate all the work Mitch has put forth to the poll and then add
> the extra effort of adding my own parts to it, or do it by hand which
> would probably take me the better part of the year to complete. No
> thank you.
This led to an interesting thought. I wonder if Mitch would be willing
to make his code open source? To some degree his poll already is almsot
there, since all the data and the algorithm itself are already competely
public. However anybody wishing to implement improvements still would
have to put a great deal of work to even get to the point that Mitch has
even with all the public info. At the very least the code that reads
and parses the ballots into some kind of array structure, on which
anybody (or at least any programmer) could then try their own processing
method. It would be interesting. (I should mention that once for work
as a research project I did attempt to substantially duplicate Mitch's
algorithm myself. However it was working on a different data set that
already came preprocessed to some degree. Parsing Mitch's data set
would have involved quite a bit more work, quite possibly more than
applying the algorithm itself.)
Nashville Mike
That's one serious shitload of text. I thought I was wordy back in the
day.
My view is pretty blunt: You come off like a guy who can't handle that
your favorite coaster doesn't rank highly with the consensus of riders
and thus feel compelled to prove how the other guys opinion is wrong.
I could be wrong of course, but I tend to doubt it. I've seen this
*exact* behavior in other forums in other hobbies all the time. The
thing is - who gives a flying fuck if the ride you like is hated by
most or vice versa? You ride them, we hope, for your own personal
enjoyment. I know when I ride the Voyage, my personal favorite, that I
don't really care if TPR, CB, ACE, or whomever ranks it as #1, or
#100, or #1000, because at this point I know my tastes, I've ridden a
lot of wooden coasters, and I don't need a poll to direct me to the
next woodie I might like. Mitches poll isn't perfect, but it's
reasonably accurate, as best as a poll can be. Sure - a small
contingent of riders can skew results to get their favorite in there,
but really, when you think about it - so the fuck what? I mean,
seriously, if you look at any of the top coasters, I think a
reasonable argument could be made that the majority of coaster
enthusiasts would enjoy the ride - so whether Voyage, Grizzly, Far-Off-
Korean-Coaster-Of-The-Month or whatever is #1 and one of the others is
#2, it's irrelevant unless you just feel the need to participate in
mental masturbation knowing "your" coaster is #1 in the poll. These
days I'm more curious how some "not #1 but still good" coaster (like
Prowler in KC, one hell of a ride) ranks with the folks who have
ridden it. I'm long over losing sleep if my personal fave is ranked #1
or not, cause these days, I could care less.
The poll is fun, and yes, I vote, but it's usefulness has diminished.
Long ago before the growth of the internet and the dissemination of
information that came with it, polls were a nice way to determine
where, perhaps, we might travel to. Now there is so much out there
it's irrelevant. Making 400 paragraph arguments to try and disprove
the poll or investigate it's "accuracy" seems like a wasted exercise
to me. Go out and ride more and type less dude!
-m
Jeremy
mile around Rome or visiting Florence for a weekend sleeping in a room
where the temperature is less than 50 degrees F, especially since it
helps put my mind on other things I'm more familiar with and
momentarily distract me from the culture shock (not to mention I was
in bed sick from after said weekend visit to a frigid Firenze which is
where I found time to write out all of the above...)
Anyway I'll keep this reply quick. If I feel like writing a lot on a
subject then that's my business. If you've got something on your mind
want to try to articulate as clearly as you know how then I wouldn't
begrudge you that just because it seems excessive given the importance
of the topic to me. Besides, how am I supposed to read your thoughts
on which topics you believe are and aren't worth discussing?
