Theory of aces: high score by skill or luck?
John Bailey
Theory of aces: high score by skill or luck?
Authors: M.V. Simkin, V.P. Roychowdhury
Abstract: We studied the distribution of WWI fighter pilots by the
number of victories they were credited with along with casualty
reports. Using the maximum entropy method we obtained the underlying
distribution of pilots by their skill. We find that the variance of
this skill distribution is not very large, and that the top aces
achieved their victory scores mostly by luck. For example, the ace of
aces, Manfred von Richthofen, most likely had a skill in the top 29%
of the active WWI German fighter pilots, and was no more special than
that. When combined with our recent study (cond-mat/0310049), showing
that fame grows exponentially with victory scores, these results
(derived from real data) show that both outstanding achievement
records and resulting fame are mostly due to chance.
Gordon
airmen he faced. In ~100 combats, only one aviator brought him down.
v/r
Gordon
tomcervo
Gordon wrote:
> The theory is flawed, as Richtofen was obviously better than 98% of the
> airmen he faced. In ~100 combats, only one aviator brought him down.
Actually 2, or 3 if you count the pilot of the Fe 2--a massively
inferior aircraft, BTW.
WaltBJ
in physics and coming up with a similar result. I spent 22 years trying
to be the best fighter pilot I could and poring over all rpt all the
pertinenent literature. Ability, attitude/aggressiveness, physical
fitness, eyesight, 'situational awareness' meaning keep your head out
of your . . .; never stop learning/practicing/ what you don't know can
kill you . . . know your airplane's limits and be able and eager to fly
it to them; and know when to fold 'em. Richtofen was a superb shot as
was Udet. These yoyo's can't explain the Mannock, Bishop, Guynemer, etc
- luck? Yeah, right. 'Luck' is what theese pilots made themselves.
What really brought Richtofen down? Tunnel vision plus mental fatigue
from too many missions - he forgot everything he'd learned about aerial
warfare in his pursuit of that next kill. I'll bet these yoyos churned
numbers and never even talked to one fighter pilot about what it took
to be good. And then they had the monumental gall to call it scientific
and even signed their names to the trash. They need a couple back seat
rides with some good sticks to see what it's all about. Not that they
personally would see much . . .
Walt BJ
WaltBJ
assumed skill doesn't vary much between fighter pilots. Gentlemen, I
assure you from experience gained while serving in nine count 'em nine
fighter squadrons over 22 years and 5000 hours of flying time in 5
different fighters, much of that time instructing and administering
tactical evaluation and instrument checks to fighter pilots, yes indeed
quality differs widely. As much as say a high school physics teacher
with his PhD differers from Richard Feynman or Julian Schwinger. FWIW
most fighter squadrons have about 3 top notch fighter pilots. OTH there
will be 3 or so guys who shouldn't be there, usually because of
attitude; they just aren't interested in a career as a fighter pilot or
won't take the time to get better than the day they first flew a
fighter. Yes, there are people who volunteer for a fighter slot just to
punch that hole in the career card. On the RFC side there were a lot of
losses in 1915 simply because the brass had their young pilots flying
crappy aircraft. OTH the Germans had parachutes and the RFC did not,
ever, not in fighters. Lots of factors that would never show up in a
simple statistical analysis, especially if they were ignored to start
with. What a travesty of a scientific study.
Walt BJ
Vanwall
factor, but top fighter pilots, like the top race car drivers and
professional athletes, have an almost exponentially superior abilty
compared to even their less successful contemporaries. It has always
seemed to me that journeymen are at a certain level because they lack
the extra something, that may be undefineable, that the best seem to
have at certain points in their careers. Skill can be augmented by
study and practice, but there will always be a certain few who are just
better on a vastly hgher scale, I don't care what statistics imply.
Rob in Dago
Gernot Hassenpflug
don't have a lot of data... well, we do :-) MEM is far to low
resolution to zoom in on the cirtical differentiating skills.
That said, in the top 1% exprts in any activity, it will be pretty
difficult for a non-expert to judge who is better, be it ballet,
painting, musicianship, or some other pursuit. In martial pursuits,
winning and losing include luck, so in a one-off it may not be so
clear-cut who is better. But taken over a series of encounters between
similarly-skilled participants, those who are the best clearly stands
out. Not that they will win 100 out of 100, but they will win more
percentage-wise than any of the others.
And so you said, skill differs widely, so when such an expert meats up
with the upper 50% of the field rather than only the top 1%, well, he
has a "field day". I think the study would do well to try and separate
out the top 1% from the top 50% and find some interesting conclusions
there.
--
Gernot Hassenpflug (ger...@rish.kyoto-u.ac.jp) Tel: +81 774 38-3866
JSPS Fellow (Rm.403, RISH, Kyoto Uni.) Fax: +81 774 31-8463
www.rish.kyoto-u.ac.jp/radar-group/members/gernot Mob: +81 90 39493924
Leadfoot
"Gordon" <krzta...@aol.com> wrote in message
news:1152915955.5...@m79g2000cwm.googlegroups.com...
> The theory is flawed, as Richtofen was obviously better than 98% of the
> airmen he faced. In ~100 combats, only one aviator brought him down.
>
Richtofen was brought down by ground fire
>
> v/r
> Gordon
>
Cub Driver
<john_...@rochester.rr.com> wrote:
>We find that the variance of
>this skill distribution is not very large, and that the top aces
>achieved their victory scores mostly by luck.
Which rather overlooks the most important factor in a shooter, whether
he's in the air or on the ground: courage, or if that's too difficult
a concept, then a disregard for personal safety.
In every battle, there's a very small percentage of men (women don't
seem to share this characteristic as much, and anyhow the numbers
aren't sufficient to draw a conclusion) who get up close and do the
business.
In the air, this generally means (or used to mean) withholding fire
until the victim is 100 meters distance or less.
Sure, luck helps, and so does skill. But both fade away when the
shooter is up close. As the Irish guerrilla leader Dan Breen famously
remarked: "There are no good or bad shots at twelve paces." Get close
enough, and you create your own luck and skill.
And the funny part is, you may even be safer there.
-- all the best, Dan Ford
email: usenet AT danford DOT net
Warbird's Forum: www.warbirdforum.com
Piper Cub Forum: www.pipercubforum.com
In Search of Lost Time: www.readingproust.com
John Carrier
"WaltBJ" <walt...@mindspring.com> wrote in message
news:1152927994.9...@i42g2000cwa.googlegroups.com...
I like to think of "luck" as what happens when preparation meets
opportunity. Only a few guys really have that total package of
"preparation" and they tend to be dispersed across the entire force. The
Israeli's do better, they have a very competitive evaluation system that
puts the best sticks in their best air-to-air jets.
R / John
Peter Twydell
writes
As tomcervo alluded to in a previous post, he was shot down on 16 July
1917, by 2nd Lt A.E. Woodbridge, the observer in an F.E.2d. The aircraft
was from No. 20 Sqn RFC, and was piloted by Capt. D.C. Cunnell.
Von Richthofen was flying his Albatros D.V (1177/17 ?). A bullet grazed
his skull and it was a close call.
--
Peter
Ying tong iddle-i po!
Ed Rasimus
John Carrier wrote:
>
> I like to think of "luck" as what happens when preparation meets
> opportunity. Only a few guys really have that total package of
> "preparation" and they tend to be dispersed across the entire force. The
> Israeli's do better, they have a very competitive evaluation system that
> puts the best sticks in their best air-to-air jets.
>
> R / John
While at Northrop working training system development for the ATF
(YF-23) proposal, we had a mandatory PhD on the staff to give all of us
fighter pilot types instant credibility. He never quite understood when
we tried to explain to him that a large measure of the difference
between WW II air-to-air kills and Vietnam War A/A kills was that issue
of opportunity.
You can be the best trained, best motivated, most skilled fighter
driver in the world, but if the enemy doesn't show up at the dance, you
don't get kills.
Conversely, you can simply be in the right place at the right time and
the enemy flies in front of you. That's pretty much what happened to
Lt. Dave Waldrip in his F-105 double kill day.
I always believed that common sense and prudence were big players in
the game. While I couldn't always win, it was critical to NEVER lose.
That being said, and having a chance again last May to sit across a
table from guys like Robin Olds (12.5 kills in WW II, 4 in SEA), and
Bob Titus (3 in SEA), it becomes very apparent that skill, motivation,
preparation and aggressiveness are very much in play rather than luck.
When the enemy showed up on their watch, the enemy died.
Rob Arndt
Peter Twydell wrote:
> In message <QV0ug.778$_M.370@fed1read04>, Leadfoot <not...@nospam.net>
> writes
> >
> >"Gordon" <krzta...@aol.com> wrote in message
> >news:1152915955.5...@m79g2000cwm.googlegroups.com...
> >> The theory is flawed, as Richtofen was obviously better than 98% of the
> >> airmen he faced. In ~100 combats, only one aviator brought him down.
> >>
> >
> >Richtofen was brought down by ground fire
> >
> >>
> >> v/r
> >> Gordon
Sorry Gordon but it has since been proven the bullet ending the Red
Knight's life came from enemy ground fire, probably from the same
Australian ground forces that eventually recovered his body and gave
the "Red Baron" a funeral with full military honors.
Rob
BernardZ
john_...@rochester.rr.com says...
Fascinating problem.
