I have no doubt that many HN readers will wonder why I've posted this. It's not specific to hackers, I know. It's only tangentially related to technology, and it certainly doesn't seem to have anything to do with startups.
But there are questions to ask, and lessons to be learned for startups, and for entrepreneurs in general. Here are some of the questions raised for me. Perhaps you can think of more:
What processes do you do by hand that could be automated? The timing was done by hand - a notoriously error-prone process. relentless automation can systematically weed out systematic errors, provided they are properly monitored and tested. Then you can forget them, and let them do their job.
Do you have the right people doing the right jobs? The report is that the time-keeper was an inappropriately chosen volunteer. It's notoriously easy to leave a job in the hands of someone who may not get it right when it's important. And just when isn't it important? (actually, sometimes)
Do you have the right error reporting processes in place? The team lodged an appeal, and clearly that didn't go anywhere. Why not? The entire bout could surely have been replayed and the timing checked. The result, in truth, should never have been in doubt, and could easily have been overturned.
Do you have the right corrective measures in place? After reviewing the bout, surely it should be straight-forward to correct the result. When you have a mistake reported, can you roll-back and make it right?
And finally (for me - perhaps you have more points to make), always, always consider the individual. This poor woman was left distraught on the piste, unable to leave, in full view of the audience, in tears, her lifetime's work hanging in the balance, out of her control.
So let me ask - how well do you look after your users?
Personally I'm glad this made HN simply because it is an interesting story and I virtually never check the mainstream media. To me, the failures in the mainstream media, including focusing on kitch or faddish things that are of no particular interest to me, are always informative in the sense that I think they can help people building the next generation of news/media platforms -- empowering users to get the information that is relevant and valuable to them.
In this sense, I think this is extremely relevant, simply because I think we all know that the old newspaper model is dead, but nothing obvious has risen to replace it (besides news aggregators, either as semi-free or targeted Bloomberg-esque services).
This is, also, fairly obviously, one of those problems that has the problem of scale immediately rather than post facto, so it is interesting to explore those aspects of the problem -- insofar as there are a lot of places already reaching out to a niche audience, but nothing that obviously brings them together.
Thanks for submitting this story, this is a great reflective comment. The one point that I'd add is that, going into the matches, no-one felt time-keeping could be a big thing that would break, or be significant. It's always (maybe tautologically) the things we think that are too small to notice that trip us up the most.
I read somewhere that the team tried to lodge an appeal but didn't front the $500USD required for the appeal to be considered. I found it extremely odd that at a place like the Olympics, you'd be asked to front money for an appeal.
On another note, I wonder when computerized judges will start to replace humans for stuff like synchronized diving. All of those post-dive slow-mo footage I've seen seems to suggest that computers would be much more accurate/fair/subjective than the human eyes that are currently being used for judging.
I read somewhere that the team coach actually had to run out and find an ATM to withdraw money to lodge the appeal, while the athlete was sitting on the mat.
I guess a minor fee is OK to prevent troublemakers appealing every single ridiculous thing.
I believe just the starting of the timing is done by hand. The stopping is done when a contact is automatically detected.
I am on the "fence" about this article. It seems within the capabilities of a 15 year old to press a button when a referee shouts a word. The responsibility is great when the controversy hits, however.
> I believe just the starting of the timing is done by hand. The stopping is done when a contact is automatically detected.
That's exactly the issue, though - there are a handful of mechanisms in place to ensure that the clock stops and points are awarded properly - electrical equipment is used, buttons on the tips of the Epees are tested with a 750g weight, fencers test on eachother's masks before each match, testing kits are used to check wiring, etc., and the judges are there to watch movements to account for possible disputes. The task of starting the clock has been significantly overlooked compared to the task of stopping it.
Aside from this, the rules and the scheduling of matches are done with the assumption that there will be no major disputes, which needlessly compounds the damage from any dispute that happens.
I imagine the complexity of the system above is not primarily about stopping the clock, but to accurately detect whether a hit was placed.
