Ties for medals at Sochi 2014

Think a Tie for DH Gold will finally push FIS to time to .001.
Someone knows who actually went faster. Sure clocks were recording to the ten thousands.
Truncate don't even use the data?!?!
OMG!
NBC must be stunned!

This (obviously) isn't a technological question, it's a philosophical question. Anybody with the skills necessary to time a ski race to 0.01 would most certainly be capable of timing one to 0.001 or 0.0001, if the greater resolution was what the rules called for.

Here's the reason.

The FIS Timing Working Group has been trying to get rid of mechanical start gates since the 80's. They maintain until that happens, the random mechanical slop inherent in mechanical switches and the random flex inherent in start wands makes timing to .001 simply a random draw. And they've proved it. Repeatedly.

The FIS TWG made their decision based on data collected by my technical group at FIS World Cup and the World Alpine Chmps in the 90's (as TAG Heuer) and then revisited based on data we collected in the 00's (as Rolex). We installed several sets of cells downhill of the start gate (on the start ramp) at 1m intervals and collected data for both men and women across all the disciplines. Analysis of the data clearly showed that mechanical start gates are, to a relevant resolution, random number generators.

This is a question much like the one brought up at the summer Olympics in Munich in 1972. A few months before the Olympics, FINA announced they'd obtained the technology to time swimming to .001, and would start at The Games. FINA were subsequently contacted by the engineering firm who'd designed and supervised construction of the pool in Germany. Representatives from the firm sent FINA a mathematical proof showing that at speeds typical for Olympic swimmers, the pool wasn't built to sufficient tolerances to where all lanes were of equal length to a degree where .001 would be fair. And notice...to this day...swimming also still publishes results only to 0.01. FINA have shelved the idea of 0.001 for over 40 years because no mechanical engineer will certify a pool with walls and touchpad mounts so precisely built that 0.001 would be consistent and fair across all lanes.

Think about it from an engineering standpoint.....let's say you had a time base accurate to 0.0000001 and photocells only accurate to 1.0 seconds. Sure, you could publish results to 0.000001, but anything beyond a full second would be random and therefore useless.

Mechanical start gates are an anachronism, but the TWG has to date been unable to get rid of them. It's a tradition FIS hasn't been willing to part with. Until that happens, publishing results to resolutions beyond 0.01 simply isn't fair because it's not accurate. It's proven to be random.

On another note, at the 1999 World Alpine Chmps at Beaver Creek, where we (TAG Heuer) were official timing, there was a tie for first in the mens SG between Kjus and Maier. Naturally we had the tapes, so for fun we calculated who won without truncation. Of course we kept that tidbit of information to ourselves. Later that night, persons unknown (still unknown to this day) broke into the timing bldg at Birds of Prey and stole the tapes. The next day, the "real winner" was published in a bunch of newspapers in Europe, along with photos of the stolen tapes.

James,
Thank you for the history and explaining the issue. When I test start gates I find different trigger points between most A & B outputs based on the position of the wand not to mention the flex in the wand.
You think we will go to photo cells sometime in future???
Sorry your tapes got stolen. No, I don't want to know how you protect them now.
Nick

Nick:

Yes, start gates are a technological mess. Not to mention there is no consistency, nor any flex standard, nor any thermocompensation standard, from wand to wand. So if you were to replace a wand mid-race, which most of us have done, you could be unknowingly changing your race results significantly. Certainly enough to break or make ties.

In the 90's, TAG Heuer had some very expensive experimental carbon fiber wands manufactured for World Cup because, in theory, carbon wands would be way stronger and hopefully more consistent than the fiberglass wands we were using at the time. This particular batch of wands was built by a Formula One supplier to a very tight tolerance, so they were supposedly very consistent and came with lab test data. The carbon wands worked great until we tried them at World Cup in Lake Louise @ -37C, whereupon they shattered like icicles every 5 racers or so. Working as an arm of TAG Heuer with factory support was a lot of fun back then because the big cheeses at the time, Jean Campiche and Ted Savage, were very interested in advancing the level of engineering, so we could get budgets to design and build new widgets and try new technologies from time to time. Some of the ideas worked, some of them didn't.