Reading the article is trying to explain how Manfred von Richthofen got
89 kills.
The situation is like tosting a coin and getting heads 89 times in a
row. Well good luck in a casino.
Now it states that in WW1 there were 2890 fighter pilots in the German
military.
Assuming random chance, all are equally good so every battle its 50/50
whether they win or lose. The highest expected highest score after 89
battles is about 14.
To get by random chance, 89 kills you need over 20 million German pilots
a number greater then the whole German army.
Now assuming we assume that fighter pilots are rated like chess players
http://en.wikipedia.org/wiki/ELO_rating_system
or national football teams can be
http://www.eloratings.net/
If we assume that an average pilot goes from 0 to 2000 with an average
of 1000 normally distributed running a computer similuation simulation
running to 100 kills we have
32 kills with a rating of 1441 would be a pilot on the top 6.547%
35 kills with a rating of 1488 would be a pilot on the top 4.613%
35 kills with a rating of 1544 would be a pilot on the top 2.904%
56 kills with a rating of 1561 would be a pilot on the top 2.490%
76 kills with a rating of 1665 would be a pilot on the top 0.844%
100 kills with a rating of 1684 would be a pilot on the top 0.665%
According to this analysis, Manfred von Richthofen would have been in
skill over the top 99.2% of all pilots.
--
It is best to travel the path of life with some one else but its nice to
go alone for a short time.
Observations of Bernard - No 101
Ken S. Tucker
http://en.wikipedia.org/wiki/Golf_handicap
Ok, that's the analysis (Bernard's above) the author's question.
What I think the author's main thesis is,
"Skill improves luck", but luck is still a random factor.
like the cliche, "Luck favor's the prepared".
The skill is recognizing a lucky position.
I'd say a card playing game like Euchre demo's
that very well, in that, a skilled player leverages
their luck.
The authors application of that to Air Combat
reminds me of an another thread...about
McNamara.
Regards
Ken S. Tucker
W. D. Allen
Valid statistical analysis can too easily be perverted. Look at all the
bogus public opinion polls. Better yet, compare pre-election polls against
actually election results!
end
"John Bailey" <john_...@rochester.rr.com> wrote in message
news:r55gb21184hp0j3sf...@4ax.com...
Simkin
> Reading the article is trying to explain how Manfred von Richthofen got
> 89 kills.
MvR scored 80 kills.
> The situation is like tosting a coin and getting heads 89 times in a
> row.
A highly biased coin it should be.
The total number of victories credited to Jasta pilots is 6745. The
number of casualties, however, is a lot smaller. They amount to 630 KIA
(killed in action), 52 WIA/DOW (wounded in action and later died of
wounds), 137 POW (prisoner of war), and 431 WIA (wounded in action and
survived). One can safely count KIA, WIA/DOW, and POW as defeats. These
add up to 819. This is by a factor of 8.2 less than the number of
credited victories. It is not clear what fraction of WIA one can add to
defeats, as many wounds were not severe, and the pilots could soon
resume flying combat missions. However, even if we count all of the WIA
as defeats we get 1,250 defeats, which is still by a factor of 5.4 less
than the number of credited victories.
Some possible explanations of why the number of German victories
exceeds the number of German defeats are given in the article. Another
came up in the post by WaltBJ: Germans had parachutes, while allied
pilots did not (I would appreciate a reference).
Assuming that the ratio of the probability of credited victory to the
probability of defeat is 8.2, the probability of victory in each
decisive engagement is 8.2 /9.2 = 0.89. The probability of 80 victories
in a row is ( 0.89)^80 =0.0001. The probability that at least one of
2890 German fighter pilots will achieve 80 or more victories is 1
-(0.9999)^2890 =0.25. Richthofen's score is thus within the reach of
chance.
(BTW all this is explained in the article)
> Assuming random chance, all are equally good so every battle its 50/50
> whether they win or lose.
The 50/50 assumption is wrong as discussed above.
>The highest expected highest score after 89 battles is about 14.
The only possible score after 89 battles is 89.
---
Mikhail Simkin
http://www.ee.ucla.edu/~simkin
BernardZ
simkin_...@hotmail.com says...
> BernardZ wrote:
> > Reading the article is trying to explain how Manfred von Richthofen got
> > 89 kills.
>
> MvR scored 80 kills.
I know I just took it a bit higher
>
> > The situation is like tosting a coin and getting heads 89 times in a
> > row.
>
> A highly biased coin it should be.
>
> The total number of victories credited to Jasta pilots is 6745. The
> number of casualties, however, is a lot smaller. They amount to 630 KIA
> (killed in action), 52 WIA/DOW (wounded in action and later died of
> wounds), 137 POW (prisoner of war), and 431 WIA (wounded in action and
> survived). One can safely count KIA, WIA/DOW, and POW as defeats. These
> add up to 819. This is by a factor of 8.2 less than the number of
> credited victories. It is not clear what fraction of WIA one can add to
> defeats, as many wounds were not severe, and the pilots could soon
> resume flying combat missions. However, even if we count all of the WIA
> as defeats we get 1,250 defeats, which is still by a factor of 5.4 less
> than the number of credited victories.
>
> Some possible explanations of why the number of German victories
> exceeds the number of German defeats are given in the article. Another
> came up in the post by WaltBJ: Germans had parachutes, while allied
> pilots did not (I would appreciate a reference).
A myth really. Various reasons delayed the introduction of parachutes in
WW1 partly a WW1 parachutes were too bulky to be used by pilots in
fighters until the end of the war at a time when weight was very
important and also a widespread mistaken belief that they inhibited
aggressiveness. So its only very late in WW1, that we see the Germans
using parachutes widely but even then, not always.
Actually thinking about it some more, the main reason for the 89% ratio
is over claims is pilots. In action, pilots do tend to overclaim
particularly when they fight, as the red baron did, in large groups.
Check a topic here
North Africa! Where was the Luftwaffe?
rec.aviation.military
>
> Assuming that the ratio of the probability of credited victory to the
> probability of defeat is 8.2, the probability of victory in each
> decisive engagement is 8.2 /9.2 = 0.89. The probability of 80 victories
> in a row is ( 0.89)^80 =0.0001. The probability that at least one of
> 2890 German fighter pilots will achieve 80 or more victories is 1
> -(0.9999)^2890 =0.25. Richthofen's score is thus within the reach of
> chance.
>
> (BTW all this is explained in the article)
> > Assuming random chance, all are equally good so every battle its 50/50
> > whether they win or lose.
>
> The 50/50 assumption is wrong as discussed above.
Following my logic before assume it is random chance that is 50/50 with
2890 pilots over 80 missions, this is to me random then the simulation
program ran
1,000 times gave me the highest score of 25.
20,000 times gave me the highest score of 25.
1,000,000 times gave me the highest score of 30.
No one makes it to 80 kills. So presumably to make it random with 80
kills we need WW1 to run considerably more times then 1,000,000 times
over.
Assuming we go to this concept of 89% (which you argue above as per this
article) with 2890 pilots over 80 missions, the simulation program ran
the chance of 80 kills
1,000,000 times gives a 26%
About 26% is acceptable to suggest random chance but it is not likely.
He would have to be about 90.1%. Which is what the article is saying.
I will change it tonight when I have some time, with this new assumption
as the concept does seem interesting and report back.
>
> >The highest expected highest score after 89 battles is about 14.
>
> The only possible score after 89 battles is 89.
>
Highest expected score does not mean possible score.
I might add that the ELO ratio is an extremely high regarded system
http://en.wikipedia.org/wiki/ELO_rating_system
> ---
> Mikhail Simkin
> http://www.ee.ucla.edu/~simkin
>
>
--
Gordon
Both times? Not.
Gordon
Gordon
Rob Arndt wrote:
> Peter Twydell wrote:
> > In message <QV0ug.778$_M.370@fed1read04>, Leadfoot <not...@nospam.net>
> > writes
> > >
> > >"Gordon" <krzta...@aol.com> wrote in message
> > >news:1152915955.5...@m79g2000cwm.googlegroups.com...
> > >> The theory is flawed, as Richtofen was obviously better than 98% of the
> > >> airmen he faced. In ~100 combats, only one aviator brought him down.
> > >>
> > >
> > >Richtofen was brought down by ground fire
> Sorry Gordon but it has since been proven the bullet ending the Red
> Knight's life came from enemy ground fire, probably from the same
> Australian ground forces that eventually recovered his body and gave
> the "Red Baron" a funeral with full military honors.
Understood, and you will find no greater advocate for the illustrious
Sgt Popkins than myself. However, on an earlier flight, a British
-pilot- managed to shoot the Baron in the head, causing temporary
blindness and resulting in a crash - that was the incident I referred
to in my post.
We display an immaculate DR.I in our museum and are honored that the
family of the fallen Red Baron has given us his actual WWI medals and
awards. Alongside a restored "Spandau" MG, visitors can see the
amazing chest full of medals from many countries and other artifacts.
During meetings on proposed changes to our WWI gallery, I have pushed
for and it appears we may soon get the addition of a life-sized
mannequin of Sgt. Popkins, posed with his Lewis, either beside or
opposite the Baron's DR.I.
Gordon
Cub Driver
wrote:
>The total number of victories credited to Jasta pilots is 6745. The
>number of casualties, however, is a lot smaller.
I suspect that the number of British/French casualties is also a lot
smaller!
vincent p. norris
How do you know which ones are bogus and which ones are not?
