The Flawed Athens Starting System

[gh]

See front page headlines for an informative article we commissioned on the subject.

[Daisy]

>See front page headlines for an informative article we commissioned on the subject.

Good job on giving this story so more life, it is very clear there is a problem. If IAAF does not look at this seriously maybe they should stop timing races to 100th of a second?

[gh]

doesn't matter if they're timing in millionths or whole seconds; the placing decision is based on when you cross the line, and if you start with a disadvantage, you may finish with it.

[Daisy]

>doesn't matter if they're timing in millionths or whole seconds; the placing
>decision is based on when you cross the line, and if you start with a
>disadvantage, you may finish with it.

True. So the IAAF better take this seriously.

[gh]

the Technical Committee is meeting this weekend; they have been apprised of the existence of the article, although who knows if at this late date they'll have time to get it onto what is probably a crowded agenda.

[Daisy]

>the Technical Committee is meeting this weekend; they have been apprised of the
>existence of the article, although who knows if at this late date they'll have
>time to get it onto what is probably a crowded agenda.

I know this sounds crazy but are the ignoring this issue because there is too much math involved in the problem? Evidence from dubious stats, no mater how obvious the conclusion, is always easy to ignore.

I say ignore since this has been a problem since 1996. Can IAAF really not know that this problem exists?

[Kevin]

Terrific article! The kind of work that makes an engineer/runner proud. If not addressed at this next meeting, this arguement will not fade with age. Thanks for commisioning the work.

[Jon]

Good article.

I definitely wouldn't doubt what Jesus has to say.

[sluhtaf]

Great article. Though I think IAAF won't be able to get this on their agenda at such a late date, so this problem (which should recieve some publicity because of its relevance to faster times) will have to go undiscussed until the next meeting of the Tech Committee.

And by the way, how often does the Technical Committee meet?

[Powell]

The Olympic Games aren't under IAAF's jurisdiction, though - I'm not sure how much influence they have over them. It's no accident that silent gun systems are used at all World Championships, but not at the Olympics.

[Daisy]

>The Olympic Games aren't under IAAF's jurisdiction,

You're right. it is harsh to blame them since they have their own house in order.

i suppose I would have thought they would have had some say in how the olympic track meet is run. Afterall it's still run under their regulations.

[wineturtle]

Bob Hersh may have some insite as he was a member of the Tech Comm.in years past.

[gh]

>The Olympic Games aren't under IAAF's jurisdiction, though - I'm not sure how
>much influence they have over them. It's no accident that silent gun systems
>are used at all World Championships, but not at the Olympics.>>

Powell is correct in that who's providing the equipment is out of the IAAF's hands at the Olympics. However, at least a hint of blame should be tossed their way for not taking the one step mandated by the rules that could have obviated the problem, and that's the stipulation that when you don't have speaker blocks you position the starter so he/she is equidistant (or at least a reasonable facsimile thereof) from all lanes.

I can understand (sort of) why they would say "we've got the technology in place, so we don't need to worry about that" but as a backup, the rules should probably mandate the starter is always in such a position, no matter what.

I believe, by the way, that in some of Dapena's original calculations, he was able to look at the reaction times in various races, and from that, able to plot with relative accuracy (as confirmed by photo evidence we dug up later) where the starter actually was.

[gh]

I don't think I ever mentioned, by the way, that anybody who tries to use a stopwatch in conjunction with the "good" WC starting system hates it, given that it gives virtually no flash and accurate timing is virtually impossible.

[26mi235]

Is it possible for the starting device to provide a visual signal (such as the starting gun's puff of smoke)?

[tafnut]

Some systems have strobe lights as a signal.

[dsrunner]

Jesus,
Just out of curiosity, did any of the apparent discrepancies alter a single race result by place?

[Daisy]

First, I'm from Britain so don't all take this as whining.

The US could well have lost the gold in the mens 4x100m due to the slower reaction time.

[dukehjsteve]

>Some systems have strobe lights as a signal.


That's what they have at all the swim meets.

[J. Dapena]

>Jesus,
Just out of curiosity, did any of the apparent discrepancies alter a
>single race result by place?

Dsrunner: I have only checked the Athens men's 4x100 final for place changes. For that race, the answer is yes: The USA would have beaten Great Britain by 0.03 seconds, and Trinidad/Tobago would have beaten Australia by 0.09 seconds.