I'll assume you didn't read all of that text before jumping in to
criticize me since one of the points I've repeatedly stressed is that
my motivation is neither to see my favorite rides show up in the #1
spots (which if they were to show up in those positions one year I'd
be one of the first to cry foul play since I know that wouldn't not
reflected by the actual ballots submitted), nor that those things are
particularly relevant to the argument I actually am trying to put
forth, which is potentially major inaccuracies with Mitch's poll that
haven't generally gone acknowledged. I know full well what my opinions
on certain rides are and that they frequently disagree with majority
opinions, that's why I started a personal website devoted to that
subject. The reason this topic is important to me is because I
constantly encounter people that reference Mitch's poll as being as
accurate as a poll could be (frequently used in conversations
discussing which coasters are generally the most popular amongst
people which is an important conversation to have if we're to
determine what elements within a coaster generally have the strongest
affect on people on the average) and I personally feel that I cannot
be sure if the coaster that ranked #1 or whatever is in that spot
because that's what the voters actually feel or if there's some random
ghost data interfering with the actual results. I'm fine with a
coaster I don't agree with coming in #1 as long as I can be sure that
it's the closest to a consensus as the data could possible reach and
large sections of votes aren't being included or given randomly
unequal weight, which is exactly what I'm *not* sure of.
"Mitches poll isn't perfect, but it's reasonably accurate, as best as
a poll can be. Sure - a small contingent of riders can skew results to
get their favorite in there..."
You seem to believe that these are self-evident truths apparent to
everyone, when that's exactly what I'm arguing against (at least that
the poll is as accurate as it possibly can be, which I've been saying
some minor changes could make the poll much more accurate). You give
an example of one thing that is wrong with the poll apparently on the
presupposition that that's the only thing wrong with the poll and it's
not a big deal so I shouldn't be complaining, when I've been saying
nothing but there are many other things wrong with the poll that
aren't as obvious at first glance because they don't affect the data
in as dramatic ways that grab out attention like a random Asian
coaster ending up in the top ten, but these other issues could
possibly have a much larger impact on the overall final results than
that one problem (although I'm really not sure what changes in the
results the theory would produce in reality, that's the main thing
I've been having to reconsider since starting this discussion).
I agree with your last point about the practical usefulness and
importance of the poll over time, but it's still a pretty major annual
event in enthusiast circles and I think is worthy of at least the
amount of time I've put into it so far, although I definitely am
getting close to reaching my limit of interest on the subject.
David H.--REMOVE "STOPSPAM" to reply
wrote:
>On Jan 24, 2:53?am, "David H.--REMOVE \"STOPSPAM\" to reply"
><davidhhhSTOPS...@bellatlantic.net> wrote:
>> On Wed, 20 Jan 2010 22:31:35 -0500, "Wolf" <bill.buss...@gmail.com> wrote:
>> >> Or how about Magnum? ?#9 on the GTA's and #69 on Mitch's poll. ?Sure, I
>> >> love it. ?And I know that a lot of the old-school enthusiast types who
>> >> tend
>> >> to be on the AT survey love it. ?But it's certainly no longer considered a
>> >> top 10 ride. ?Or even a top 4 ride AT THE SAME PARK!
>>
>> >This is probably more of a generation gap in the voters than an indication
>> >of the sample weighting, which was your point.
>>
>> Actually, I think it's more an effect of the fact that Magnum has been
>> eclipsed by many more similar rides in the years since it's been built.
>> (And also many dissimilar coasters, which would still knock it down in the
>> rankings.)
>>
>> After all, if you're a big fan of Hypercoasters, there are hell of a lot
>> more of them to choose from than there used to be.
>>
>> >And what rides do you put ahead of Magnum at CP? I could see MF and an
>> >argument for Raptor, but what else? And I still have Magnum in my top-10
>> >overall, and I've ridden a sizeable chunk of both the wood and steel
>> >top-20s.
>>
>> I was saying that in the general opinions, it's fallen. ?Even at the Point.
>> Dragster, MF, Maverick and Raptor all are pretty clearly more popular among
>> both enthusiasts and the GP.
>>
>> Just look at the lines at the park during coaster events. ?Magnum late
>> night rides used to be legendary, leading to HUGE lines at events. ?Now,
>> not so much.