> Better yet, compare pre-election polls against actually election results!
My impression is that they are generally pretty close.
vince norris
Billy Beck
BernardZ
> as the concept does seem interesting and report back.
>
>
I am stunned the paper figures work out.
If one assumes an 89% kill rate is standard for German pilots.
W. D. Allen
close.
They are NOT! In fact, the pollsters acknowledge they don't even meet their
own self imposed error margins. The reasons are relatively obvious - first
they poll anyone and everyone, not just likely voters - second they limit
their sample to about 1,000 people. They could improve their results by
significantly more sampling, which is too expensive, whereas they simply
claim plus or minus 4% error [justified by arcane sampling theory] because
the public really has NO idea what a plus/minus 4% error means. As one
polling company CEO admitted to me, the sampling error is always 100%
Got it?
WDA
end
"vincent p. norris" <vp...@psu.edu> wrote in message
news:cvgkb2ta85vr8jtqi...@4ax.com...
Simkin
>
> I suspect that the number of British/French casualties is also a lot
> smaller!
>
Me too. Although I am not sure where one can find the numbers. German
casualties are catalogued in the Grab Street book "Jasta Pilots",
however with regard to the French and British they only published the
lists of their aces.
According to these books, the French and the British fully credited
their pilots for shared victories. In addition RAF officially credited
their pilots for "moral" victories. Not surprisingly, the British got
the highest number of aces of all countries.
So moral/shared victories can explain why Allied victories exceed
Allied casualties.
In contrast, under the German scoring sistem the opponent aircraft must
have been destroyed or forced to land on the German territory and its
crew killed or taken prisoners. Thus the fact that German victories
exceed German casualties is more of a puzzle than in the Allied case.
One possible explanation is that the official scoring rules where not
followed in practice.
Guy Alcala
> Cub Driver wrote:
>
> >
> > I suspect that the number of British/French casualties is also a lot
> > smaller!
> >
>
> Me too. Although I am not sure where one can find the numbers. German
> casualties are catalogued in the Grab Street book "Jasta Pilots",
> however with regard to the French and British they only published the
> lists of their aces.
<snip>
FWIW, here's what wikipedia has to say about RFC casualties in Bloody April
and some other periods:
"During April the British lost 245 aircraft, 211 aircrew killed or missing
and 108 as prisoners of war. The German Air Services lost 66 aircraft from
all causes. As a comparison, in the five months of 1916's Battle of the
Somme the RFC had suffered 576 casualties. Under Richthofen's leadership,
Jasta 11 scored 89 victories during April, over a third of the British
losses.
"The month marked the nadir of the RFC's fortunes. However, despite the
losses inflicted the German Air Service failed to completely stop the RFC
carrying out its prime objectives. The RFC continued to support the army
through the Arras offensive with up-to-date aerial photographs,
reconnaissance information and harassing bombing raids. Within a couple of
months the new technologically advanced generation of fighting scouts (the
SE.5, Sopwith Camel, and SPAD S.XIII) entered service in numbers and
quickly gained ascendency over the over-worked Jastas. More able to
adequately protect the lumbering recce and artillery observation machines,
RFC losses fell as German losses rose."
Guy
bernardz
Maybe May should get credit too for carrying out his instructions. As a
result when Baron von Richthofen followed him, he flew straight too
low over the Australian lines where their sharp shooters could get a
crack at him which could have been Sergeant Popkin or a lucky shot
from an unknown soldier firing his rifle.
bernardz
There is an old saying the harder you work the more lucky, you become.
In thsi case luck is important but even if this analysis is correct its
still means that the Red Baron was a much better, then average fighter.
How much better is the issue?
Simkin
> What absolute drivel.
>These yoyo's can't explain the Mannock, Bishop, Guynemer, etc
> - luck?
>I'll bet these yoyos churned
> numbers and never even talked to one fighter pilot about what it took
> to be good. And then they had the monumental gall to call it scientific
> and even signed their names to the trash.
You protest too much.
>That would be like analyzing Nobel Prize winners
> in physics and coming up with a similar result.
This was already done. See:
http://xxx.lanl.gov/abs/cond-mat/0305150
>Ability, attitude/aggressiveness, physical
> fitness, eyesight,
Actually three German WWI aces wore eyeglasses
Kurt Wintgens (19 victories), Otto Kissenberth (20 victories) , Walter
Kypke (9 victories).
> They need a couple back seat
> rides with some good sticks to see what it's all about. Not that they
> personally would see much . . .
I wouldn't mind. I have private pilot license.
Simkin
> I might add that the ELO ratio is an extremely high regarded system
> http://en.wikipedia.org/wiki/ELO_rating_system
Thanks for the interesting link. In chess we always no who played with
who. In airfight we don't know who downed who (with only a few
exeptions). So ELO can't be used in theory of aces.
Cub Driver
<g_al...@junkpostoffice.pacbell.net> wrote:
>FWIW, here's what wikipedia has to say about RFC casualties in Bloody April
>and some other periods:
Wikipedia: the charming notion that ten thousand monkeys bashing at
ten thousand keyboards can create a reference work.
John Bailey
<john_...@rochester.rr.com> wrote:
>http://xyz.lanl.gov/abs/physics/0607109
>
>Theory of aces: high score by skill or luck?
>Authors: M.V. Simkin, V.P. Roychowdhury
>
>Abstract: We studied the distribution of WWI fighter pilots by the
>number of victories they were credited with along with casualty
>reports.
(snipped)
>When combined with our recent study (cond-mat/0310049), showing
>that fame grows exponentially with victory scores, these results
>(derived from real data) show that both outstanding achievement
>records and resulting fame are mostly due to chance.
One factor in addition to those such as aircraft superiority, tactics,
attitude, etc that has not been adequately emphasized is that of
skill gained by experience.
An engagement won not only results in the losing side needing to
replace the lost pilot with a newbie but gives the winning side a
pilot with one more engagement's worth of experience. If this factor
counts for anything (and I believe it does,) the situation is one with
reinforcing feedback such that eventually, a slight superiority is
amplified. The better a squadron is, the less pilots it loses and at
the same time, the worse a squadron is, the more pilots are replaced
by inexperienced ones. The average skill of the winning side grows
exponentially!
I modeled this situation with Excel and confirmed this intuitive
insight with a numerical demonstration. With only a 1% difference in
initial skill and the same initial skill for all pilots and identical
size of respective squadrons; the slightly superior squadron gains
overwhelming superiority after 100 days of 25 sorties/day.
Extending this aggregate modeling to track individual pilots is a bit
much but it is not hard to see that in such a situation SOMEBODY has
to become the leading scorer and likely by an overwhelming lead.
This is a nature vs nurture situation. The Baron may not have had an
inherent initial superiority but as time passes, his skill resulting
from experiences while winning would result in his having become an
overwhelmingly superior pilot.
A Google spreadsheet is at http://tinyurl.com/paxmc
I will grant access as requested.
For my email go to http://home.rochester.rr.com/jbxroads and click on
the email icon.
John
Peter Twydell
<john_...@rochester.rr.com> writes
One factor skewing results in favour of the German forces was their
practice of concentrating the more skilled pilots in elite units like
Jasta 11. I don't know how widespread this was with the French, but
Groupe de Combat 12 (Les Cigognes) was one elite unit of theirs. AFAIK
the RFC/RNAS/RAF didn't do this at all.
Simkin
John Bailey wrote:
> One factor in addition to those such as aircraft superiority, tactics,
> attitude, etc that has not been adequately emphasized is that of
> skill gained by experience.
It was mentioned in the footnote 4 on page 4 of the paper.
Figure 2 on page 4 shows the average probability of a defeat as a
function of fight number. The probability of a defeat in the first
fight is 23% and it drops strongly for first few fights and does not
change after 10 fights after reaching about 5%.
A hard-core idealist can attribute all of the decrease of the defeat
rate with the number of fights to learning.
In the paper an opposite approach was taken. Learning was completely
ignored. Each pilot was supposed to have an intrinsic skill level (a
concept similar to Darwinian fitness), which does not change with time
The decrease of the defeat rate with the fight number was attributed to
low-skilled pilots being eliminated by natural selection.
If one takes into account learning, MvR will turn out even less unique.
> I modeled this situation with Excel and confirmed this intuitive
> insight with a numerical demonstration. With only a 1% difference in
> initial skill and the same initial skill for all pilots and identical
> size of respective squadrons; the slightly superior squadron gains
> overwhelming superiority after 100 days of 25 sorties/day.
Interesting. But what is your definition of skill? Looks like some
arbitrary variable as they use in ELO rating. You need to tie it to
some observables.
In the paper the definition of skill was the probability of winning a
fight.
Simkin
>
> Wikipedia: the charming notion that ten thousand monkeys bashing at
> ten thousand keyboards can create a reference work.
Apes can produce Modern Art. See:
vincent p. norris
>close.
>
>They are NOT! In fact, the pollsters acknowledge they don't even meet their
>own self imposed error margins.
I've never heard one acknowledge that. Can you supply a source?
>The reasons are relatively obvious - first they poll anyone and everyone, not just likely voters
That's not true of all polls.
>- second they limit their sample to about 1,000 people. They could improve their results by
>significantly more sampling, which is too expensive, whereas they simply
>claim plus or minus 4% error......