To have this level of precision I imagine the judge's "allez" would have to be replaced with a racetrack style countdown and green. Otherwise attempting to detect when a specific spoken word has been said (either predicting it just started, or detecting that it has finished) sounds like a complex problem.
Fencer here. There are several questions on HN, so I'll explain the rules that led to this.
A normal bout goes for nine minutes or until someone scores 15 touches. In this case, time expired with a score of 5-5.
In the event of a tie, one fencer is randomly selected to have "priority" and the bout resumes in sudden death for one minute. Whoever scores in that minute wins. However, if time expires again, then the fencer with priority is declared the winner by default.
The reason for priority is to force one competitor to be the attacker. Epee fencing in particular favors defense, which is why the score is routinely so low. The priority rule is a way of preventing passivity in overtime.
For this bout, the Korean fencer was given priority. The clock ran down to one second and the two fencers "doubled", meaning they hit simultaneously. In overtime, a double touch is annulled; nobody scores. The clock had one second remaining, so the bout resumed. The fencers doubled again.
But oddly enough the clock still had one second listed! So the fencers had to try it again. They immediately doubled. At this point, it looked like the clock had run to zero, and it really isn't clear what was actually happening with the time keeper. Eventually, the referee reset the clock back to one second!
Finally, the German fencer scored a single touch and was declared the victor. The appeal from the Korean team is that the clock should never have been reset to one second; the numerous double touches beforehand had been more than a second long.
The committee that overseas the referee (the Directoire Technique) refused to overrule. From personal experience, the DT never overrules.
At any rate, the entire fencing community is really upset about this. Many of us were routing for Korea today in the team women's epee event because of this.
"En cas de défaillance du chronomètre ou de faute du chronométreur,
l'arbitre devra évaluer lui- même le temps restant à tirer."
In English:
"In case of failure of the chronometer or of error by the chronometer,
the arbiter can, all by himself, decide the time remaining."
So, I can see a reason why the arbiter was not overruled, especially since he, apparently (http://news.ycombinator.com/item?id=4339668), informed both parties that he adjusted the clock. If they had changed the result, surely the other party would appeal to that.
> Unfortunately for Shin, the clock never started after the referee signaled to restart the match, giving Heidemann more than one second to land the winning touch.
> the timekeeper for the event was a 15 year old British volunteer
Wow, this is just absurd. Does anyone know on what basis the decision not to award the Korean fencer the victory was made? This isn't even a situation in which a do-over would have been necessary - the correct result was obvious.
And I can't help but feel that if this had happened to an American fencer, the decision would have been very different.
> Unfortunately for Shin, the clock never started after the referee signaled to restart the match, giving Heidemann more than one second to land the winning touch.
If I recall correctly, this is wrong. If there is less than a whole second left for this 're-start' of the match, the clock is reset to a whole second. In the German broadcast of the event, viewers were told that Heidemann landed her hit 0.8 seconds after the clock was officially started again, so it counted.
Now what's left to discuss is whether it's possible that the clock was started those 200ms late by whoever was keeping the time. However it's probably not regulated down to the split-second at which point or at which hand or mouth movement by the referee the clock is to be started again.
The real technical question concerning the judgement made is:
Just how late was the restart of the clock.
If it was more than 200ms then the hit should not be allowed to stand. Was it more than 200ms late in being restarted? The video should make that clear - there should be sufficient logging to know for sure.
But the question shouldn't arise - the processes involved should not allow the clock to be restarted at a time other than when the official says "Allez"
Secondly, the contestant should never be treated like that
Fencing used to be scored entirely manually. This was a disaster (see anecdotes below) and the problems were solved with slightly-expensive but relatively simple electronic touch-detection technology, i.e. those crazy lights that you see flashing green, red, and white.