>>
>> Personally, I'm with you. ?I haven't ridden Maverick yet (probably this
>> year, before HWN), but I agree that the only coaster I've ridden at the
>> park I'd put above Magnum is Raptor.
>>
>> But as much as I love both coasters, which were certainly once in my top
>> 10, they've fallen as I've ridden more and more coasters, particularly so
>> many newer ones.
>>
>> >> How can you say that it's almost as accurate when the top ten steel
>> >> coaster
>> >> lists don't even RESEMBLE each other?!?
>>
>> >Here you're just presupposing that Mitch's poll is accurate.
>>
>> >> And you seem overly concerned about how mediocre coasters will rank.
>> >> Honestly, who cares? ?Wouldn't you much rather see a poll that has a
>> >> pretty
>> >> accurate top 10 and top 40 listings, rather than a slightly more accurate
>> >> middle 50?
>>
>> >Why would it be mutually exclusive? Accurate is accurate.
>>
>> Well, for one thing, very few people bother to rank the coasters outside
>> their top 10 or 20 or so (or at least their top 50 or so, especially on the
>> steel poll), simply grouping the rest to save time. ?As a result, you can't
>> get a truly accurate result on how coasters that people can't be bothered
>> to rank accurately should be ranked. ?There isn't much meaningful data
>> there.
>>
>> And that's fair. ?Not only do I not have the time or energy to figure out
>> whether I think that Dahlonega Mine Train is better than Gold Rusher, I
>> don't care.
>>
>> And that's the same reason I don't care about the results past 50 or so. ?
>>
>> And let's be honest. ?Just about no one else does either.
>>
>> That was my point.
>>
>> This actually makes me wonder if the poll wouldn't be more accurate by NOT
>> listing all of the low-ranking -- or maybe even medium-ranking coasters. It
>> would also avoid the issue of less-traveled coasters having a lower
>> percentage because they don't have enough valid comparisons with low-ranked
>> coasters.
>>
>> And it would probably bring a lot more people to fill out the poll,
>> especially the steel one. ?Let's face it, the poll ballot is pretty
>> daunting for both die-hard enthusiasts and newbies alike. ?The only reason
>> most of us can be bothered to rank more than 20-30 coasters is because we
>> can cut and paste last year's ballot!
>>
>> If it brings people like Richard to take the poll, then that would be
>> particularly valuable to the date set.
>>
>> Just something to think about.
>>
>> >> If you think that your method would yield significantly better results,
>> >> why
>> >> not do the calculations yourself? ?Or write a program that will do it for
>> >> you, if you'd rather not deal with nearly 1.5 million data points or
>> >> 135,000 matchups by hand.
>>
>> >Mitch's poll fundamentally uses a sound algorithm. I'm curious what the GTA
>> >sample would generate using Mitch's tabulation mechanism. The issue Mitch's
>> >is prone, too, though, is the monoculture visiting an out of the way park
>> >and generating a unanimous positive or negative ranking. That tends to
>> >bypass the primary strength of the mutual-comparison system, which assumes a
>> >random population of visitors. If that population isn't actually random, a
>> >lot of the validity breaks down.
>>
>> True. ?but I can't honestly think of any way to avoid this. ?After all,
>> there aren't a lot of Koreans reading this newsgroup, for example. Or even
>> all of the internet coasters forums combined. ?And certainly not a
>> statistically significant sample.
>>
>> Therefore, group tours of faraway countries will still end up providing
>> most of the people who have ridden the coasters in those countries.
>>
>> And that can end up skewing the results. ?
>>
>> For instance, it will be interesting to see how Katun scores next year. The
>> ECC -- which seems to have more B&M looper fans than some groups and forums
>> -- has done two trips to Italy that I know of (one of which I've been on.)
>> And let's face it, anyone going on a coaster trip JUST to Italy and the
>> surrounding areas is probably a big B&M looper fan, since that's really the
>> only notable coaster in the country! ?That's probably why Katun scores more
>> highly that other great B&M's particularly Montu, which has had a LOT more
>> riders in the polls.