You're right about that, but I'd call 4% "pretty close." It might not
be close ENOUGH in a tight election, but for most commercial polling
purposes, it's close enough.
>the public really has NO idea what a plus/minus 4% error means.
I'd hazard a guess that the *interested* public understands that.
> As one polling company CEO admitted to me, the sampling error is always 100%
That's a clever remark but it has no scientific meaning. If it did,
poll results wouldn't be as close as they usually are.
vince norris
bernardz
Disagree.
I would not use it to determine what someone rating is XXX in this case
but you can use it in a simulation, if you know the distribution.
Say you distribution is normal (As this one assumes. Which I am sure is
wrong as it would be shrewed but anyway.) and a S/D which in this case
I confess to fudging a bit.
The mean is arbitrary in the ELO system so I use 1000. I can then
calculate the what the ratings will be in that simulation for every
pilot.
(a)
Rating of enemy pilots =(rnd+rnd+rnd+rnd)*1000 where rnd=random number
>0 and <1.
To find the rating of German of 2890 pilots =
(rnd+rnd+rnd+rnd)*(1000+yyy) where yyy is the Extra rating because of
the reasons which are partly stated in the article.
Then just run the simulation program with various ranges of YYY until
my German pilots it achieves an 89% success rate. I use a binary search
pattern running each value about a 1,000,000 times until I get to 50%
chance of getting a 89% success rate. Which is about 263 from memory.
>From this, I know the distribution of German pilots ratings.
I then take a German pilot and make him BvR. And start him off as
somewhere between a rating of 1263 and 5000 facing the enemy at (a).
Then run the simulation again until he gets 50% of the time, 80 kills.
That is BvR min rating. He could, of course be much higher.
Then I see on a min, how many German pilots are worse then BvR. Which
is what the article was saying
All this does take a long time even with a super fast powerbasic
software running on a good machine but it works. You set it up to run
over night, go to sleep, get up, go to work and then come back that
evening to the answer.
WaltBJ
up Shomo.
Walt BJ
bernardz
But monkeys cannot write Shakespeare.
http://news.bbc.co.uk/2/hi/uk_news/england/devon/3013959.stm
So they cannot write Wikipedia which by the way I find quite good
overall but I do have some issues with parts.
Cub Driver
<john_...@rochester.rr.com> wrote:
>An engagement won not only results in the losing side needing to
>replace the lost pilot with a newbie but gives the winning side a
>pilot with one more engagement's worth of experience.
Very likely, however, the majority of victims were themselves newbies.
Cub Driver
<john_...@rochester.rr.com> wrote:
>One factor in addition to those such as aircraft superiority, tactics,
>attitude, etc that has not been adequately emphasized is that of
>skill gained by experience.
Wartime literature is full of stories about the newbie who joins his
first combat with the rest of his squadron and eventually disengages
with it, never having seen even one enemy aircraft.
I suppose we should distinguish between eras here. The original
question and some responses involved WWI. Most of my knowledge centers
on WWII. Much of that no longer has any relevance; for example, the
then-rule that killers got up close. I believe that now a fighter
pilot rarely even sees his opponent visually, let alone gets within 50
or 100 feet as WWII pilots sometimes claimed to do. (I stressed
"claimed"; in reality, I suspect it was 50 or 100 meters/yards.)
John Bailey
wrote:
>
>John Bailey wrote:
>> One factor in addition to those such as aircraft superiority, tactics,
>> attitude, etc that has not been adequately emphasized is that of
>> skill gained by experience.
>
>It was mentioned in the footnote 4 on page 4 of the paper.
>
Instead of adequately emphasized, I could have said that the dynamic
effect of changing skill was not accounted for at all.
(snipped)
>
>A hard-core idealist can attribute all of the decrease of the defeat
>rate with the number of fights to learning.
>
>In the paper an opposite approach was taken. Learning was completely
>ignored. Each pilot was supposed to have an intrinsic skill level (a
>concept similar to Darwinian fitness), which does not change with time
>The decrease of the defeat rate with the fight number was attributed to
>low-skilled pilots being eliminated by natural selection.
With a fixed skill assumption, the basis for making a Bayesian
inference becomes questionable.
>Interesting. But what is your definition of skill? Looks like some
>arbitrary variable as they use in ELO rating. You need to tie it to
>some observables.
I used the relation p = x/(x+y) where p is the probability the
adversary X will win an engagement with Y where x is the skill level
of X and y is the skill level of Y. For a new pilot, the skill level
is set to 1. In principle these are tied to observables and could be
determined by Bayesian inference.
Simkin
>top fighter pilots, like the top race car drivers and
> professional athletes, have an almost exponentially superior abilty
> compared to even their less successful contemporaries.
A champion weight-lifter can lift just several times more than an
average person.
Similarly, a champion runner can run only several times faster than an
ordinary man.
Thus their ability is not exponentially, but merely linearly superior.
Besides if you compare them to smaller, but still very large group of
relatively able people with proper training - the champions exceed them
by mere percents (in speed, weight, etc)
>It has always
> seemed to me that journeymen are at a certain level because they lack
> the extra something, that may be undefineable,
Yes, undefinable. And this extra something has the wonderful property
of remaining invisible to everyone who is unfit for the office he
helds, or who is extraordinarily simple in character.
Simkin
bernardz wrote:
> I then take a German pilot and make him BvR. And start him off as
> That is BvR min rating. He could, of course be much higher.
> Then I see on a min, how many German pilots are worse then BvR. Which
Who in hell is that? I only know MvR, LvR, and WvR.
Ed Rasimus
wrote:
>Vanwall wrote:
>>top fighter pilots, like the top race car drivers and
>> professional athletes, have an almost exponentially superior abilty
>> compared to even their less successful contemporaries.
>
>A champion weight-lifter can lift just several times more than an
>average person.
>Similarly, a champion runner can run only several times faster than an
>ordinary man.
>
>Thus their ability is not exponentially, but merely linearly superior.
A reasonable parallel, but not quite apt. You are citing simple
physical capabilities and your measure of performance is linear. Your
champion runner can only run one or two times as fast as an average
runner, but what incremental percentage of all runners can run a sub
3'50" mile? or a sub 9.7 second 100 meters?
>
>Besides if you compare them to smaller, but still very large group of
>relatively able people with proper training - the champions exceed them
>by mere percents (in speed, weight, etc)
A small, but still significant group of "relatively able" folks with
"proper training" can become qualified to drive a tactical aircraft.
They will have to have some level of intellectual capacity, some
measure of physical qualification, and some intangible measure of
motivation. At that point, they are probably in the sub 1% of all
folks around. They still aren't "fighter pilots" and they are a long
way from warriors.
>
>>It has always
>> seemed to me that journeymen are at a certain level because they lack
>> the extra something, that may be undefineable,
Journeymen are competent at what they do, but seldom outstanding. In
some instances they lack the skill set, in some they lack the
motivation and for some skilled and motivated, they lack the
opportunity.
But to ascribe success in aerial combat in complex systems as being a
function of "luck" is to not understand the total dynamic
environment.
>
>Yes, undefinable. And this extra something has the wonderful property
>of remaining invisible to everyone who is unfit for the office he
>helds, or who is extraordinarily simple in character.
This last philosophical statement seems more political than related to
the discussion at hand. Possibly I'm overly sensitive today.
Ed Rasimus
Fighter Pilot (USAF-Ret)
"When Thunder Rolled"
www.thunderchief.org
www.thundertales.blogspot.com
bernardz
Whatever
Simkin
Ed Rasimus wrote:
> On 19 Jul 2006 12:56:16 -0700, "Simkin" <simkin_...@hotmail.com>
> wrote:
>
> >Vanwall wrote:
> >>top fighter pilots, like the top race car drivers and
> >> professional athletes, have an almost exponentially superior abilty
> >> compared to even their less successful contemporaries.
> >
> >A champion weight-lifter can lift just several times more than an
> >average person.
> >Similarly, a champion runner can run only several times faster than an
> >ordinary man.
> >
> >Thus their ability is not exponentially, but merely linearly superior.
>
> A reasonable parallel, but not quite apt. You are citing simple
> physical capabilities and your measure of performance is linear. Your
> champion runner can only run one or two times as fast as an average
> runner, but what incremental percentage of all runners can run a sub
> 3'50" mile? or a sub 9.7 second 100 meters?
The parallel is absolutely apt. You are asking how rare a particular
level of ability is, while the issue being discussed is the magnitude
of that level.
In the historical past they did not keep reliable records of people's
height. However, it is safe to say that Robert Pershing Wadlow is the
tallest person of the 20th century. This makes him one out of a few
billions: 0.0000001 percentile level.
However it does not make him a titan as high as a mountain. He was only
8' 11.1", less than two times the height of an average person.
> But to ascribe success in aerial combat in complex systems as being a
> function of "luck" is to not understand the total dynamic
> environment.
Do you yourself understand what you are writing? What is "aerial combat
in complex systems"? What is "total dynamic environment"?
Sounds like some fashionable nonsense to me.
Simkin
> With a fixed skill assumption, the basis for making a Bayesian
> inference becomes questionable.
When one works with real data everything is questionable.
Note, however, that we can make to extreme approximations.
A) Everything is learning. Initial skill is 0.77, if one survived one
battle the skill becomes 0.87 and after ten battles it becomes 0.96.