The main point, though, is not who got an advantage or a disadvantage from the bad system in this particular race, but the fact that the system does not work. If it helped your team today, it may very well come back to bite your team tomorrow, and vice versa. We need a system that works and is fair to all teams and individual athletes all the time.

[Daisy]

>We need a system that works and is fair to all teams and individual athletes all the time.

Thanks for the great expose. I hope it all pays off with a silent gun in the next olympics. I'm certainly convinced by your argument.

[J. Dapena]

Thanks, Daisy.

Am logging off for the night.

[wineturtle]

It's the Buzzer lap and fasthips starts his kick........nah!

[Paul1]

For what it's worth, I know that I was taught to focus upon the sound of the gun. Could it be that the hardware is operating properly but the runner is mentally focused upon the "gun sound" and not a ping? In this scenario, the ping is ignored because the mind believes it may be a false sound. In the silent system there is no confliction of sound. Having said this, it seems logical that not all runners would have this problem, but it's worth a thought.

[gh]

The blocks don't "ping"; they emit a gun-like sound.

[Wayne T. Armbrust]

I'm surprised that no one has a theory why the "Loud Gun" system doesn't work. Wouldn't it be a rather simple matter to obtain one of those systems and test it. If it really doesn't work, it ought to be easy to find out why.

[Suso2]

Wayne:

It's not so easy to get a hold of the system. Normally such systems are just rented out for a big track meet, and then they are carted away by the manufacturer. But if you want theories, here are some:

(1) I just had a talk with an electronics specialist at my university, Larry Robinson, who read the web page article, and he had some interesting ideas. If the microphone is sound-activated, it takes some time for the activation to occur. The sound of the gun may be brief enough that, by the time that the microphone starts trying to pick up sound, there is no sound to be picked up at all, except perhaps for some residual from the "bang" --and I don't know what that residual might sound like.

This may fit with the story that Phil Henson once told me about Ed Moses. When starting blocks first came into use, it was noticed that Ed Moses, in addition to running like a demon, also had much better starting times than anyone else. When asked about it, he answered that the sound he heard through the speakers sounded to him like shhhhhhhhh-BANG! The other runners started with the BANG; he started with the "shhhhh". If the story is true, Moses may have been reacting to the residual of the bang that came through the loudspeaker (the "shhhhh"), while everyone else was reacting to the full BANG that came later through the air. Just a thought ...

(2) Larry also said that certain kinds of microphones (called condenser microphones) may stop sound transmission altogether if the loudness of the sound goes above a certain threshold. Maybe the sound of the gun temporarily mutes the microphone.

(3) Maybe somebody is deliberately disconnecting the loudspeakers after the official finishes the verbal commands and before the gun goes off, to prevent the danger of any athlete thinking mistakenly that the separate sounds from the speaker and from the air represent a false start. This would be an unconscionable practice, and I am not accusing anybody of doing it. I am just trying to be thorough in all the possibilities!

(4) Maybe the athlete hears a bang coming from the loudspeaker, but decides not to react to it because the sound is supposed to come from the direction of the guy with the gun. I don't believe this possibility at all. Again, I am just trying to be as thorough as possible.

(5) Maybe the wires are not capable of transmitting a strong enough electrical impulse to the speaker.

There are probably other possibilities.

[Wayne T. Armbrust]

Of the five theories mentioned, only the first two seem at all possible.

It's hard to understand how a major manufacturer of timing equipment would field apparatus for something like the Olympics that didn't work consistently. However, the evidence presented by Dr. Dapena is compelling.

>Wayne:

It's not so easy to get a hold of the system. Normally such systems
>are just rented out for a big track meet, and then they are carted away by the
>manufacturer. But if you want theories, here are some:

(1) I just had a
>talk with an electronics specialist at my university, Larry Robinson, who read
>the web page article, and he had some interesting ideas. If the microphone is
>sound-activated, it takes some time for the activation to occur. The sound of
>the gun may be brief enough that, by the time that the microphone starts trying
>to pick up sound, there is no sound to be picked up at all, except perhaps for
>some residual from the "bang" --and I don't know what that residual might
>sound like.

