Photo Finish
July 2, 2012 8:42 AM Subscribe
At the USA Track & Field Olympic Trials last week in Eugene, OR, Jeneba Tarmoh and Allyson Felix tied for third in the Women's 100m. In an 11 second long race, how is a tie determined? Roger Jennings, photo finish evaluator, explains how it's done.
USATF, up until last week, had no tie-breaker plans on the books. A run-off is scheduled for tonight.
USATF, up until last week, had no tie-breaker plans on the books. A run-off is scheduled for tonight.
The video was pretty interesting, and the article should be, too, but the prose is a fine example of everything that's wrong with modern sports journalism.
posted by muddgirl at 8:50 AM on July 2, 2012
posted by muddgirl at 8:50 AM on July 2, 2012
Here's a photo of the finish - I guess the runner's torso crossing the line is what counts.
posted by exogenous at 8:54 AM on July 2, 2012
posted by exogenous at 8:54 AM on July 2, 2012
Here's an article about slit-scan finish line cameras, with this race as an example.
These cameras just record a vertical line exactly over the finish line, so every racer's body is recorded only as that part crosses the line.
From the article:
OK, so now imagine taking a flatbed scanner sensor and setting it up vertically, looking across a racetrack at the finish line. Start a "scan", and it'll authoritatively tell you when every body-part of every runner makes it to the finish, by simply showing that part of that person before any part of anyone else.
...
Runner #2's foot got to the finish first, and because it was then planted on the ground it looks ridiculously elongated. Runner #1's left foot was moving forward as it crossed the line, and so it's shrunk.
posted by jjj606 at 9:04 AM on July 2, 2012
These cameras just record a vertical line exactly over the finish line, so every racer's body is recorded only as that part crosses the line.
From the article:
OK, so now imagine taking a flatbed scanner sensor and setting it up vertically, looking across a racetrack at the finish line. Start a "scan", and it'll authoritatively tell you when every body-part of every runner makes it to the finish, by simply showing that part of that person before any part of anyone else.
...
Runner #2's foot got to the finish first, and because it was then planted on the ground it looks ridiculously elongated. Runner #1's left foot was moving forward as it crossed the line, and so it's shrunk.
posted by jjj606 at 9:04 AM on July 2, 2012
They can also toss a coin to decide who goes through, but opted for a head to head race instead.
posted by notseamus at 9:07 AM on July 2, 2012
posted by notseamus at 9:07 AM on July 2, 2012
Wow, just crazy.
I applaud the timekeeper, who made a guess as to the torso positions, which initially called the race for Tarmoh, but also immediately appealed his own ruling, because of the stakes and the fact that he had to make an educated guess.
I can understand not having a tie breaker in place when you're getting 1/3000second resolution -- though it did make everyone look silly. However, I think the tiebreak procedure is pretty clear -- if both agree, they'll do one option, if they disagree they run, if neither cares, they'll flip a coin.
And really, at this point? They're less than 1/3000th apart. That's the point where how and where the athlete's name and number are attached to their clothing can change the result.
Calling it a tie, I think, is the right thing to do.
I guess the runner's torso crossing the line is what counts.
Yes, and that's the great untraceable problem -- the race is how fast a runner you are, not how fast you are to a certain point where you can throw a hand over the line, but unlike cycling, racing or skating, there's no clear fixed point that you could use to decide "This runner is across the line."
There is a point that, theoretically, could be used -- the center of mass. But how to mark that when the surface of the person is in constant motion.
Probably closest would be a waist belt, which I'm guessing is the hardest part of your body to throw forward without seriously compromising your ability to run.