I predict this will happen again. Perhaps the referee will get a little box that, when a button is pressed, simultaneously shouts "Allez" in an authentic French accent and starts the clock. If the clock doesn't start, the referee will stop the bout, debug the technology, and reset the clock. (Welcome to fencing, where one's kit of electronics parts is almost as important as one's weapon. ;)
Now for anecdotes:
I fenced saber (terribly) back when saber wasn't electrified at the NCAA level - which is to say: there was no electric circuit to detect contact between Fencer A's saber and Fencer B's target area.
(Saber got electric scoring last, I suspect, because the head and arms are part of the target area and you need conductive gloves and masks, which cost money, particularly because it's harder to share gloves and masks between fencers than it is to share a simple mesh vest.)
Moreover, at a regular-season NCAA fencing match, there's no budget for five impartial judges per bout, and no spare unaffilated fencers sitting around. So the rule was that we had to make touch/no touch calls in matches involving our own teammates. This was, to say the least, a no-win situation for a beginner. I quickly learned to pretend I was half-blind and clueless – which, alas, did not require much acting talent – and round off all non-obvious calls in favor of my teammate. To do otherwise was to earn the undying enmity of one's own team.
The stress drove me crazy. In retrospect, I should have demanded to be switched to foil or epee – epee, in particular, with its electronic scoring and lack of right-of-way rules, was a gloriously straightforward sport; though I would still have been a lousy fencer at least I would have lost for clear reasons – but instead I fled after one season.
If there's a moral to this story, it's this: There is no way that this incident is the most unfair thing that has happened in the history of Olympic fencing. ;) They must have been really, really good sports back in the old days.
Do you enjoy watching those matches? Looks like an enjoyable sport to practice, but as an layman watching on TV, it gets very boring, because I can't see when someone touched, when someone defended, etc.
At least if the swords where painted in bright colors...
Frankly, back in the old days when Olympic fencing never turned up on TV, I used to wish I could watch it. But then it appeared, on streaming video. And I've tried to watch it every four years. And I always turn it off after a couple of minutes, because it is the most boring spectator sport I can think of.
Even bowling, which is far emptier of visible strategy than fencing, is perhaps more fun to watch, because it makes pleasant sounds and has a soothing rhythm and you can immediately see the difference between a good frame and a bad frame. And I actually spent more time watching archery last week, which is really boring as a spectator, but again it has a hypnotic rhythm, and at least you can see where the arrows land.
In fencing, most of the time the participants are hovering looking for openings and baiting each other, and then something happens that you barely see, and half the time that's a no-action anyway because they didn't hit the target, or they moved past each other, or the equipment broke.
I see, from our more-informed commenters, that this controversial match seems to have involved a tie, three tiebreakers, two dubious judgement calls and a denouement that happened too fast to see and that may have been a glitch. I wish this were atypical of my experience with fencing, but that pretty much sums up what it all looks like to me. ;)
So, not surprisingly, it turns out that a lot of sports are more fun to do, or read about, than to watch.
Maybe it would be different if I really understood fencing, rather than knowing some but not all of the rules.
I think on-field in the context of the article meant live vs tape delayed or written about afterward. (It was immediately after complaining about NBC not covering it for Americans).
That's also not strictly speaking true. The match was even at this point. Lam had previously been selected by /lottery/ to win in the case of an even standing at the end of the one-minute-long extension to the match. So, she would have won had that second passed, but she was not leading.
This is your second comment that has details I've been unable to find elsewhere. I'm interested in trying to track down what actually happened here, and your sources seem better than mine, unless I'm just missing something.
> Csar ging zur Planche, befragte beide Fechterinnen, ob sie damit einverstanden seien, die Uhr nochmals auf 00:01 zurückzusetzen und das Gefecht ein letztes Mal freizugeben. Heidemann und Shin A Lam stimmten zu. Csar gab das Startsignal. Heidemann attackierte - und traf nach 0,84 Sekunden. Die Uhr zeigte 00:00.