Having our tapes stolen at the WASC wasn't a catastrophe, but it was certainly amusing. They're not a secret. Any athlete or coach has the right to examine race tapes and do their own math, which is one of the successes of the TWG. It may seem like a big pain in the tush for timing geeks to fill out timing forms and submit their forms & tapes to the Chief of Timing, and I've certainly heard a ton of complaints about it. But in an era where there is extensive betting on ski racing and a lot of corruption & conflict of interest in the sports headlines, transparency is important.

I have no better idea than you as to whether wands will ever be replaced with photocells. I'm not on the TWG and I'm an engineer, not a politician. It's a FIS decision, a phrase which makes us all cringe. As a practical engineering matter, it's a no-brainer. Keep an unplugged start gate on the start post for TV and start the race with a photocell mounted 1m down the hill. Duh.

-37c
hope you were inside.
Bet they wrecked every pair of skis used that day.

Thanks for the detailed response. Although the technical reasons are clear, the "FIS decision" part is anything but... just as I'd suspected all along.

I think the starting gate is just one of the issues.
Photocells (many models are based on electromechanical switches) likely suffer similar problems. It would be nice to know if someone did the same tests (described by the previous users) at the finish.
I suspect that at some degree, even photocells deliver results which are almost random.
Maybe not at 0.01 but probably 0.001 would be close to the physical limit of most electromechanical switches.

Then what about the clocks? The FIS rules only require the initial homologation of the device, but there is no periodic check of its stability, calibration etc. as it happens for other sports.

So.... I guess that the TWG is stuck between technical constraints, rather than political.
Also because the sport authority must evaluate the sustainability of the changes.

FIS TWG should first try to address more important issues.
An example: by the diagrams, ski is probably the only sport which has different ways to do timekeeping, as the level of the race increases.

Which is a little bit insane, my humble opinion.
It means that the same skier, may have a different time -with the same equipment- if the level of the competition changes. Don't you think that this is all but normal?

Sorry Aaron, either you have not read the FIS homologation rules regarding photocells or you do not understand the specifications. You are demonstrably wrong.

Electromechanical photocells are not even eligible for submission to the FIS- and FEI-approved homologation lists. Submitted designs must be fully electronic, with a minimum sampling rate of >= 200,000 hertz and a max warble of < 0.5 ppm. That resolution is ((TOD resolution X 2 orders of magnitude)x 2). Such photocells are, in and of themselves, very accurate, thermocompensated time bases.

There are of course plenty of electromechanical photocells on the market. Such products are inexpensive and are perfectly adequate for, say, triggering a door chime in a sandwich shop. But for high-resolution sports timing....not so much.

It is for this reason that sporting disciplines with no timing equipment homologation process nor timing procedural specifications are doing their athletes a disservice. Those who don't understand the specs or the certification process are left scratching their heads when they see some of the nicely-packaged integrated timing systems available on the market for a few hundred dollars. They then look at approved hardware at 10x the price and can't understand why.

This is why.

TheMightySkunk, why do you think I am wrong?

I kindly ask you:
-Do all the photocells in the FIS list of currently homologated equipment comply with the specifications you wrote?
-Do current (fall 2013) FIS requirements for photocells guarantee that we can safely go to 0.0001?

I think that right now -I mean with today’s homologated equipment- even 0.001 might be risky due to the photocells, not only due to the starting gate.

And I state again, I feel that FIS TWG’s efforts should be directed to fix things like the differences of the timing between level 0 competitions and level 1, then to fix some evident contradictions in the rules and so on.
Although I may guess the reasons behind this, I feel that the diagrams for level 0 competitions are deeply wrong if not insane, from many points of view.