But mod something like that, there's always going to be a judgement element to the race -- where is the torso? And to be honest, that's a standard that's worked very well in the past, but here, it's just not clear, and I think when you have to stare at something for more than a minute and you can't be sure who crossed first, when you're looking at a 3000fps still, at that point, it's a "beyond resolution" tie. It may be that one runner did cross a fraction before, but that fraction is less than 1/3000th, and I don't really have a problem saying "If that's not enough resolution, it's a tie."
posted by eriko at 9:10 AM on July 2, 2012 [1 favorite]
I applaud the timekeeper, who made a guess as to the torso positions, which initially called the race for Tarmoh, but also immediately appealed his own ruling, because of the stakes and the fact that he had to make an educated guess.
I can understand not having a tie breaker in place when you're getting 1/3000second resolution -- though it did make everyone look silly. However, I think the tiebreak procedure is pretty clear -- if both agree, they'll do one option, if they disagree they run, if neither cares, they'll flip a coin.
And really, at this point? They're less than 1/3000th apart. That's the point where how and where the athlete's name and number are attached to their clothing can change the result.
Calling it a tie, I think, is the right thing to do.
I guess the runner's torso crossing the line is what counts.
Yes, and that's the great untraceable problem -- the race is how fast a runner you are, not how fast you are to a certain point where you can throw a hand over the line, but unlike cycling, racing or skating, there's no clear fixed point that you could use to decide "This runner is across the line."
There is a point that, theoretically, could be used -- the center of mass. But how to mark that when the surface of the person is in constant motion.
Probably closest would be a waist belt, which I'm guessing is the hardest part of your body to throw forward without seriously compromising your ability to run.
But mod something like that, there's always going to be a judgement element to the race -- where is the torso? And to be honest, that's a standard that's worked very well in the past, but here, it's just not clear, and I think when you have to stare at something for more than a minute and you can't be sure who crossed first, when you're looking at a 3000fps still, at that point, it's a "beyond resolution" tie. It may be that one runner did cross a fraction before, but that fraction is less than 1/3000th, and I don't really have a problem saying "If that's not enough resolution, it's a tie."
posted by eriko at 9:10 AM on July 2, 2012 [1 favorite]
this photo of a triple dead heat horse race from Wikipedia is cool.
posted by scose at 9:14 AM on July 2, 2012 [3 favorites]
posted by scose at 9:14 AM on July 2, 2012 [3 favorites]
Does anyone really think that 1/3000 of a second, or even 1/100 of a second, is meaningful in a single measurement of human speed? It's like one part in 30000, or .003%.
I think they should hold these races over a matter of days and histogram them or something.
posted by DU at 9:15 AM on July 2, 2012 [2 favorites]
I think they should hold these races over a matter of days and histogram them or something.
posted by DU at 9:15 AM on July 2, 2012 [2 favorites]
Ties do happen in track. But a tie isn't an option here, because the third-place finisher gets the spot on the U.S. 100m team.
posted by Doofus Magoo at 10:07 AM on July 2, 2012 [1 favorite]
posted by Doofus Magoo at 10:07 AM on July 2, 2012 [1 favorite]
I think they should each be asked to write a 400 word essay on why they want the place and why it would make them a better person, the Olympic team stronger and America a better country to live in.
It could then be judged on live on breakfast tv to see who should go.
posted by MuffinMan at 10:39 AM on July 2, 2012 [2 favorites]
It could then be judged on live on breakfast tv to see who should go.
posted by MuffinMan at 10:39 AM on July 2, 2012 [2 favorites]
USA Track & Field says Tarmoh is officially withdrawn from the run-off.
posted by Lame_username at 10:40 AM on July 2, 2012
posted by Lame_username at 10:40 AM on July 2, 2012
Well, that kind of bums me out. Why not run the run-off? You're a runner. It's what you do.
posted by vibrotronica at 11:07 AM on July 2, 2012
posted by vibrotronica at 11:07 AM on July 2, 2012
Didn't she qualify for the 200? I heard that she doesn't want to risk injury.
posted by muddgirl at 11:13 AM on July 2, 2012
posted by muddgirl at 11:13 AM on July 2, 2012
No, that was Felix who won the 200 qualifier. Tarmoh is going to the games, but as a relay runner only.
posted by JanetLand at 11:18 AM on July 2, 2012
posted by JanetLand at 11:18 AM on July 2, 2012
I am sure that people are lined up there right now, and disappointed.