Rough translation:
Csar walked up to the planche and asked both fencers whether they would agree to reset the clock to 00:01 and re-start the match one last time. Heidemann and Shin A Lam agreed. Csar gave the start signal. Heidemann attacked - and hit after 0.84 seconds. The clock showed 00:00.
If you search for "lam 0,8 sekunden" on the German Google News, you can find similar articles. However, they usually are an exact copy of the article mentioned above. The real source of the above claim is probably the "Sport-Informations-Dienst (sid)", a German sports news agency that provides content to those publications. I haven't yet found any other primary sources with the exact time.
That's brilliant - thanks. It doesn't seem conclusive, because if the clock didn't get started, and the touch happened at 0.84, then the question is whether the delay was more than 0.16s, and further, why did the clock on the video still show 0:01, while the claim there is that the clock showed 00:00.
And if the clock showed 00:00, wasn't the touch out of time?
Completely unclear, but it's useful to have the extra information. Thanks again.
No problem. My guess is that they round the displayed clock up instead of down, so 00:00 automatically means the end of the match, and any split second left shows 00:01 to let everyone know that it's not over yet.
Given that if the scores remained as they were then player A would be declared the winner, then in some very real sense I would say that player A was in the lead.
Semantics and interpretation.
To me it's not really the point. To me, the lessons are about process, problems, how you deal with them, and then treating the people with respect.
Clearly this item has been heavily flagged. There are currently 20 items above it on the front page, and of them, 10 are older and have fewer points. The ranking algorithm says those items should rank lower. The fact that they don't indicates that this item has been flagged.
It would seem there are those on HN who consider this inappropriate. Enough that they don't simply ignore it, but that they take the effort to flag it (although one click is hardly much of an effort)
But the lessons to learn from this are there to be seen by those who would take the time. That's why I wrote my extensive commentary, and that's why a few people also commented in support.
Perhaps I should have written this up as a blog post so that I could have:
* controlled the title (since the moderators are now so aggressive on this)
* made the commentary the first thing to read, rather than the story.
Perhaps a lesson to take away from the story of the submission (rather than the story of the fencer) is that there is a vocal minority on HN who will flag things regardless of whether there are lessons to be learned.
I'm disappointed by that, although not really surprised.
You're counting during the extra second that was awarded. The problem was the timer was not started during the previous double-touch. The timer should've already elapsed by that time.
Why "reluctantly" if the judges were correct in the first place? Anyway, the real crime is the ridiculous random sudden-death coin flip. Whose idea was that?
I get that, if that was not the case, both contendants would stay too much on defense. OTOH if one is declared winner 'by default' in case of a tie, the other is more inclined to attack.
The criticism of NBC is tiresome and probably inaccurate. Fact is, the vast, vast majority of US tv watchers, including on both coasts: 1) prefer to watch US athletes, 2) like the biographical segments and 3) like to watch marquee events in prime time. The vocal minority are media and tech elites (and wannabes).
The criticism of NBC is tiresome and probably inaccurate.
And the reason I submitted this isn't because of the lack of coverage. The reason I submitted it is because of the many, many lessons that entrepreneurs can learn from the incident itself.
Good points (except I think premature automation is not always or even usually wise; and, see "Knight Capital Group").
I guess I was responding to the post itself (or its need/decision to highlight "American TV's" neglect. And also a good many responders taking an opportunity to bash NBC.
I used to fence a bit and watched this match live on BBC's iPlayer on Wednesday night. For those of us who're honorarily British for the Olympics, here's the link:
If you don't have access to it, the salient points of the match are:
48:25, the "sudden death" round is started due to non-combativity or timeouts on the previous rounds. Shin is given "priority" based on the toss of a coin, which means at the end of a minute if no single hits have been scored, she'll win. Shin has been playing much more defensively or tactically all match, so this maintains the tempo up to this point.
49:02, a double-touch happens with 24s left on the clock. Double-touches (ones that occur within 1/25th of a second of each other) are not scoring, so play resumes.
