Automatic Cheating Detection in Human Racing

This is a fascinating glimpse of the future of automatic cheating detection in sports:

Maybe you heard about the truly insane false-start controversy in track and field? Devon Allen—a wide receiver for the Philadelphia Eagles—was disqualified from the 110-meter hurdles at the World Athletics Championships a few weeks ago for a false start.

Here’s the problem: You can’t see the false start. Nobody can see the false start. By sight, Allen most definitely does not leave before the gun.

But here’s the thing: World Athletics has determined that it is not possible for someone to push off the block within a tenth of a second of the gun without false starting. They have science that shows it is beyond human capabilities to react that fast. Of course there are those (I’m among them) who would tell you that’s nonsense, that’s pseudoscience, there’s no way that they can limit human capabilities like that. There is science that shows it is humanly impossible to hit a fastball. There was once science that showed human beings could not run a four-minute mile.

Besides, do you know what Devon Allen’s reaction time was? It was 0.99 seconds. One thousandth of a second too fast, according to World Athletics’ science. They’re THAT sure that .01 seconds—and EXACTLY .01 seconds—is the limit of human possibilities that they will disqualify an athlete who has trained his whole life for this moment because he reacted one thousandth of a second faster than they think possible?

We in the computer world are used to this sort of thing. “The computer is always right,” even when it’s obviously wrong. But now computers are leaving the world of keyboards and screens, and this sort of thing will become more pervasive. In sports, computer systems are used to detect when a ball is out of bounds in tennis and other games and when a pitch is a strike in baseball. I’m sure there’s more—are computers detecting first downs in football?—but I’m not enough of a sports person to know them.

Posted on September 21, 2022 at 6:35 AM28 Comments

Comments

Q September 21, 2022 7:33 AM

I tried to watch the video in the article and got this error:
“Video unavailable. The uploader has not made this video available in your country”

But I expect it is just the normal broadcast frame rate of 25 or 30 frames/s. So trying to see something timed down to 1ms won’t be possible when each frame is 33ms or 40ms apart.

MrC September 21, 2022 8:19 AM

Many years ago now I was a swimmer. What Mr. Allen is accused of — anticipating the starting gun — was considered 100% legal, and sometimes even encouraged. If you had so much as a single toe touching the starting block at the moment of the beep, your start was legal. If I recall correctly, at some of the more high-stakes competitions they had a camera wired up to the starting equipment to generate a photo they could check. One’s skill at anticipating the gun was simply considered part of the competition.

This seems to me a better rule. An easy to enforce, bright-line rule. With no arbitrary pseudo-scientific thresholds baked in.

Somewhat related: In my youth, I could consistently achieve eye-to-hand reaction times of 110ms. Even though ear-to-leg is generally slower than eye-to-hand (longer nerve, bigger, slower muscles), it strikes me as entirely plausible that a world-class athlete could hit 99ms on a good day.

ALV September 21, 2022 8:24 AM

Computerized ball tracking is used to adjudicate LBW (leg-before-wicket) out decisions in cricket. But they have a human backstop. Before it goes to the automation, the human umpire says whether in their opinion it is out or not. The umpire’s call is overruled if it is obvious, but if it is marginal, the call stays as the umpire said.

Austin September 21, 2022 8:26 AM

Humans’ ability to anticipate things is one of our most advanced features. This athlete was punished for his.

commenter September 21, 2022 8:37 AM

If there is no typo in a value, we should talk about one hundredths of a second reaction time and not one thousandth as it has been written. Nonetheless, even that value doesn’t change anything in terms of human capability.

Christof September 21, 2022 8:42 AM

If the athlete knows the rules and had the skills to anticipate the gun (by guessing it mind reading?), I would assume he should also be able to time his start to be legal.

He gambled, he lost.