It was going to be free admin, with only the West Grandstand open.
Glad I did not plan on going.
posted by Danf at 11:26 AM on July 2, 2012
It was going to be free admin, with only the West Grandstand open.
Glad I did not plan on going.
posted by Danf at 11:26 AM on July 2, 2012
Does anyone really think that 1/3000 of a second, or even 1/100 of a second, is meaningful in a single measurement of human speed? It's like one part in 30000, or .003%.
For the sake of accuracy: It's like one part in 3000, not 30,000. 1/3000=.000333... or .033%
Take it a little further: running 100 meters in 11 seconds is 9.1 meters per second. Because it starts from a standstill, at the finish they are probably going 10 meters per second or better. So the 3000 frame-per-second camera is capturing intervals that are 10/3000 meters apart, which is 3.33 millimeters (about 1/8 inch for the metrically-challenged).
posted by beagle at 12:27 PM on July 2, 2012
For the sake of accuracy: It's like one part in 3000, not 30,000. 1/3000=.000333... or .033%
Take it a little further: running 100 meters in 11 seconds is 9.1 meters per second. Because it starts from a standstill, at the finish they are probably going 10 meters per second or better. So the 3000 frame-per-second camera is capturing intervals that are 10/3000 meters apart, which is 3.33 millimeters (about 1/8 inch for the metrically-challenged).
posted by beagle at 12:27 PM on July 2, 2012
She was scared of losing, which is likely to have been the outcome. But in racing, you never know!
I'm super bummed she quit.
posted by ReeMonster at 1:09 PM on July 2, 2012
I'm super bummed she quit.
posted by ReeMonster at 1:09 PM on July 2, 2012
beagle: "For the sake of accuracy: It's like one part in 3000, not 30,000. 1/3000=.000333... or .033%
Take it a little further: running 100 meters in 11 seconds is 9.1 meters per second. Because it starts from a standstill, at the finish they are probably going 10 meters per second or better. So the 3000 frame-per-second camera is capturing intervals that are 10/3000 meters apart, which is 3.33 millimeters (about 1/8 inch for the metrically-challenged)."
3.33 millimeters out of 100 meters is 3.33/100,000 or approx 1/30,000. So one racer was less than 1/30,000th faster than the other. QED
posted by m@f at 2:56 PM on July 2, 2012
Take it a little further: running 100 meters in 11 seconds is 9.1 meters per second. Because it starts from a standstill, at the finish they are probably going 10 meters per second or better. So the 3000 frame-per-second camera is capturing intervals that are 10/3000 meters apart, which is 3.33 millimeters (about 1/8 inch for the metrically-challenged)."
3.33 millimeters out of 100 meters is 3.33/100,000 or approx 1/30,000. So one racer was less than 1/30,000th faster than the other. QED
posted by m@f at 2:56 PM on July 2, 2012
Across the distance, if one runner were 3.3 millimeters faster than the other, that's 3.3/100,000 or 1/30,303 or .000033 or .0033%. So, OK. it depends on what we're comparing with what.
posted by beagle at 3:18 PM on July 2, 2012
posted by beagle at 3:18 PM on July 2, 2012
For the sake of accuracy: It's like one part in 3000, not 30,000. 1/3000=.000333... or .033%
1/3000 of a second out of a total of 10 seconds is 1 part in 30000.
posted by DU at 4:35 AM on July 3, 2012
1/3000 of a second out of a total of 10 seconds is 1 part in 30000.
posted by DU at 4:35 AM on July 3, 2012
« Older La Création du Monde | Comics Pride Newer »
This thread has been archived and is closed to new comments
posted by troika at 8:43 AM on July 2, 2012 [1 favorite]