The CardBoard
Touchpad touch cost Katie Ledecky outright win - Printable Version

+- The CardBoard (https://thecardboard.org/board)
+-- Forum: C-House! (https://thecardboard.org/board/forum-4.html)
+--- Forum: The CardBoard (https://thecardboard.org/board/forum-5.html)
+--- Thread: Touchpad touch cost Katie Ledecky outright win (/thread-13638.html)


Touchpad touch cost Katie Ledecky outright win - M T - 03-29-2017

ESPN finally showed a 2 hour portion of the NCAA Women's Swimming & Diving Championships.

Rowdy Gaines, in his commentary seconds after the 200 Free race, as he is seeing the replay for the first time:  "Katie Ledecky.  I thought she had it [the win] from this vantage point.  I thought it was all hers."  But the times came up the same:  1:40.36, which is considered a tie.

I've seen many instances where I felt that swimming touchpads do not accurately record finishes.  I believe that the touchpad in Katie Ledecky's lane was slow to record her touch.  The video of the camera directly above the finish seems to indicate that Katie touched the wall more than 0.01 seconds ahead.  The apparent distance that Mallory is from the wall when Katie touches agrees with this.

In swimming, by definition, two swimmers who record the same time to the 0.01 second are declared to have tied.
Because of the discrete nature of the clock, two swimmers can be arbitrarily close in actual finish (say, 0.000001 seconds) and yet receive different times.  Likewise, a swimmer whose arbitrarily precise finish is 0.010000... or more seconds faster than another swimmer should get a faster time by 0.01 or more seconds.  A swimmer that finishes 0.01 - 0.000001 seconds behind may receive the same time.  A swimmer that finished 0.005 seconds behind has a 50% chance of receiving a slower time and a 50% chance of receiving the same time.  In theory.

At swimming speeds, what distance is covered in 0.01 seconds?

Swimming a 200 yard race in 1:40 is almost exactly 2 yards/second = 72 inches/second = 0.72 inches/ 0.01 seconds.  The actual swimming speed is a little slower (starts & turns are faster than swimming), so let's just look at the last 2 lengths. The timer catches when her feet hit the wall with 50 yards to go, and the race ends  with the touchpads register her hand touch.  Katie Ledecky swam the last length (50 yards minus (her body length+arm length), so about 47.33 yards) in 25.62 seconds (Comerford swam it in 25.53).  That averages out to be 66.5 inches per second, or 0.665 inches/0.01 seconds.  One inch is about from the tip of your index finger to the middle of the first knuckle.  In 0.0167 seconds, the swimmer travels 1.11 inches.

If you want to correlate the distance between the position of the fingertips versus the time between touching, you also have to consider if the arms are moving faster than the average for the swimmer.  They certainly are moving faster at times during a stroke, but when you've stretched out to touch the wall, the fingertips will be traveling at the speed of the body.

So, between Katie Ledecky and Mallory Comerford, the two swimmers should be within 2/3 of an inch of each other for them to possibly correctly get the same time.  If they're more than 1/3 of an inch apart, there is a more than 50% chance that they will get different times.

Attached are consecutive frames of a camera that looked almost perfectly straight down over the finish (you can see the front of the touchpad, so the actual touch is not hidden).   

To calculate the frames (images) per second, I took a recorded basketball game, also broadcast by ESPN and counted the frames in one second as an on-screen clock counted down the tenths of a second at the end of a quarter.  There were 60 distinct images in one second.  When I watch the real time feed of the race (recorded on a TiVo), there are 8 frames (distinct images) from the splash of Katie's last arm entry to the splash of Mallory's last arm entry.  When ESPN steps through the consecutive frames of the overhead view, there are also 8 frames.  So, the overhead frames are effectively snapshots every 0.0167 seconds.

In frame 1, Katie is about to touch.  Her fingers appear to be about 0.5-1” from the wall, but it is hard to be sure. [I don’t really trust my estimated distances here.  Use your own judgement.]

In frame 2, 0.0167 seconds later, Katie has clearly already touched.  It appears that the 3 middle fingers have touched but the little finger and thumb have not yet touched.  Jam your hand, fingers first, into a wall and see what happens.  As the momentum of the body pushes the main part of the hand toward the wall, the hand will travel downward causing the fingers to  bend, as though they are bending upward (but they really don't move upward - Sorry, Rowdy). Mallory's fingers appear to be 1.5-2” from the wall.

In frame 3, 0.0167 seconds after the image that shows Katie touched, Mallory has not yet touched as there is still blue between her hand and the wall. Her fingertips appear to be about 0.5“ from the wall.