This may fit with the story that Phil Henson once told me about
>Ed Moses. When starting blocks first came into use, it was noticed that Ed
>Moses, in addition to running like a demon, also had much better starting times
>than anyone else. When asked about it, he answered that the sound he heard
>through the speakers sounded to him like shhhhhhhhh-BANG! The other runners
>started with the BANG; he started with the "shhhhh". If the story is true,
>Moses may have been reacting to the residual of the bang that came through the
>loudspeaker (the "shhhhh"), while everyone else was reacting to the full BANG
>that came later through the air. Just a thought ...

(2) Larry also said
>that certain kinds of microphones (called condenser microphones) may stop sound
>transmission altogether if the loudness of the sound goes above a certain
>threshold. Maybe the sound of the gun temporarily mutes the
>microphone.

(3) Maybe somebody is deliberately disconnecting the
>loudspeakers after the official finishes the verbal commands and before the gun
>goes off, to prevent the danger of any athlete thinking mistakenly that the
>separate sounds from the speaker and from the air represent a false start.
>This would be an unconscionable practice, and I am not accusing anybody of
>doing it. I am just trying to be thorough in all the possibilities!

(4)
>Maybe the athlete hears a bang coming from the loudspeaker, but decides not to
>react to it because the sound is supposed to come from the direction of the
>guy with the gun. I don't believe this possibility at all. Again, I am just
>trying to be as thorough as possible.

(5) Maybe the wires are not capable
>of transmitting a strong enough electrical impulse to the speaker.

There
>are probably other possibilities.

[Daisy]

>Of the five theories mentioned, only the first two seem at all possible.

>It's hard to understand how a major manufacturer of timing equipment would
>field apparatus for something like the Olympics that didn't work consistently.
>However, the evidence presented by Dr. Dapena is
>s compelling.


I have to say that my favored one is number 3. Or they just forget to plug them in.

My reasoning is that if you look at the reaction times for the athens 4x100m semi finals the reaction times do not seem that skewed, espcially when compared to the final. If this is so, then the speakers could be working in some races but no in others. That would rule out a design flaw and mean human error could be significant.

Here are the reaction times for the 4x100m

lane.....semi#1........semi#2.........final ----> semi time
1..........0.152..........0.162..........0.199 ----> 0.199
2..........0.211..........0.212..........0.159 ----> 0.138
3..........0.156..........0.146..........0.232 ----> 0.162
4..........0.149..........0.206..........0.266 ----> 0.211
5..........0.237..........0.153..........0.300 ----> 0.289
6..........0.158..........0.289..........0.295 ----> 0.152
7..........0.199..........0.406..........0.372 ----> 0.149
8..........0.252..........0.138..........0.344 ----> 0.156

The final colomn is a repost of the semifinal reaction time for each athlete in the final (My assumption is that the lead off runner is the same in each case. I think this is true for the eight teams concerned). Note that the differential between the athletes final and semi reaction time increases as the athletes are further from the starter (lane 8 furthest from starter).

Again, I think it is possible that the speakers were working in the semi finals but not for the final. Human error? For sure, it can explain how the system passes the required tests and yet can fail when it really counts.



Finally:
An easy solution to this problem is to adjust the postion of the starter relative to the atheletes! This has been touched on in a previous thread. If the starters position is optimal, i.e. roughly equidistant from all the athletes, then none of these problems would exist. Apparently, when speakers used, the starter is positioned some where on the straight, thus, maximising the problem if the speakers do fail. If IOC insist on using the loud gun then starters should at least be required to position themselves more optimally to safeguard against the speakers failing.

[dj]

The one problem -- not that it should be -- with using a focal starting position, particularly for the relays, is that schedulers seem to love to place a high jump simultaneously.

The high jump is usually in the first turn, and a focal point for the relay starts would often posiiton the starter in or near the right-hand j approach.

Solution: reschedule the high jump and put the starter where he always ought to be.

[26mi235]

Great series of posts. This should be required reading, with a quiz afterwards, for the IAAF/IOC. I still have not seen an answer to how the results would be affected (except that a beginning thread was that the US 4x100 would have been ahead of the UK).

[Suso2]

Daisy's data to some extent raise doubts about whether the starting system actually failed in the Athens men's 4x100 semifinals. However, there are other indications that the system did fail in the Athens semininals. From Daisy's data, we can calculate that:

The median of all lanes in semi#1 was (0.158+0.199/2=) 0.179.
The median of all lanes in semi#2 was (0.162+0.206/2=) 0.184.
The median of the last column was (0.156+0.162/2=) 0.159.
(The final was awful, of course, I will not discuss it here.)