>>
>> It will be interesting to see what happens next year, when TPR goes to
>> Mirabilandia, as part of a large trip all over Italy. ?Not only do I see
>> much less of a bias towards B&M loopers among TPR members, but Mirabilandia
>> is not the most notable park on the trip, since it's going a great distance
>> to many great parks. ?Thus, I think that fewer of the people on that trip
>> will be as pre-disposed to love Katun as those who've already ridden it
>> previously in past polls.
>>
>> But I don't really see any way to account for that in any poll.
>>
>> Unless, of course, someone wants to create a fund to send lots of coaster
>> enthusiasts around the world! ?;-)
>>
>> >For instance, for the steel poll, it would be trivially easy to make John
>> >Ivers' Blue Flash the #1 steel coaster.
>>
>> Not now that it's off the ballot! ?;-)
>>
>> >This is why we have a minimum sample size requirement, and why it keeps
>> >getting adjusted upward. It would be interesting to plot ranking volatility
>> >as a function of voters per ride.
>>
>> True.
>>
>> I think that ultimately that until a coaster has at least 50 and preferably
>> 100 riders, you're likely to see a lot of variance in its rankings.
>>
>> What's ultimately most interesting is to look at the 16 year results table
>> of the woodie poll. ?There's tremendous volatility in the less-ridden
>> coasters.
>>
>> But when you look at most of the ones with the most riders, preferably over
>> 100, and certainly over 200, you'll find remarkable consistency in the
>> rankings, with the inevitable declines as more new coatsers are built and
>> as older coasters need maintenance and retracking. ?This is particularly
>> true if you just look at the last 10 years or so, as much larger numbers of
>> people have taken the poll.
>>
>> Pick just about any coaster with over 200 riders, and you'll see a very
>> strong consistency, when you allow for the natural slide as new coasters
>> are added.
>>
>> And you certainly won't see the level of inconsistency of a coaster with 10
>> riders like La Fiera in Mexico, which in the past 5 years has jumped from
>> 31 to 106 to 89 to 50 to 39! ?Compare that to say Colossus at SFMM, which
>> has ranked between 140 and 145 for the past five years. ?Or Racer at
>> Kennywood, which has ranked between 73 and 79 for the past 5 years. ?And
>> the same holds true for most coasters with over 200 riders.
>>
>> What that tells me is that the algorithm is sound. ?It just needs more
>> riders and more data points.
>>
>> David H, ?david...@STOPSPAMbellatlantic.net, ?Boston, MA
>> PLEASE remove "STOPSPAM" from my address when replying via e-mail.
>>
>> "Patriotism means to stand by the country. It does not mean to stand by
>> ?the president or any other public official, save exactly to the degree
>> ?in which he himself stands by the country. It is patriotic to support
>> ?him insofar as he efficiently serves the country. It is unpatriotic not
>> ?to oppose him to the exact extent that by inefficiency or otherwise he
>> ?fails in his duty to stand by the country. In either event, it is
>> ?unpatriotic not to tell the truth, whether about the president or
>> ?anyone else."
>> ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?-- Theodore Roosevelt
>
>Dave, you really need to take more time to explain your opinions! ;^)
I would, but I'm shy.
David H, davi...@STOPSPAMbellatlantic.net, Boston, MA
Wolf
"David Sandborg" <sand...@Spam-away.ix.netcom.com> wrote in message
news:sanddave-03872D...@earthlink.us.supernews.com...
> This led to an interesting thought. I wonder if Mitch would be willing
> to make his code open source? To some degree his poll already is almsot
> there, since all the data and the algorithm itself are already competely
> public.
I think Mitch is reticent to do so. I asked about it years ago.
I seem to recall it was in Perl, which makes sense. Considering it's just a
pairwise comparison, it seems easy enough to implement the algorithm itself.