Anonymous Coward September 21, 2022 9:29 AM

I think the article has their decimal in the wrong place.
0.1 is a tenth of a second.
0.099 was the recorded reaction time, not 0.99.

iAPX September 21, 2022 9:36 AM

Aren’t the announcements done before the gun especially needed for the runners to synchronize themselves and have a start as fast as possible, are they?

This is ridiculous.

Chelloveck September 21, 2022 10:10 AM

The speed of sound in air is about 1100 feet/second, or 1.1 ft/ms. If the dispute in reaction time is really down to a thousandth of a second the first thing I’d want to know is where the gun, microphone, and the runner’s ears are relative to each other. Then I’d ask what the air pressure, ambient temperature, relative humidity, and even wind speed and direction are. A combination of those could easily affect the speed of sound by 1ms over the distance from the gun to the runners.

Even if it’s a hundredth of a second, 11 feet is roughly the distance from once side of the track to another. Is the runner on the far side of the track disadvantaged by 1/100 of a second compared to the runner on the near side, just because it takes the sound of the gun that long to get to them?

I’m just an engineer, not an athlete, but this seems like a classic misuse of significant digits. Is human reaction time measured as 0.1s or 0.100s? Those right-hand zeroes do make a difference.

Steve Friedl September 21, 2022 10:22 AM

I’d want to know is where the gun, microphone, and the runner’s ears are relative to each other.

In high-level competitions, there’s a speaker right behind every starting position to provide equal gun-to-ear time for all the competitors.

JonKnowsNothing September 21, 2022 10:52 AM

@All

E-Sports and computer contests all have cheating problems. Sometimes they are called Exploits and sometimes they are called Cheating. It’s a spectrum of problems with reaction times.

In PVP E-Sports combat games there are factors that govern reaction times.

Graphics issues come under all sorts of labels: lag, hitching, rubber banding. Client-Server packet processing. Key pressing and mousing movements. Any delays in computer controlled processing can swing the results of interactions: a winning encounter turns into a defeat. A player with a more powerful set of processors or enhanced aspects may have a better score card than players without such equipment.

There are human reaction times involve. It takes time to process which keys to press and in which order. Hard Core PVP E-Sports players practice long hours to build muscle memory to speed or reduce the delay in keying in specific sequences. There is still human time needed to visually track and analyze the play board. Even with advanced strategy planning, things don’t go according to plan and improvisation is needed.

Winning Strategies

Within each version of E-Sport PVP games is a set of winning strategies. The games are far simpler than chess and the game plans between similar types varies by the distribution of “attack and defense” skills assigned to the computer avatars available in that game. Classical combat games have the Trinity: Tank, DPS (damage dealer), Healer.
Each avatar has some combination of these but their statistical effects will be tilted towards where they sit in the Trinity.

So players learn not only the reaction issues but also which “classes, alts, avatars” fit into a particular combat scenario.

Except this:

With the advent of “Win Always” view points, scripting of many types gets used. It’s not usually forbidden. One key press unleashes 4-10 automated skills. In advanced scripts the skills are executed or played according to the computer timing allotted to that skill by the game developers: (immediate + 2 second cool down), (.08 delay + 0 cool down).

When playing an opponent with such scripts and not having an equivalent script or equivalent equipment or internet connection, this isn’t invisible. Being blasted by 10 unanswerable skills with near guaranteed defeat, is obvious to all. The process supersedes human response time.

An under class of reaction issues happens when very specific tailored programs are used to intercept the packet to-from the server and stuff extra commands into the packet string. Big Security Hole. The player uses their (1key-10actions) and stuffs n-extra sequences of this into the packet going to the server forming a quick iteration loop, resulting in ((1key-10actions)*N). This is also obvious and far beyond human reaction time, even partially assisted reaction times.

There are attempts to use video cameras to monitor player actions, but these are just as easily manipulated to drop a frame or two. There have been flame wars over whether a dropped frame at a crucial time in a sequence was an Exploit, Cheat or Accidental.

For casual PVP, it’s an annoyance. It’s also a learning set of which players to avoid.