All these values are higher than what we would expect: medians of 0.145 and 0.146 for Goeteborg and Paris --meets with silent gun.

Also, I don't have more detailed data for Goeteborg, but in Paris, out of the total of 51 starting times in the men's seven 4x100 races (3 rounds), only ONE time was over 0.200 (0.229 in lane 8 of the 2nd SF). By contrast, in the three races in Athens there were 13 starting times over 0.200, and 7 of them were in the semis. So, even though, as Daisy shows, the starting times in the Athens semifinals were not as bad as those in the final, they were still bad in relation to silent gun meets, so it seems that something was wrong in the system in the Athens semis, not only in the final.

With respect to the position of the starter, yes, it would be good for the starter to be in a position more or less equidistant from all the athletes ("focal starting position"). That would make it fair for all the athletes relative to each other. However, it would still be detrimental to all the TIMES --equally detrimental, but still detrimental. So the ideal would be to fix the starting system.

26mi235: See J. Dapena's email in this thread: the positions of USA and Great Britain got switched, and also the positions of Australia and Trinidad. One way to look at this is that half the teams in the Athens men's 4x100 final got the wrong places.

[Daisy]

>So, even though, as Daisy
>shows, the starting times in the Athens semifinals were not as bad as those in
>the final, they were still bad in relation to silent gun meets, so it seems
>that something was wrong in the system in the Athens semis, not only in the
>final.

Hi Suso, yes i had considered the fact there were still slow reaction times. It's interesting to hear how they compare with the Paris and Gothenburg times. How do the reaction times compare if you average the middle four times? I am entering into the realm of how to lie with statistics here, of course.

Although i did not address the issue above, my original reason for discounting your argument was the 0.138 in lane eight for semi final 2. Doesn't that seem a bit fast if the speakers are not functioning? Of course, it could easily have been a flier and this is the obvious problem of looking at individuals in different races. Each circumstance is bound to be different. Nevertheless, that same athlete had a respectable 0.159 reaction time in the final when we can be sure the speakers were not working.

[26mi235]

An additional "input" into any analyses is that the center lanes are usually seeded with higher ranking/placing entrants. They might be expected to have, on average, better startng reactions.

[jla]

>An additional "input" into any analyses is that the center lanes are usually
>seeded with higher ranking/placing entrants. They might be expected to have,
>on average, better startng reactions.

Better sprinters do not have better reaction times. Reaction times are not "trainable" but due to how the human being is "wired". The spread in true reaction times is extremely small - something of the order of 0.02/0.03 - between individual sprinters while their 100m times may vary by half a second or more.

It is worth noting that a certain Ben Johnson had the best reaction times already in the 1983 World Championships but no one noticed it because he was not good enough as a runner then being eliminated already in the semis.

[Suso2]

Jla has got it pretty much right. If there is any lowering of the start times in the middle lanes, it is imperceptible.

Daisy: I'm at home now, and the data are in my office. I'll try to remember to get you the data tomorrow.

[jla]

>The one problem -- not that it should be -- with using a focal starting
>position, particularly for the relays, is that schedulers seem to love to place
>a high jump simultaneously.
The high jump is usually in the first turn,
>and a focal point for the relay starts would often posiiton the starter in or
>near the right-hand j approach.
Solution: reschedule the high jump and put
>the starter where he always ought to be.

Although the focal point position for the starter might have the advantage of being fair in the sense of providing a bang that is delayed by the same amount for all runners it is actually an absolutely impossible place for performing the starter's principal duties!

IAAF Rule 129.3 states that "The Starter shall position himself so that he has full visual control over all athletes during the starting process. ....
Note: The Starter shall place himself so that the whole field of athletes falls into a narrow visual angle. For races using a crouch start it is necessary that he is so placed that he can ascertain that all athletes are currently steady in their set positions before the gun is fired."

The focal point position makes it absolutely impossible (even with Marty Feldman type of eyes) to simultaneously see more than three (perhaps four) lanes out of the eight. A starter using such a position is forced to guess if the runners in all lanes are "Set" and still (and he will not be able to observe a false start in the at least 50% of lanes outside of his visual control).

He is then violating Rule 162.2 which states that "All races shall be started by the report of the Starter's gun ... fired upwards AFTER HE HAS ASCERTAINED that athletes are steady and in the correct starting position".