I will probably never be the timekeeper of a level 0 competition so this will not change my life.
But since the original question was about the reasons behind the ties in the olympics, our opinions about the FIS diagrams are legitimate, I hope.

Answers:

1) I did not write the specs. But I personally know, and work with, the guys who did. I am a software engineer, not an EE. I was involved in the process because I was asked to write some of the data-collection software used to test the various devices and technologies.

You won't get an answer on this forum from any of those guys, none of them belong to this forum.

Yes, commercial timing hardware appears on the list only after it has been tested in a FIS-approved lab, and testing has shown it meets the spec.

2) No. FIS timing is source TOD 0.001 with truncation math to NET 0.01. That's the resolution of the spec.

If you believe that the current photocell spec is invalid for the resolution claimed, post your rebuttal and math proof here, and I will see that it gets to the right people for review.

Well, first of all thank you.

There is no need of big math, after your answers. All this thread started from sly_kharma’s question about ties.

In your first post, you stated that the starting gates were the element preventing going down to 0.001.
My impression instead was that with the current homologated equipment, the starting gate was not the ONLY constraint.
Now, I think it is clear that the photocells are another limitation, and probably not the last one. You confirmed that. Thanks. That is all what I asked to clarify.

But now I would like to take the opportunity you offered, to ask a couple of questions to the “right people” through you.

The first is about the diagrams for timing.
As far as I can understand, for competitions of level 0, there seems to be a “strange” redundancy,
as any fault in some elements will result in a quite wider failure of the timing system, since systems A and B are deeply interconnected. Which is something going in the opposite direction of the traditional interpretation of redundancy.

Please correct me if I am wrong: I am referring to the Timing Booklet version 2013, however diagrams did not change since many years.

Let’s say the starting gate has an issue, like a permanent short on one channel. This will block the start channel of both timers. TOD will be available, but only on a 3rd device.
Yes, there might be electronics detecting permanent shorts, but basically, any issue on any channel, of the main and backup timer, will be reflected on both timers.

And there is an even more tricky/subtle situation. What about the TDR in level 0 competitions?
It might clearly happen that 2 different skiers are measured with different parts of the two systems.
Since I assume that the same rules about times coming from systems A and B are valid for ALL competitions, what happens on Level 0 competitions?
Times are all certified as coming from System A even if i.e. the photocell that actually stopped the time was B?

Hi everyone.

I did an analysis recently which I think gets to the very heart of this topic of timing precisions. Last season I noticed that the difference between our SysA and SysB times were larger than I expected. So I examined the distribution of these differences for all of our races, and what I found surprised me:

1. ~50% of the run times differ by >= 0.01s (both positive and negative).

2. ~10% of the run times differ by >= 0.02s (both positive and negative).

3. Virtually all of the difference occurs at the finish, which is contrary to the study noted previously in this thread. 99% of the start TOD difference is <= 0.005s, and most start time differences are <= 0.002s.

4. The fact that the start TOD differences are so well behaved indicates that the timers are working well. They are not drifting relative to each other.

5. The finish TOD differences between the SysA and SysB vary randomly. The difference can go from +0.03s to -0.02s for consecutive racers.

Naturally, I was concerned that our club's results were atypical. I wanted to see another hill's results, and so I analyzed the race files for the March 2012 Men's Eastern Cup GS at Burke which I had from the TC1 test prep. Low and behold, the results were virtually identical: 55% of the A/B time difference were >= 0.01s, and 8% were >= 0.02s.

I spoke with David Iverson (BMA) about this, and he thought that the results are entirely reasonable. Like me, David suspected the cause of the difference was due to different parts of the body breaking the two beams at different times. Fred Patton has actually proven this with a high-speed photofinish study done at the 1999 Worlds at Vail.

So, if two completely legal timing systems give times which differ by more than +/- 0.01s 50% of the time, I would conclude that our timing precision is worse than 0.01s – maybe 0.02s or 0.03s, depending on how much variation you are willing to accept.