When the stakes are $1,000,000 things can get more heated.

Shoal Creek September 21, 2022 1:23 PM

I’ve personally known an amateur cowboy quick-draw shooter that had typical reaction times (hand-eye) in the 80-130 millisecond range. Many fighter pilots are the same. My youngest brother, with no formal training, had a natural hand-eye reaction time of approx. 90 milliseconds when he was in his late teens. His eyes and brain work so quickly that, to this day, he can see and recall details in individual frames of movies playing at 30 fps, aka 33.33 milliseconds per frame. (It’s a very interesting party “trick,” especially with old Disney movies and the occasional subliminal advertising frame slipped into some older movies.) Boxers and martial artists also tend to develop extremely fast reaction times. Bruce Lee is reported to have had a hand-eye reaction time of about 50 milliseconds.

I think that World Athletics didn’t calculate their error using an appropriate statistical model that would account for athletes with reactions that are more than 2-3 standard deviations faster than normal for someone doing a sprint. Statistics say that such people should exist. Is it any surprise that it exists in an athlete whose success may depend on getting up as soon as possible after a ball is snapped?

I hold a conjecture that as we learn more about the relationships between quantum physics and multiple dimensions and how they relate to the human body, we may find that we might not have a speed limit on how the human body can develop ever faster reaction times.

arf'n'arf September 21, 2022 2:10 PM

With high speed cameras at both ends of a running track, the actual elapsed time of a runner can be measure with great accuracy. So it isn’t really necessary use anything to signal the start, just measure the time each runner takes (in longer distance running the terms gun time and mat time are used to describe this).

While it is arguable that it wouldn’t be much of a race if runners could start whenever they like, and it certainly wouldn’t be such a spectacle, there are some sports, such as long jump, where a camera could measure the lift-off point and the touch-down point and completely eliminate the need for a foul line.

SpaceLifeForm September 21, 2022 5:44 PM

Upon further review

I learned long ago that a really good reaction time is .15 seconds, aka 150 milliseconds. One should not expect faster results normally.

Average is higher. The reason that 150 milliseconds is good, is because based upon studies, that is typically the lowest firing time of neurons. A human can not react faster than their neurons can fire.

But, it is a spectrum, and the average does not preclude outliers, that have faster reaction times. Or slower.

I see many people that have reaction times over 1000 milliseconds. They typically are the driver closest to the red light. They are the last to know that the light is now green.

Avoid drama. Drama is stress. Save your brain energy.

Do not allow yourself to be distracted. It is a waste of your brain energy.

‘http://www.neuwritewest.org/blog/4541

‘https://aiimpacts.org/rate-of-neuron-firing/

MrC September 21, 2022 11:11 PM

@SpaceLifeForm: In my youth I could consistently achieve eye-to-hand response times of 110ms. While ear-to-leg is generally slower than eye-to-hand (longer nerve, bigger muscles), I find it totally plausible that a person a couple standard deviations faster than average could hit 99ms on a good day. In fact, I wouldn’t be surprised if such a person’s fastest single sample was quite a big faster than that.

MrC September 21, 2022 11:19 PM

@ Christof:

It’s not mind reading. I wrote a more detailed explanation, but the moderator ate it. So this will have to do: In order to avoid unduly fatiguing the competitors holding the ready position, most starters will hit the buzzer as soon as (1) all the competitors are down into position, (2) all the competitors are still, and (3) the crowd is quiet enough. You can see #1 and #2 in your peripheral vision. You can hear #3. You can attend other events/heats earlier in the day to get a sense of how quiet a particular starter wants the crowd.

Dave September 22, 2022 12:00 AM

F1 racing has a similar problem, if it were up to the engineers the only thing any driver would have to do is press a start button and then the car would drive itself around the track vastly better than any human driver ever could. So the electronics are all heavily crippled to ensure that there’s still a human making a lot of the driving decisions rather than the car driving itself. And, where I came into it, a ton of security on the electronics to make sure they stay crippled so the car doesn’t have an unfair advantage over the crippled electronics in the other cars.