Such reasoning can "explain" all the data in Fig. 2. MvR is like
everyone else.
B) No learning. The extreme case considered in the article. MvR is in
top 29%.
The reality is somewhere in between.
Ed Rasimus
wrote:
>
>Ed Rasimus wrote:
>
>> But to ascribe success in aerial combat in complex systems as being a
>> function of "luck" is to not understand the total dynamic
>> environment.
>
>Do you yourself understand what you are writing? What is "aerial combat
>in complex systems"? What is "total dynamic environment"?
>
>Sounds like some fashionable nonsense to me.
Aerial combat has always taken place in vehicles which press the
existing state of the art. The complexity of those vehicles has
increased and they have evolved into "weapons systems" rather than
simply airplanes. For someone to succeed in combat in these systems
takes a lot of factors coming together and luck isn't one of those
factors.
The total dynamic environment is the place where aerial combat occurs.
Three dimensional maneuver, physical challenge (see how long you can
function effectively at five to nine Gs), Rules of Engagement, weapons
employment parameters, mutual support between own team members,
awareness of support of enemy team members, weather factors,
predictions of future relationships between aircraft, fuel states,
sensor inputs and capabilities, etc. etc. etc.
If you think that winning in aerial combat is dependent upon luck, you
are failing to appreciate the situation.
If that sounds like fashionable nonsense to you, I can offer you some
suggested reading that might aid you in understanding what aerial
warfare is about. It ain't video games or XBox.
Simkin
> Simkin wrote:
> > Apes can produce Modern Art. See:
> >
> > http://reverent.org/an_artist_or_an_ape.html
>
>
> But monkeys cannot write Shakespeare.
>
written by an illiterate man from Stratford.
Ed Rasimus
<ball...@adelphia.net> wrote:
>"...My impression is that they [public opinion polls] are generally pretty
>close.
>
>They are NOT! In fact, the pollsters acknowledge they don't even meet their
>own self imposed error margins.
Margin of error is a statistical calculation based on sample size. It
isn't "self-imposed." In a scientific poll (which means every member
of the population polled has an equal chance for selection), you can
get 95% reliability (same result 95 out of 100 iterations) and +/- 3%
margin of error with a sample of around 1500. Reduce the sample size
to about 800 and your reliability goes to 93% and margin of error to
+/- 5%.
>The reasons are relatively obvious - first
>they poll anyone and everyone, not just likely voters
The population polled should be described in the documentation of the
poll. "They" don't poll "anyone and everyone" unless they are simply
doing a straw poll for some odd reason. A scientific poll for a
political issue or election typically will state that they polled
"registered voters" or "registered voters who indicate that they are
likely to vote". Man-on-the-street questioning doesn't meet the
criteria for scientific polling.
>- second they limit
>their sample to about 1,000 people.
See above regarding sample size. As long as every member of the
population has an equal chance of selection, it doesn't matter what
the total population size is. Most folks can't grasp that conundrum.
> They could improve their results by
>significantly more sampling, which is too expensive, whereas they simply
>claim plus or minus 4% error [justified by arcane sampling theory] because
>the public really has NO idea what a plus/minus 4% error means.
Beyond that 95% and +/-3 MOE, you really begin to increase sample size
to gain higher number reliability and lower MOE. It does get expensive
and it isn't usually meaningful data.
While the public may not generally have any idea what those numbers
mean, the people paying for the polls know and the ones conducting
them know what standards must be met.
> As one
>polling company CEO admitted to me, the sampling error is always 100%
Did he define what the term "sampling error" meant? Never heard it
before and I've been teaching political science for about ten years
now. Never heard it in grad school either.
>
>Got it?
Got it back?
Rob Arndt
> Some possible explanations of why the number of German victories
> exceeds the number of German defeats are given in the article. Another
> came up in the post by WaltBJ: Germans had parachutes, while allied
> pilots did not (I would appreciate a reference).
> Mikhail Simkin
> http://www.ee.ucla.edu/~simkin
Um... the Germans DID have parachutes from around 1917 forward in
fighter a/c but they did not become standard until 1918; even so, many
failed to open and they were somewhat difficult to control.
Here is a 1918 photo of a German pilot and his parachute:
http://www.spartacus.schoolnet.co.uk/FWWparaP.jpg
Walt is wrong about allied fliers not having them. The Brits did not,
but the French and Americans who flew French a/c were issued early
parachutes. They, however, were used on a smaller scale than the German
type because allied types were considered unreliable. The German ones
weren't that much better but the Germans used them regardless, just
like the ejection seats in WW2.
Rob
Harry Andreas
"Luck is the residue of design"
- Branch Rickey, 1950
If you have tended to your training, and are mentally prepared
then you can take advantage of an opportunity if it presents
itself.
--
Harry Andreas
Engineering raconteur
vincent p. norris
vince norris
Cub Driver
<rasimus...@verizon.net> wrote:
>For someone to succeed in combat in these systems
>takes a lot of factors coming together and luck isn't one of those
>factors.
Well, for any given encounter, luck would be a fairly important
factor, though no doubt diminishing over the decades. In WWII, luck
surely was a significant part of each individual shoot-down.
If the shooter kept on doing it, however, we should be able to
subtract out luck as a factor in his total score. But he was still
subject to luck: as I recall, Der Stern von Afrika was killed in a
crash in a borrowed airplane. Luck got him in the end.
I don't know very much about contemporary weapons systems, but from
what I do know, it all seems much more technical. Now, I suppose, the
luck consists of being an American (Israeli, whatever) fighter pilot
instead of a Libyan (Iraqi, whatever) one.
John Bailey
wrote:
>When one works with real data everything is questionable.
>Note, however, that we can make to extreme approximations.
>
>A) Everything is learning. Initial skill is 0.77, if one survived one
>battle the skill becomes 0.87 and after ten battles it becomes 0.96.
>Such reasoning can "explain" all the data in Fig. 2. MvR is like
>everyone else.
>
>B) No learning. The extreme case considered in the article. MvR is in
>top 29%.
>
>The reality is somewhere in between.
Indeed. After your earlier remarks I have spent some amount of time
trying to find a natural skill measure that grows with experience and
can be used in a symmentrical two valued function to compute the
probability of a victory. The best I came up with is "kills" or
victories. Then p(A downing B)= (k_a+1)/(k_b +1+k_b +1)
In any case, note that my crude model also resolves the issue toward
luck. Pilots with initially equal skills in initially equal squadrons
evolve into a situation in which one squadron has the least
replacements and as a result, its pilots have a higher likelihood of
winning engagements. While the ultimate winningest pilot such as MvR
does indeed have a higher skill level (at the end) than anyone else,
he got there by the luck of facing an opponent's squadron having newer
pilots and by his own luck of winning his first few
engagements--thereby unbalancing the balance of equality. Whether one
prefers the Simkin-Roychowdhury model or accepts the underlying logic
of my crude model, luck wins this engagement--IMHO.
Ed Rasimus
net> wrote:
>On Thu, 20 Jul 2006 21:47:17 GMT, Ed Rasimus
><rasimus...@verizon.net> wrote:
>
>>For someone to succeed in combat in these systems
>>takes a lot of factors coming together and luck isn't one of those
>>factors.
>
>Well, for any given encounter, luck would be a fairly important
>factor, though no doubt diminishing over the decades. In WWII, luck
>surely was a significant part of each individual shoot-down.
>
>If the shooter kept on doing it, however, we should be able to
>subtract out luck as a factor in his total score. But he was still
>subject to luck: as I recall, Der Stern von Afrika was killed in a
>crash in a borrowed airplane. Luck got him in the end.
>
>I don't know very much about contemporary weapons systems, but from
>what I do know, it all seems much more technical. Now, I suppose, the
>luck consists of being an American (Israeli, whatever) fighter pilot
>instead of a Libyan (Iraqi, whatever) one.
Possibly true. But now we're getting into the issue of defining
"luck". Being in the right place at the right time? Certainly an
element of luck involved.
But, knowing your system thoroughly isn't "luck". Consider, for
example, the success or failure statistics of A/A missiles in SEA.
When you factor in the number fired out of parameters, you begin to
adjust the reliability conclusions upward. If you don't employ the
weapon properly, it simply won't work.
Consider the US aces--guys like Ritchie and Cunningham were Fighter
Weapons School and Top Gun instructors. They trained and practiced to
specialize in the air superiority mission. When they got the
opportunity, their victories weren't luck-dependent.
Or equipment--put the TCTO 556 and 566 mods on an F-4E and you quickly
get a much more combat effective A/A system with better ergonomics,
improved switchology, and increased performance. Not luck that makes
the aircraft more likely to win.
Or add dedicated GCI controllers that the aircrew worked with every
day to build confidence and trust. Organize the mission to employ
those assests and use other elements of the force to herd the enemy
aircraft to your designated primary MiG killers. Not luck there
either.
And then throw in that ephemeral quality of courage/aggressiveness to
makes some folks better warriors than others.
Luck might be a player, but other than that rare instance when an
enemy aircraft blunders in front of you and you reflexively "spray and
pray" the usual kill mechanism is skill, training, knowledge,
equipment and courage.
tomcervo
Ed Rasimus wrote:
> Possibly true. But now we're getting into the issue of defining
> "luck". Being in the right place at the right time? Certainly an
> element of luck involved.
>
the world won't help a pilot stationed in a no-combat backwater. Dick
Audet is the poster boy for that situation.