The problem, I believe, is the rather nebulous definition of what constitutes a finish:

[ACR 611.3.1] With electronic timing, the time is taken when the competitor crosses the finish line and triggers the beam between the photo cells.

What part of the competitor? The tip of the skis? The shins? A ski pole? And even if the ACR did more precisely define what constitutes a finish (for example, the shins breaking the PLANE between the finish posts), we have no way of reliably capturing that with our finish eyes. The only way to do that is with a photofinish system, which is prohibitively expensive.

The timing systems that we have are not bad. My only point is that there is really no sense in expending effort on improving the precision of the start gate or the timers before we find a way to more reliably and repeatably define and capture the finish.

You can view the analysis details for one of my typical races as well as for the Eastern Cup GS at the following link (I hope that this works ...):

https://onedrive.live.com/redi...BF9D2FF1FF3A5A%21140

jacunskim, you are absolutely right.

I tried to highlight this situation, but so far, I found nobody willing or knowing to admit that the issue exists, is serious, and evidence shows that it is not related (maybe only in part) to the starting gate.

I remember we had a couple of TdC4000 in the 80s, the machines were drifting between themselves no more than 5 ms every 2 hours. This was checked on all channels. Such devices were used mostly in ski competitions.

On any ski race, the start gate impulses followed this drift on 99% of the times.
But the net times measured were, as you said, randomly different. And this was not only due to the then available photocells. Although we did not have any hi-res camera and/or photo finish, the reason was clearly suspected to be in the way skiers crossed the finish line: hands, arms, etc..

30 years have passed, and I see that this has not changed a lot. Now the reduced size of some photocells allows us to put them as near as possible and thus, the chances that
the two beams are cut at the same time are higher.
There are sometimes some crosstalk problems, but this another issue.
As expected, sometimes we get sometimes big differences (+/-0.02), that not surprisingly match exactly the length of half an arm.

Let’s say one skier is crossing one beam with the hand and the other with the legs.
This means: @ 50 km/h it takes 29ms to cover 40cm, 36ms to cover 50cm.
At 100 km/h, it takes 15 ms to cover 40 cm, 18 ms to cover 50. I have no data about speed events (DH/SG), but the tapes of giant slaloms show a perfect match with these values.

My sensation is that in level 0 competitions, this odd situation has been "fixed” in a quite rough way. Instead of working on the quality of the sampling, which would be the first recommendation, the fix has been to put a logical “or” gate between the two photocells and the two clocks A and B. Shortly, the 1st photocell giving the pulse will stop the time.
I guess that in L0 competitions, the A and B systems show totally identical TODs between systems A and B, maybe only the drift between the two clocks appears.

That was -I think- a too easy solution. Because it had some important side effects:
a) one channel “hides” the other (at least for A and B) which is quite an issue for the redundancy that the 2 systems are supposed to insure (and thus 3rd and 4th clocks have been added).
b) there is a more subtle problem. Doing so, you are actually “timing” differently the skiers. For some of them, the time will stop with system A, for others with system B.
This clashes hardly with the basic principle of sport, that all competitors must be judged in the same way.
c) there seems to be no “special" TdR for level 0 competitions, so when the chief of timing fills the forms, how can he state if the times were from system A or B, since one system’s impulses always hide there other’s?
d) The situation is so unbelievable, that with the same equipment, in the same ski race, if you set “level 0 mode” you can have results quite different from “level 1 mode”. But you are timing the same race with the same skiers with the same equipment.

I personally find this quite disturbing and I don’t know how much fair it is. As I stated before, I am not timekeeper of L0 competitions, so it will not change my life.
But when I see ties on the results of big competitions, or podiums with three skiers that fit within 2/100ths of a second, I think: in 2014, are those rules still appropriate and fair?
Today 1/100 means big money, olympic gold, etc. etc.. The ties in the OG are true ties or just the output of something that still has a wide margin of uncertainty inside?