JonKnowsNothing September 22, 2022 12:09 AM

@MrC, @Christof

re: Starting Box Protocols

There must be thousands of different timed events that require Starting Protocol. There is almost always a penalty for leaving early. Sometimes the penalty is large enough to effectively act as a disqualifying score but not causing any negative attributes to the athlete.

USA horse racing generally uses a Starting Gate. The horses are loaded into stalls with a front swing door and a back swing door. The front door does not open until the starter fires the gun. Horses know what’s coming and some will try to chest-bump the door to get it to open. Horses don’t understand the part about the gun, they just know that when the door opens they get to run fast. The doors are built to limit chest-bumps while still having safety aspects where if the horse forces the door open none of the other doors open.

The starter fires the gun when they see all or nearly all the horses have 4 feet touching the ground. The starter doesn’t wait too long for this because the horses get more anxious the longer they are in the gate. And they are off and running…

Some timed events use a string snap barrier. In rodeo roping, the calf gets a head start. A snap string is attached to the calf and when the doors to the calf chute open and the calf runs out the snap string will pull down a separate string barrier that blocks the horse and rider from leaving their starting box. If the horse and rider leave early, the penalty barrier is tripped and a time penalty is applied to the run. Often this time penalty is enough to put the horse and rider far far down the lists so the rider does their best to control the horse from leaving early. The horse knows about chasing the calf but doesn’t understand about the penalty flag. This last part causes a great problem between rider and horse, one that doesn’t end well for the horse. A world class time is 3-5 seconds.

iirc(badly) There was a documentary that examined some interesting indentations in Ancient Greek foot race stadiums. It was a common set of indentations but no one knew what they were for until someone Got A Clue. It was an elaborate race barrier that crossed the entire race path. The indentations held small posts. There was a cord around the top of these posts and it went round and back to a starters box. It was knotted in such a way that the entire gate complex dropped forward along the entire width of the pathway at the same time when the starter yanked the cord.

Getting starter’s orders correct has been a long time issue.

SpaceLifeForm September 22, 2022 12:15 AM

@ MrC

In my youth (long ago), I learned that 150ms was good. That was on a driving simulator. Sometimes, but rarely, I could reach near 140ms. This was Eye to Leg response time. Avoiding an idiot walking in front of your car and hitting the brake before you run them over. I have had it happen. The pedestrian was drunk. He was lucky that I was paying attention and spotted him before it was too late. I had to lock up my brakes. This fortunately was in a 35mph zone where you should never have to lock up your brakes. The guy would have never made it across a highway.

NickP September 22, 2022 3:49 AM

This is worth a listen to:
BBC More or Less: Does the World Athletics Championships have a false start problem?
https://www.bbc.co.uk/programmes/p0cnjn16

In it they show that start times have been decreasing over the last few years and Devon Allen’s start was well within the expected range statistically speaking.

Scott Lewis September 22, 2022 9:05 AM

@arf’n’arf

Your proposal would bypass the technical abilities that separate some world class athletes from others. The fact that one long jumper can get a little closer to the line and rarely foul is in fact a competitive advantage he or she is entitled to. The fact that a sprinter has a faster reaction time versus another is a competitive advantage they are entitled to.

A sprint is not merely about who can run the fastest. Nor is a long jump about solely who can jump the furthest.

Whether it should be acceptable to anticipate a starter gun, is a very interesting question. But eliminating the starter gun who remove part of the skill involved.

David in Toronto September 22, 2022 10:32 AM

@Dave (the other one) you may recall F1 also has a solution to mitigate anticipating the start.

For those unfamiliar with F1 starts, there is a non-racing formation lap when the tires get warmed up and the drivers take their assigned grid positions. Once they are all there the start begins. Five lights go red one at a time until they are all lit. When the lights go all dark racing begins. The timing between all-red and all-dark is randomized.