In frame 4, Mallory has finally touched.  Note that in the four frames, Mallory's arm does not appreciably extend further.  Therefore the estimate that her fingers are traveling at the same rate as her average swimming rate is reasonably accurate.

Because we can see that Katie has touched in frame 2 and that Mallory has not touched 0.0167 seconds later in frame 3, Katie's time to the hundredth of a second should have been at least 0.01 (likely 0.02) seconds faster than Mallory's.  Assuming a sane design of the touchpad's timing, the only explanation I can come up with is that the touchpad did not register Katie's touch when it should have.

We see the main part of Katie's hand drive all the way to the wall.  She didn't jerk her hand back.  She didn't swipe at the wall.  I am very confident that her fingertips exceeded the rated touch sensitivity of the touchpad.    In my opinion, the current touchpad technology does not accurately record swimmer's touches to the hundredth of a second.

In this case, in my opinion, the touchpad cost Katie the outright win.  Rowdy, you were right in your first assessment of the outcome.

(The attached images are unfortunately not screen caps.  They are photos of a TV screen, and then cropped & reduced in quality to meet the size requirements for attachments.)

[Edit: I initially thought the frames were 0.033 seconds apart but I realized they were 0.0167 seconds apart.  This doesn’t change the argument that the touch didn’t register, and that Katie’s time should have been 0.01 or more faster than Mallory’s, but it does halve the time difference (from >0.033 to >0.0167).  It also made me realize I was slightly overestimating the distance to the wall.]


Re: Touchpad cost Katie Ledecky outright win - 81alum - 03-29-2017

I know nothing about swimming, but that is fascinating analysis, MT, and seems like quite the indictment of electronic timing of swimming.

For those of you who have been competitive swimmers, how have you been instructed with regard to these timers?  Do they have to be touched with a certain technique?

Re: Touchpad touch cost Katie Ledecky outright win - M T - 03-29-2017

81alum, I made a significant edit.  You may want to re-read it.  The conclusion doesn't change, but magnitudes do.

Re: Touchpad touch cost Katie Ledecky outright win - fullmetal - 03-29-2017

As a middle school swimmer, I often had issues hitting the touchpad hard enough to register a time, but I was also much lighter.  I would guess that the amount of force required to get each touchpad to respond might vary a bit?  I don't know their operational priniciples.  Or maybe you need a touch over a sufficient surface area...or something.  I sympathize!

Re: Touchpad touch cost Katie Ledecky outright win - CTcard - 03-29-2017

A very similar issue arose at the Olmpics in 2012 where it appeared that Phelps finished second in the 100 fly but ended up winning in a brobdingnagian surprise that many thought was bogus. Here is an article on how timing and photo finishes are used at the Olympics. 

http://olympics.time.com/2012/07/27/technologys-touch-how-a-photo-finish-in-the-olympic-pool-gets-resolved/

Unfortunately, there are some ambiguities in the article (beyond the fact that it calls kg a unit of force).
We have both,
Quote:That’s how Phelps’ win in the 100 butterfly in Beijing was determined. The camera’s shutter clicks 100 images per second, and by comparing Phelps at the wall to Cavic, officials could see that Phelps finished first — by one image.

But also,
Quote:Fingers have to register at least 1.5 kg of force to be a legitimate touch;...
Which means that in the Aquatics Center in London, coming to the wall doesn’t just mean putting your hand on the touch pad first. It means pushing it — by 2 mm — as well

That seems to leave open whether the photo or the touch pad are the official determinant of the winner.

[I had always thought it was touch pad, but from that particular Phelps finish there was lots of talk about the photos being the ultimate arbiter. Somewhere I suppose the official rulebooks of FINA and the NCAA define this, but I don't have time to look just now.]


-------------------------
EDIT

So I looked up what is available on the web. FINA's rulebook is so short I cannot quite believe it is the actual rulebook, and doesn't really specify anything beyond automatic timing be used for official time and placements.
Here is the section on timing:
http://www.fina.org/content/sw-11-timing

USA Swimming has a more complete rulebook available.
https://www.usaswimming.org/docs/default-source/rules-regulations/2017-rulebook.pdf?sfvrsn=13
(see pages 39-40)
It is pretty clear that the primary system for determining times and placement is the touchpads (unless you don't have them - clearly not the case at an NCAA championships), and the overhead cameras are a secondary (backup) system. However, they maintain confusion by stating,
Quote: If  manually  operated  watches  are  not  the  primary  system,  a  secondary  system  of  precedence  equal  to  or  lower  than  the  primary  system  must  be  used.
I don't know how the secondary system can have equal precedence - if they disagree you have to use one or the other. Still, I am pretty sure that means you go by the touchpads are primary and the overhead camera is only if you have reason to believe the touchpads failed.