Ed Rasimus
Is it time to introduce: "Fortes fortuna juvat!"
Gordon
> In any case, note that my crude model also resolves the issue toward
> luck.
A good pilot doesn't get "luckier" over time, he becomes more skilled,
his reflexes sharpen, and he gains a better understanding of what will
work against his usual foe. Utilizing a skilled and successful
nightfighter pilot to attack heavily defended bombers in daylight will
get him killed because the first two advantages are unable to offset
the lack of that third component. But if he can use those three
critical advantages in concert, his chances of besting a particular foe
on any given day markedly improve. Think Schnaufer, Glunz, Godfrey,
Cunningham (both of them, Cats Eye and Prisoner #334914).
I think you vastly overrate luck in aerial combat. Why don't we simply
man our interceptor squadrons with repeat lottery winners?
v/r Gordon
Rob Arndt
Gordon,
I'm all with you on daylight warfare but the German nightfighter pilots
had a "bit" of luck with the Schrage-Musik installations!!!
Nothing like auto-racking the belly of an RAF Lancaster from underneath
;)
Rob
Ed Rasimus
To respond to the subject line and not the post, one of the most
experienced F-105G Wild Weasels is "Lucky" Ekman. He was a 1/Lt when
he got his first 100 NVN missions in the F-105D. He extended for about
30 more counters until he was shot down on 30 May 1966 over Yen Bai.
He instructed at McConnell for a short time, then became an F-105F
Weasel and returned (again and again) to SEA.
When I went back to Korat in the F-4E, Lucky was deployed with the
561st WWS from George, flying NVN combat missions still and at that
time wearing a patch on his party suit that said "200
Missions--F-105G" over NVN! Dunno how many total NVN missions he
accumulated, but he is certainly in the top five among all USAF combat
pilots.
But, he had the nickname before he went to war. Dunno the origin, but
it doesn't relate to luck in combat. He knew his job and he did it
well. I flew with him several times in the Hunter/Killer role during
Linebacker I/II.
He has a son and a daughter, both AF fighter pilots--one in F-15s and
one in Vipers.
Maybe it's a genetic thing rather than luck?
Ken S. Tucker
Ed Rasimus wrote:
> On Fri, 21 Jul 2006 05:48:28 -0400, Cub Driver <usenet AT danford DOT
> net> wrote:
> Possibly true. But now we're getting into the issue of defining
> "luck".
Random factors beyond control, a "God Did It"
factor, like swallowing a bird or some transistor
blows, or a tire falls off, how many random things
can go right or wrong in favor of one or the other
the list is endless.
Ken
Peter Twydell
Arndt <B44Th...@aol.com> writes
>
>Gordon wrote:
>> > In any case, note that my crude model also resolves the issue toward
>> > luck.
>>
>> A good pilot doesn't get "luckier" over time, he becomes more skilled,
>> his reflexes sharpen, and he gains a better understanding of what will
>> work against his usual foe. Utilizing a skilled and successful
>> nightfighter pilot to attack heavily defended bombers in daylight will
>> get him killed because the first two advantages are unable to offset
>> the lack of that third component. But if he can use those three
>> critical advantages in concert, his chances of besting a particular foe
>> on any given day markedly improve. Think Schnaufer, Glunz, Godfrey,
>> Cunningham (both of them, Cats Eye and Prisoner #334914).
>>
AFAIK, John Cunningham HATED being called Cat's Eyes. Never had the
nerve to speak to him on the couple of occasions I could have.
>> I think you vastly overrate luck in aerial combat. Why don't we simply
>> man our interceptor squadrons with repeat lottery winners?
>>
>> v/r Gordon
>
>Gordon,
>
>I'm all with you on daylight warfare but the German nightfighter pilots
>had a "bit" of luck with the Schrage-Musik installations!!!
>
>Nothing like auto-racking the belly of an RAF Lancaster from underneath
>;)
>
>Rob
>
You are an unpleasant person, glorifying death and destruction again.
They were no more lucky in having Schräge Musik than RAF pilots were in
having four Hispano cannon.
For a wannabe German, your spelling is as bad as your attitude.
--
Peter
Ying tong iddle-i po!
Ken S. Tucker
Check this out...
http://members.aol.com/warlibrary/vwc0.htm
A mortar round is fired, wind carries it NSE or W,
and "skill, training, knowledge, equipment and courage"
didn't help the vast majority who possessed all of that
in abundance.
Ken
Jeff Crowell
>>Well, for any given encounter, luck would be a fairly important
>>factor, though no doubt diminishing over the decades. In WWII, luck
>>surely was a significant part of each individual shoot-down.
Questionable IMO. Many of the really consistently
successful fighter jocks had worked damn hard to be ready for
those opportunities. You think Thach and O'Hare, to name
a couple, "lucked" into what they did? Smells like skill
combined with much careful thought and hard work to me.
Unless you count as luck that specific combination of talents
and attributes necessary to make a superb fighter pilot, that's
just what Heinlein called the 'meiotic dance."
"Luck" might have more to do with how they met their ends,
O'Hare at night somehow, Marseille is a defective aircraft,
etc., and not so much with their successes.
>>If the shooter kept on doing it, however, we should be able to
>>subtract out luck as a factor in his total score. But he was still
>>subject to luck: as I recall, Der Stern von Afrika was killed in a
>>crash in a borrowed airplane. Luck got him in the end.
Ed Rasimus wrote:
> Possibly true. But now we're getting into the issue of defining
> "luck". Being in the right place at the right time? Certainly an
> element of luck involved.
It seems to me that, at least from the Vietnam War forward, "luck"
was more of the "opportunity" aspect than anything else. What
Cunningham, Ritchie, Olds, et al, accomplished had
NOTHING to do with luck, other than whether that was a
"MiG Day" and if they were blessed with the chance to put all
that aggressiveness, training, and hard work to use.
Jeff
Ed Rasimus
wrote:
>
>Ed Rasimus wrote:
>> Luck might be a player, but other than that rare instance when an
>> enemy aircraft blunders in front of you and you reflexively "spray and
>> pray" the usual kill mechanism is skill, training, knowledge,
>> equipment and courage.
>
>Check this out...
>http://members.aol.com/warlibrary/vwc0.htm
>
>A mortar round is fired, wind carries it NSE or W,
>and "skill, training, knowledge, equipment and courage"
>didn't help the vast majority who possessed all of that
>in abundance.
>Ken
A web site offering Vietnam War casualty statistics doesn't really
seem all that relevant to a discussion of A/A success causes.
I can only assume that your comment after the URL is related to the
randomness of the "golden BB" that gets even the best occasionally.
Les Matheson
subject of a book entitled "Lucky's Bridge"
A great guy, great Commander (I worked for him twice), and really (!!!) into
Porsches.
Les
"Gordon" <krzta...@aol.com> wrote in message
news:1153497344.0...@i3g2000cwc.googlegroups.com...
Gordon
Peter Twydell wrote:
> >>
>
> AFAIK, John Cunningham HATED being called Cat's Eyes. Never had the
> nerve to speak to him on the couple of occasions I could have.
I can confirm that. He felt it insulting to suggest that his eyesight
was somehow more gifted than others - but folks I talked to that flew
with him said balls to that, they were absolutely convinced his
eyesight was the best in the RAF.
v/r
Gordon
vincent p. norris
>and attributes necessary to make a superb fighter pilot, that's
>just what Heinlein called the 'meiotic dance."
But wouldn't you say that the Naval Aviators who arrived in the
Pacific late in WW II, flying F6Fs or F4Us, after the best Japanese
pilots had been shot down or drowned at Coral Sea and Midway, were
"luckier"' than the ones at Guadalcanal flying F4Fs early in the war?
vince norris
Rob Arndt
> >I'm all with you on daylight warfare but the German nightfighter pilots
> >had a "bit" of luck with the Schrage-Musik installations!!!
> >
> >Nothing like auto-racking the belly of an RAF Lancaster from underneath
> >;)
> >
> >Rob
> >
> You are an unpleasant person, glorifying death and destruction again.
>
> They were no more lucky in having Schräge Musik than RAF pilots were in
> having four Hispano cannon.
>
> For a wannabe German, your spelling is as bad as your attitude.
> --
> Peter
Quite the contrary, Schrage-Musik greatly enhanced the night fighter
force firepower and success to the extent that 1/3rd of the late war NF
force were equipped with them:
http://en.wikipedia.org/wiki/Schrage_Musik
The Me Bf 110 regained some lethality and the RAF took losses. It is
said that just 3 rounds of fire from a twim 30mm installation would
bring a heavy bomber down. So skill is reduced while success is
increased- dive under and auto-burst.
Rob
p.s. I'm not using a German keyboard for German terms; apparently,
neither is Wiki looking at the link!
Peter Twydell
Arndt <B44Th...@aol.com> writes
>
>> >I'm all with you on daylight warfare but the German nightfighter pilots
>> >had a "bit" of luck with the Schrage-Musik installations!!!
>> >
>> >Nothing like auto-racking the belly of an RAF Lancaster from underneath
>> >;)
>> >
>> >Rob
>> >
>> You are an unpleasant person, glorifying death and destruction again.
>>
>> They were no more lucky in having Schräge Musik than RAF pilots were in
>> having four Hispano cannon.
>>
>> For a wannabe German, your spelling is as bad as your attitude.