JonKnowsNothing September 22, 2022 11:20 AM

@NickP, @All

re: start times have been decreasing over the last few years

I wonder if this is an effect of the GoFaster track shoes now permitted in competition?

These are the ultra high tech energy return shoes with special spikes. During early introduction of these shoes used by Elite Class Athletes, world record times dropped like hotcakes on a Saturday morning.

Along with the increased Tech-Advantage Shoes goes the improved track surfaces. Lots of funds are invested in these surfaces and they also provide energy return benefits, either from direct foot contact or less muscle fatigue as the leg pushes against the surface. (1)

iirc(badly) The effect of these shoes was most easily seen in marathon running events. World Elite runners still ran first but at amazing times. The second and third tier runners merged into a single grouping with times well beyond their normal or historical clocked times.

If either of these conditions are present, the start time in question, could well be within these allowed tech advantaged shoe and surface speed improvement scale.

===

1) Track surfaces aka arena footing are big time issues and problems in horse sports. In USA horse racing, the race track is referred to as “dirt” but today, there is little dirt in the track surface. In EU most races are run on “turf” or grass and have a different composition than USA horse racing tracks. Race horses in UK-EU have much longer racing careers with overall, fewer track produced leg and foot injuries than USA tracks where a horse’s racing career is over at 3yrs best case. Other horse sports like Dressage have very specific requirements for the surface and how it is prepared and how it is re-leveled (groomed) between competitors.

In Dressage competitions each “ride” is allotted X-time for the routine, based on the number of moves required by the level of the horse’s training. A competitor is allowed to approach the area after the previous rider exits (governed by a grounds steward) and is allowed to ride around the perimeter to give the horse a chance to spook at the judges stands and flowers decorating the arena. Spooking outside the arena has no penalty, while spooking inside is a major demerit. When the judge(s) signal they are ready a bell is rung informing the rider they may now enter the ring. They have ~60 seconds to pass the perimeter and the enter. Timers are running throughout. Failure to enter within 60 seconds is disqualification. Failure to complete all the moves in the prescribed order and in the prescribed location is also disqualification or serious demerit.

At local training or schooling shows the judges are more lenient with young riders and young horses the latter often imagine that butterflies are monsters coming to get them just when you are supposed to Halt at G.

Max September 22, 2022 6:13 PM

They’re THAT sure that .01 seconds—and EXACTLY .01 seconds—is the limit of human possibilities that they will disqualify an athlete who has trained his whole life for this moment because he reacted one thousandth of a second faster than they think possible?

No, they are THAT sure that 0.1 second is so beyond human’s reaction time capacity, they can safely choose it to be the threshold. Anything below that is GUARANTEED to be cheating without any shadow of doubt.

The Guinness World Record for the fastest response time (punch) is 0.186 seconds. Kick, which is closer to the situation at hand, is 0.476 seconds.

And before you jump on me, I understand that the saccade reaction time will be lower, but it’ll be maybe half that – karate kick measured in the record takes less than 0.2 seconds to execute from beginning of the motion.

Dave September 23, 2022 6:07 AM

@Max: No, they are THAT sure that 0.1 second is so beyond human’s reaction time capacity, they can safely choose it to be the threshold. Anything below that is GUARANTEED to be cheating without any shadow of doubt.

The thing is, they’re not 100% certain of anything involving humans, they just know that 99.999% (or whatever) of humans are above 0.1s. So all that could be occurring here is that they’ve found someone who is a sufficient number of standard deviations away from normal that they can legitimately react in under 0.1s.

And then penalise them for that.

Leave a comment

Login

Allowed HTML <a href="URL"> • <em> <cite> <i> • <strong> <b> • <sub> <sup> • <ul> <ol> <li> • <blockquote> <pre> Markdown Extra syntax via https://michelf.ca/projects/php-markdown/extra/

Sidebar photo of Bruce Schneier by Joe MacInnis.