This ascertaining could only be done by the starter personally because if he would use any of his assistants to cover some lanes and then "report" the procedure would take several seconds which would result in the situation having changed significantly in the mean time.

So while the focal point position might solve one problem it will create several others that are more severe.

However, the "silent gun" set-up solves all problems at the same time. Not only does all the athletes get the starting sound simultaneously, they get it with minimal delay - and the starter is free to position himself wherever he finds best to be able to perform all his duties.

PS. However, it is correct that one always should avoid scheduling the HJ when relays are starting in the same bend. However, this is for the good of the high jumpers who always seem to end up having to wait and wait and wait for sprinters running across and then walking back across (the HJ run-up path) while practicing starts.

[Suso2]

Daisy:

Here are the data that you asked for (Athens vs Paris):

Athens:

lane.....semi#1........semi#2.........final
1..........0.152..........0.162..........0.199
2..........0.211..........0.212..........0.159
3..........0.156..........0.146..........0.232
4..........0.149..........0.206..........0.266
5..........0.237..........0.153..........0.300
6..........0.158..........0.289..........0.295
7..........0.199..........0.406..........0.327 <---notice, 0.327 instead of 0.372
8..........0.252..........0.138..........0.344

averages of middle four times:
SF1 = 0.181
SF2 = 0.183
F = 0.273


For Paris, the corresponding values were:

lane.....semi#1........semi#2.........final
1..........0.172..........0.125..........0.132
2..........0.163..........0.144..........0.160
3..........0.134..........0.145..........0.146
4..........0.148..........0.139..........0.141
5..........0.150..........0.150..........0.150
6..........0.138..........0.131..........0.138
7..........0.138..........0.145..........0.145
8..........0.179..........0.229..........0.172

averages of middle four times:
SF1 = 0.150
SF2 = 0.143
F = 0.146

So, basically, we are getting about the same values as with the medians. It's pretty hard to lie with these statistics. Also, notice how consistently small the Paris times are.

>Although i did not address the issue above, my original
>reason for discounting your argument was the 0.138 in
>lane eight for semi final 2. Doesn't that seem a bit fast if
>the speakers are not functioning? Of course, it could easily
>have been a flier and this is the obvious problem of looking
>at individuals in different races. Each circumstance is bound
>to be different. Nevertheless, that same athlete had a
>respectable 0.159 reaction time in the final when we can be
>sure the speakers were not working.

That was Souza, from Brazil. Yes, I would say that his 0.138 from lane 8 was almost certainly a flier. In the final, he was in lane 2, so his disadvantage with respect to the gun was smaller. And he made a perfectly legal start in the final, in 0.159. I don't see any problem with this.

When you compare single races, you can easily get unclear results, because of the variability between athletes' physiological capabilities, and because of the variability of one athlete from one race to another. That is why it is important to look at a whole bunch of races, to allow this random variability to cancel itself out. (Large values in one race will tend to get partially cancelled out by small values in other races.)

[gh]

<<3) Maybe somebody is deliberately disconnecting the
>loudspeakers after the official finishes the verbal commands and before the gun
>goes off, to prevent the danger of any athlete thinking mistakenly that the
>separate sounds from the speaker and from the air represent a false start.
>This would be an unconscionable practice, and I am not accusing anybody of
>doing it. I am just trying to be thorough in all the possibilities!>>

I was chatting with an Apple engineer and while this is far from the kind of work he puts his EE degree to, he suggested that the first part of your first sentence is right. Yes, they're purposely turned off after the verbal comands. Or, in fact, not actuall hooked to the gun at all in the first place. Not because of "mistaken sounds" problem, but because the technology to carry something as huge as the gun sound down the wire (and reproduce it accurately in such small speakers) isn't there. That make any sense?

[wineturtle]

I was chatting with
>an Apple engineer and while this is far from the kind of work he puts his EE
>degree to, he suggested that the first part of your first sentence is right.
>Yes, they're purposely turned off after the verbal comands. Or, in fact, not
>actuall hooked to the gun at all in the first place. Not because of "mistaken
>sounds" problem, but because the ***TECHNOLOGYto carry something as huge as the
>gun sound down the wire (and reproduce it accurately in such small speakers)
IS NOT THERE*** That make any sense?

Gary are you saying that the impulse electricly starts the timer but the runners go at the sound from the gun not from the sound from the speakers behind the blocks?
Guess we need to talk with the builders of these systems not just the PR people.