>> --
>> Peter
>
>Quite the contrary, Schrage-Musik greatly enhanced the night fighter
>force firepower and success to the extent that 1/3rd of the late war NF
>force were equipped with them:
>
>http://en.wikipedia.org/wiki/Schrage_Musik
>
>The Me Bf 110 regained some lethality and the RAF took losses. It is
>said that just 3 rounds of fire from a twim 30mm installation would
>bring a heavy bomber down. So skill is reduced while success is
>increased- dive under and auto-burst.
>
>Rob
>
>p.s. I'm not using a German keyboard for German terms; apparently,
>neither is Wiki looking at the link!
>
Whoosh again.
Get a clue.
Gernot Hassenpflug
universe, practice, ability and equipment can place one so far on one
side of the (perhaps very narrow) probability curve that the
likelihood of success is almost 1 (Even when likelhood is 1 this does
not mean certain success). Still, the statistics over many events are
what is important, not one event. I would define bad luck as failure
in the conditions that place you in a position of likelihood of
success greater than 0.5, and good luck as success in conditions in
which likelihood of success is less than 0.5
--
Gernot Hassenpflug (ger...@rish.kyoto-u.ac.jp) Tel: +81 774 38-3866
JSPS Fellow (Rm.403, RISH, Kyoto Uni.) Fax: +81 774 31-8463
www.rish.kyoto-u.ac.jp/radar-group/members/gernot Mob: +81 90 39493924
Gordon
Gernot Hassenpflug wrote:
> On a statistical basis, which is what I think really describes the
> universe, practice, ability and equipment can place one so far on one
> side of the (perhaps very narrow) probability curve that the
> likelihood of success is almost 1 (Even when likelhood is 1 this does
> not mean certain success). Still, the statistics over many events are
> what is important, not one event. I would define bad luck as failure
> in the conditions that place you in a position of likelihood of
> success greater than 0.5, and good luck as success in conditions in
> which likelihood of success is less than 0.5
Now that actually makes sense. Although my brain -does- hurt...
v/r
Gordon
Ken S. Tucker
Yup, stretches neurons. I'm trying to compare
A/A combat to chess, euchre and poker.
There is no luck in chess as there are no
random factors, oops, except who goes first,
aside from that, two equals will draw or average
equal. However suppose once one looses,
(shot down by analogy), then what factor
determines that!? It would be equivalent to a
flip of a coin, or high card wins, IOW's luck.
Suppose we have 8 people playing high card
wins, then 4, then 2, then 1 left holding the
pot, pure luck.
I think the paper adds a sophisticated skill factor
to that, and thus places MvR in the top 29%,
with the balance explained by luck.
WOW...my brain hurts :-)
Ken S. Tucker
Ed Rasimus
wrote:
>Yup, stretches neurons. I'm trying to compare
>A/A combat to chess, euchre and poker.
>There is no luck in chess as there are no
>random factors, oops, except who goes first,
>aside from that, two equals will draw or average
>equal. However suppose once one looses,
>(shot down by analogy), then what factor
>determines that!? It would be equivalent to a
>flip of a coin, or high card wins, IOW's luck.
You are trying to compare a fairly simple, statistically measurable
generally zero-sum game with a very complex scenario. It isn't
impossible, but the branching of the program gets pretty detailed.
Pilot vs pilot, but airplane vs airplane as well. Then weapon vs
weapon and maintenance vs maintenance. Don't forget weather factors
and ROE. Add wing-men and sensors. Make choices of whether to go or
not. Decide whether to press or not. Choose whether to engage or not.
Lose wingman and choose again. Measure fuel against time of engagement
in mission. Determine whether supported aircraft or mission objective
still need high risk level or safer alternative is possible.
Lots different than seeing four clubs on the table and determining if
you should check to fill an inside straight.
Simkin
>I'm trying to compare
> A/A combat to chess, euchre and poker.
> There is no luck in chess as there are no
> random factors, oops, except who goes first,
> aside from that, two equals will draw or average
> equal. However suppose once one looses,
> (shot down by analogy), then what factor
> determines that!? It would be equivalent to a
> flip of a coin, or high card wins, IOW's luck.
The outcome of a chess match in which the two players play many games
between themselves is not random. However the outcome of each
individual game is random.
In aerial combat the chanse plays a far greater role then in chess
because the loser gets killed.
Consider a chess competition with the following rules. The 32 players
are divided into 16 pairs. Each pair plays a game (in the case of a
draw they play again untill someone wins). The 16 loosers are shot, and
16 winners are split into 8 pairs. The procedure is repeated untill
after five stages only one chess player is left alive.
The score distribution at the end of the competition will be as follows
1 player won 5 games
1 player won 4 games
2 players won 3 games
4 players won 2 games
8 players won 1 game
16 players won 0 games
This is the guaranteed distribution of the scores and it will be such
even if all 32 players are equally skilled.
This is what happens in air combat.
> I think the paper adds a sophisticated skill factor
> to that, and thus places MvR in the top 29%,
> with the balance explained by luck.
The skill factor is not sofisticated but is simply defined as the
probability to win a fight.
For example, for 22% of pilots this probability is 50% or less.
On the opposite extreme for 29% of pilots this probability is 97% or
more.
(I remind that the number of victories credited to Jasta pilots exceeds
the
number of casualties (KIA+POW+WIA/DOW) by a factor of 8.)
We estimate that MvR's skill level is between 97% and 98% with 50%
probability. Any surviving 5-victorty ace belongs to the same skill
category
with 35% probability and any 5 victory ace who was killed in his sixth
fight
belongs to the same skill category with 20% probability.
tomcervo
Rob Arndt wrote:
> Quite the contrary, Schrage-Musik greatly enhanced the night fighter
> force firepower and success to the extent that 1/3rd of the late war NF
> force were equipped with them:
>
One of the RAF mavens--Bill Gunston maybe--noted that the Defiant was
the top-scoring nightfighter in 1941, because the turret gunner could
fire into the underside of a bomber from a 45 degree angle up and
forward.
Later on, when the Luftwaffe was doing the very same thing, the Air
Ministry couldn't figure out what was happening to their bombers. The
Canadian boys in 5 Group could, and started sticking .303's and .5's
through ventral hatches.
John Bailey
wrote:
>In aerial combat chance plays a far greater role then in chess
>because the loser gets killed.
Try http://tinyurl.com/nlhj5 for a discussion of the randomness of
chess. Even chess between two chess computers is random--in the
Chaitin/Kolmogorov sense.
August Scientific American has an article The Expert Mind by Philip
Ross which reports analyses of Chess master's and other top
competitor's thinking processes. Apply the concepts provided there
to John (40 second) Boyd's observation-orientation-decision-action
(OODA) loop for aerial combat and the discussion really gets
interesting. http://www.aviation-history.com/airmen/boyd.htm
One last nugget to report. I found a way to track one individual
pilot in my Excel spreadsheet model of competing squadrons. The new
result seems to demonstrate that some initial skill would be needed
to seed MvR's career toward success. Simple equality, depletion of
the enemy squadron and learning could not account for 99 kills. OTOH,
a relatively small margin of superiority would produce apparently
outstanding results.
Which reminds me--while going through basic (class 57-C), I came to
realize that their quick washout policy was effective. It eliminated
those who could not learn quickly.
Ed Rasimus
wrote:
>The outcome of a chess match in which the two players play many games
>between themselves is not random. However the outcome of each
>individual game is random.
>
>In aerial combat the chanse plays a far greater role then in chess
>because the loser gets killed.
>
>Consider a chess competition with the following rules. The 32 players
>are divided into 16 pairs. Each pair plays a game (in the case of a
>draw they play again untill someone wins). The 16 loosers are shot, and
>16 winners are split into 8 pairs. The procedure is repeated untill
>after five stages only one chess player is left alive.
>
>The score distribution at the end of the competition will be as follows
>
>1 player won 5 games
>1 player won 4 games
>2 players won 3 games
>4 players won 2 games
>8 players won 1 game
>16 players won 0 games
>
>This is the guaranteed distribution of the scores and it will be such
>even if all 32 players are equally skilled.
>
>This is what happens in air combat.
But, the outcome of every air combat is NOT a loss for one player.
There are many more air combats in which neither player loses than
those in which a victory is claimed. And, there are also the air
combats in which both players lose--I used to tell students that they
never wanted to enter a scissors with an enemy as more than 50% of the
players lose. A mid-air is a high probability event in a fully
developed scissors.
Further, you don't allow for the 1-v-2 situation or the 1-v-many.
Tactics and doctrine can impact who gains the victory in many
situations. During the Korean War, the USAF doctrine was "fighting
wing" in which the element lead is the primary shooter and the wingman
is relegated to flying a fixed formation position, effectively
occupying the "kill zone" of the leader and denying it to the
enemy--not a good job!