The NCAA itself has a rulebook that is ... pretty hard to read. As far as I can tell, all the words about video assume that it is only used as a backup to the automatic relay exchange judging system. The rules appear to make no mention of video for determining finishes or timing, and only notes touchpad systems as automatic devices for this purpose. Yet, they obviously had a video system in place, given the still frames above.
http://www.ncaapublications.com/productdownloads/SW17.pdf

Re: Touchpad touch cost Katie Ledecky outright win - oldalum - 03-29-2017

thanks, very interesting analysis. The touchpad has a mechanical component to it, so I would suspect it would not be accurate to 1/100 of a second. Seems like photos are a better determinant.

But, from a broader perspective, I'm not upset about the two of them sharing the gold in a race over 200 meters that ended that close. I don't view Simone's 100-m Olympic gold as being diminished by her tie, either. I understand that fewer points toward the team total might make a difference in some NCAA meet, but not in this one.

Re: Touchpad touch cost Katie Ledecky outright win - DC 86 - 03-29-2017

There is another issue involved as well - the fact that the pool lanes are not necessarily identical distances. In fact, FINA permits tolerances of up to 3 centimeters between lanes:

http://www.fina.org/content/fr-2-swimming-pools

It's basically impossible for the pool wall to literally be perfectly straight without imperfections and variations, which is why times are only measured to 1/100th of a second even though technology permits us to be much more precise.



Re: Touchpad touch cost Katie Ledecky outright win - M T - 03-29-2017

(03-29-2017, 04:41 PM)DC 86 link Wrote:There is another issue involved as well - the fact that the pool lanes are not necessarily identical distances. In fact, FINA permits tolerances of up to 3 centimeters between lanes:

http://www.fina.org/content/fr-2-swimming-pools
You misunderstand.  A pool that measures 50.03 meters will measure the right length with touchpads installed:
Quote:For a pool to be included in the list [of certified pools] below,
all lanes must meet the “minimum” lane length requirement of 25.03 yards/meters or 50.03 meters
measured without touchpads.  The pools can be measured with touchpads installed
which makes the minimum length requirements 25.00 yards or meters and 50.00 meters. 
(NOTE: I believe they meant 25yards/meters + 0.03 meters.)

Pools are surveyed to much more accuracy, usually by laser. Each lane is measured separately (in 2 locations). You will not find a pair of lanes that differ by anywhere near that much.  For USA Swimming, gory details can be found here  Lane lengths are reported to at least an accuracy of 3mm (short course) or 5mm (long course).  The only actual certification survey form that I found online contained these lengths:
25.0261, 25.0307, 25.0416, 25.0391, 25.0412, 25.0409, 25.0374, 25.0433.  For this pool, the maximal variation among lanes was 0.0172 yards = 0.6".

For a 50 yard race, that 0.6" would represent maybe 0.012 seconds and is unlikely, but possible, to impact there results.  But for a 1650 (66 lengths), that adds up to 39.6", which is roughly 0.6 seconds.  For a 200, that particular difference adds up to 4.8", or roughly 0.075 seconds.

(03-29-2017, 04:41 PM)DC 86 link Wrote:It's basically impossible for the pool wall to literally be perfectly straight without imperfections and variations, which is why times are only measured to 1/100th of a second even though technology permits us to be much more precise.
I disagree.  I'm not an engineer, but I expect that the placement of touchpads could be adjusted to 2mm accuracy across a pool width, with individual tuning and measurement by a laser of the other end.
My recollection of the early 60s (when there were some questionable results from human timing & judging, leading to the development of electronic timing) is that times could be measured to the 10,000th of a second, but the decision was made to limit it more or less to human perception limits.



Re: Touchpad touch cost Katie Ledecky outright win - Langdude - 03-29-2017

This is awesome. Please find me a sports board from any other school that is having this conversation (or one of the many others). I learn a lot reading here! Stanford is great.

-m.

Re: Touchpad touch cost Katie Ledecky outright win - JJJ - 03-30-2017

I also think this is awesome, but has it been shared anywhere else? If not, why not? Let's get more mileage out of all the effort put into it already! (Except for the fact that I have no clue where it ought to be shared...)

Re: Touchpad touch cost Katie Ledecky outright win - fullmetal - 03-30-2017

I think that the coaches would have brought some kind of protest if possible at the time...it's likely that results this far after the meet are not subject to revision outside of a WADA-style violation.  I know nothing about NCAA Swimming rules!