Ken S. Tucker
> August Scientific American has an article The Expert Mind by Philip
> Ross which reports analyses of Chess master's and other top
> competitor's thinking processes. Apply the concepts provided there
> to John (40 second) Boyd's observation-orientation-decision-action
> (OODA) loop for aerial combat and the discussion really gets
> interesting. http://www.aviation-history.com/airmen/boyd.htm
This is a bit OT maybe but the Red Baron had
a manuever that would loop the plane vertically
and end up on the opponents ass, well that's a
scientific tactic. OTOH, in the Boyd article quote,
*Boyd was famous for a maneuver he called "flat-plating the bird." He
would be in the defensive position with a challenger tight on his tail,
both pulling heavy Gs, when he would suddenly pull the stick full aft,
brace his elbows on either side of the cockpit so the stick would not
move laterally, and stomp the rudder. It was as if a manhole cover were
sailing through the air and then suddenly flipped 90 degrees. The
underside of the fuselage, wings, and horizontal stabilizer became a
speed brake that slowed the Hun from 400 knots to 150 knots in seconds.
The pursuing pilot was thrown forward and now Boyd was on his tail
radioing "Guns. Guns. Guns." *
I don't believe an F-100 could do that?
Sounds like a flat spin stall and recovery at a
low speed of 150 knots, then the pursuer would
still have his energy, 400 knots, and he's gone,
but could still turn on the slower Boyd.
Ken
Simkin
> Try http://tinyurl.com/nlhj5 for a discussion of the randomness of
> chess. Even chess between two chess computers is random--in the
> Chaitin/Kolmogorov sense.
You mean pseudo-random, like those numbers generated by Excel RAND()
function?
But if you average a lot of numbers generated by RAND() the result is
pretty deterministic.
There is far less chance in chess than in air combat because, while MvR
was killed after a couple years, these desk-jockeys can hop over
chess-board for decades. The only chess-ace who burned alive was Fisher.
Simkin
>Maximum Entropy Method is good when you really
> don't have a lot of data... well, we do :-) MEM is far to low
> resolution to zoom in on the cirtical differentiating skills.
The Method of Maximum Entropy can be used on any amount of data. When
we have a lot of data (as was the case in the study being discussed)
its resolution is very high.
> That said, in the top 1% exprts in any activity, it will be pretty
> difficult for a non-expert to judge who is better, be it ballet,
> painting, musicianship, or some other pursuit.
Try to give some of these tests ( http://reverent.org/quizzes.html ) to
professional artists, writers, and musicians. The results will surprise
you.
Simkin
> But, the outcome of every air combat is NOT a loss for one player.
> There are many more air combats in which neither player loses than
> those in which a victory is claimed.
I once was landing a Cessna in Quonset airport in Rhode Island and saw
an F-16 taking off. Should I have logged it as a combat mission?
>And, there are also the air
>combats in which both players lose--
That is correct. I recall there were few cases during WWI when two
opponent aircraft collided. This should be taken into account by a more
complete theory, but I think this will give only second-order
correction to the results.
> Further, you don't allow for the 1-v-2 situation or the 1-v-many.
I am not sure that there is enough statistics to include this into
analysis, but this is an interesting question to address.
Ed Rasimus
wrote:
>
>John Bailey wrote:
>
>This is a bit OT maybe but the Red Baron had
>a manuever that would loop the plane vertically
>and end up on the opponents ass, well that's a
>scientific tactic. OTOH, in the Boyd article quote,
Actually that sounds a lot more like Raoul Immelman. A half loop (more
usually a wing-over type maneuver) followed by a half-roll at the top.
He employed it mostly against balloons.
>
>*Boyd was famous for a maneuver he called "flat-plating the bird." He
>would be in the defensive position with a challenger tight on his tail,
>both pulling heavy Gs, when he would suddenly pull the stick full aft,
>brace his elbows on either side of the cockpit so the stick would not
>move laterally, and stomp the rudder. It was as if a manhole cover were
>sailing through the air and then suddenly flipped 90 degrees. The
>underside of the fuselage, wings, and horizontal stabilizer became a
>speed brake that slowed the Hun from 400 knots to 150 knots in seconds.
>The pursuing pilot was thrown forward and now Boyd was on his tail
>radioing "Guns. Guns. Guns." *
>
>I don't believe an F-100 could do that?
>Sounds like a flat spin stall and recovery at a
>low speed of 150 knots, then the pursuer would
>still have his energy, 400 knots, and he's gone,
>but could still turn on the slower Boyd.
>Ken
The F-100 most assuredly can do that as well as the F-105--see my
discussion of the maneuver in When Thunder Rolled, first in Chap 2
"Transition" and later employed in combat in the chapter titled "Of
MiGs and Moustaches".
It can be done in almost any swept wing aircraft and is typically done
at some speed below corner velocity (otherwise you over-G the bird).
Smoothly bring the stick to full aft and when deep into buffet, apply
full rudder in the direction of roll. Done from a turn at the start,
you can roll "over-the-top" counter to your direction of inital turn
or "underneath" in the same direction. The former loses more airspeed
while the latter leaves you nose low and able to accelerate more
easily.
Each will dissipate a whole lot of airspeed in a hurry and when done
defensively in front of someone unfamiliar with the maneuver will
virtually guarantee an overshoot. It is a "last ditch" maneuver and
employed only when your demise is imminent without it!
Paul J. Adam
Simkin <simkin_...@hotmail.com> writes
>Consider a chess competition with the following rules. The 32 players
>are divided into 16 pairs. Each pair plays a game (in the case of a
>draw they play again untill someone wins). The 16 loosers are shot, and
>16 winners are split into 8 pairs. The procedure is repeated untill
>after five stages only one chess player is left alive.
You forgot those occasions where one player suddenly loses half his
pieces mid-match (mechanical failure, for example); where a third player
strolls up and blows the opponent's brains out while they're
concentrating on the board; and being forced into a game with only a few
pawns, one rook and a king against forty or fifty enemy pieces, as well
as many other factors.
Air-to-air combat is not a one-to-one trial of champions to standard
rules; it's a complicated team game where cheating is positively
encouraged.
--
Paul J. Adam
Ken S. Tucker
Ed Rasimus wrote:
> On 23 Jul 2006 10:46:16 -0700, "Ken S. Tucker" <dyna...@vianet.on.ca>
> wrote:
>
> >
> >John Bailey wrote:
> >
> >This is a bit OT maybe but the Red Baron had
> >a manuever that would loop the plane vertically
> >and end up on the opponents ass, well that's a
> >scientific tactic. OTOH, in the Boyd article quote,
>
> Actually that sounds a lot more like Raoul Immelman. A half loop (more
> usually a wing-over type maneuver) followed by a half-roll at the top.
> He employed it mostly against balloons.
Yeah thanks for the ref, ((I know you don't wiki
but it's fast and ;-))...
http://en.wikipedia.org/wiki/Immelmann_loop
I learned the tactic as a 360 vertical loop with
NO roll, ((I never did it, I'm a Cessna-naut)).
Very Cool, thanks Mr. Rasimus, I thought it might
over stress the plane and also possibly the pilot,
(black-out). I'm guessing the plane wouldn't survive
that at 500-600 knots, unless you land with bent
wings! I've tried that on a simulator and hitting 10g's
is easy to do, so I'm guessing you gotta bleed to
400 knots (in a tightning turn) before doing that?
> Ed Rasimus
> Fighter Pilot (USAF-Ret)
> "When Thunder Rolled"
> www.thunderchief.org
> www.thundertales.blogspot.com
Lookin' forward to reading your book.
The best I've read sofar is Karnow's
account of Vietnam, (I think his first
name is Stan), I read it about 5 times,
I'm sure you've read it, and if so what
do you think of it? Me think's it's super!
Regards
Ken
Simkin
> Actually that sounds a lot more like Raoul Immelman. A half loop (more
What are you talking about? Immelmann Roll (turn)? Or did Max become
Raoul in a complex system?
Gernot Hassenpflug
Hello Ed,
Just to add to your point. Statistics has found great utility partly
owing to its application to phenomena that are just to hard to measure
or model in the detail required to predict accurately. For instance,
the physics of an interaction between exactly two particles is known
precisely, but not even the greatest computer people smarter than I
have developted can handle the physics of larger amalgamations of
particles. There's a famous analogy of where does a metal ball end up
if rolled from the top of a slightly rough snowy slope? At each point
of roughness it hits it make end up going one or another direction,
thus that it is impossible for us to predict where it will end
up. Hoever, rolled repeatedly, a kind of bell curve of arrival
positions can be imagined to result, with an average or "most likely"
position, and positions to either side that are not reached as often
out of X numbers of attempts.
You're absolutely right I think, if the system is simple we don't need
statistics. We use it when we don't know all the factors, don't know
all their interactions, and/or cannot actually do the requisite
computations to get useful predictions out. That's when we lump things
together, leave things out (whose effect we don't know) and use
results from assumed similar situations to get a "statistical idea" of
the complex reality. As you know from your field of expertise no
doubt, no-one wants to help prove the statistics, because you never
know where in a statistical curve _your_ particular attempt is going
to end up :-)
Ed Rasimus
wrote:
Caught me! My increasing senility led to a merging of Max Immelman and
Raoul Lufberry. Yes, I'm talking about the Immelman Turn--today an
Immelman is a half loop with a half roll from inverted to upright at
the top. In combat in WW I it was a steeply climbing 180 turn to a
guns shot.
Lufberry's claim to fame is the Lufberry circle in which two aircraft
chase each other around opposite sides of the arc, nose to tail but
with neither gaining an advantage. Today that would be called a
one-circle fight versus a two-circle in which the general flow is
counter to the opponent. Either way, the essence of 1-v-1 theory is to
avoid chasing and get out of the opponent's plane of maneuver, i.e.
yo-yo, quarter-plane or go vertical.