49:27, another double-touch with 15s on the clock. Shin plays very defensively and scores the majority of her points by evasion and counter-attack, but uses up a lot of the piste (the area the fencers are permitted in), so the referees check that she didn't overstep it.
50:09, another double-touch with 9s left on the clock.
50:24, another double-touch with 5s left on the clock. Heidemann isn't scoring a single-touch with her attack, but she's driving Shin toward the back line where she'll lose if she oversteps.
50:48, another double-touch, 4s left on the clock.
51:03, another double-touch happens with 1s on the clock.
51:18, Heidemann immediately attacks the moment the beep occurs for another double-touch with 1s left on the clock. The commentators are bemused that this takes less than a second but to my eye it's pretty reasonable.
51:20, Heidemann and Shin go en guard, with Heidemann repeatedly being warned by the referee not to take a position too close to Shin (the call of "distance" from the ref).
51:39, another double-touch, again with 1s left on the clock. The commentators are wondering about a technical malfunction at this point, but as Heinemann takes her position the clock actually clicks down to 0s. The referee notifies Shin and Heidemann that the clock will be set to a full one second. The Korean coach comes out to argue but is overruled.
52:56, Heidemann scores the single-touch she needs to advance.
Resetting the clock to 1s is actually unusual, if I remember right, and probably shouldn't have happened - I suspect that this was the mistake by the timekeeper (which shouldn't be held against them, Olympic timekeeping is not a "best-effort" thing). It's hard to say who "should have won" the match - Heidemann certainly played a more aggressive and interesting game but Shin's conversion so many of these into double-touches or even points for her are exemplary. As another poster points out, I suppose the real answer is to voice recognition software start the clock on "allez".
Lastly, I just wanted to salute Shin's courage and dignity in coming on to play the bronze match, even though she must have known she had no chance to win the medal. Even having rewatched the match in slow-motion half a dozen times, I have a hard time feeling convinced that the right decision was made - from her perspective she must have felt absolutely robbed.
My wife and I are both disgusted with the coverage. Sure, we like to see kids from our country win, BUT, we all live on the same planet, and nationalism belongs in a previous century. The human interest stories are not bad per se until you realize how much broad international is bumped.
For the OP story: if that young woman's worst possible result if the error had not occurred was a bronze, it would have been good if the international Olympic committee could have firmly pulled their heads out of their asses and given her a bronze medal. It would hurt no one if two bronze medals were awarded.
It isn't just that the players weren't american, it's the sport as well. When I was in fencing school in the US, the coach had to order the tapes of the fencing bouts of the olympics, because they weren't even shown on TV in the US.
To round this up, we Germans just got evened out ;)
In the men's team semi final, Peter Joppich fought the final match against Yūki Ōta. In the extended time, Joppich scored a single hit which was apparently not counted because Joppich obscured his vest with his mask. Another double hit which to me and the commentator seemed like an attack by Joppich then (after a few minutes of discussion between the referees) was counted for Ōta, leading to sudden victory for Japan.
"giving Heidemann more than one second to land the winning touch"
This seems the crucial point, and without having it specified it is hard to tell if this is a very serious oversight or very minor aspect. If Heidemann had 1.01 seconds instead of 1.0 to land a hit, this would be very minor. Whereas if Heidemann had a full minute of extra time, this would be major.
I'm not doing any extra research since this isn't too important to me -- but some crucial context is very clearly missing here.
I was trying to find a clear description of what happened, with supporting evidence at the right times, and it was surprisingly hard (many arm-chair fencers).
But there are questions to ask, and lessons to be learned for startups, and for entrepreneurs in general. Here are some of the questions raised for me. Perhaps you can think of more:
What processes do you do by hand that could be automated? The timing was done by hand - a notoriously error-prone process. relentless automation can systematically weed out systematic errors, provided they are properly monitored and tested. Then you can forget them, and let them do their job.
Do you have the right people doing the right jobs? The report is that the time-keeper was an inappropriately chosen volunteer. It's notoriously easy to leave a job in the hands of someone who may not get it right when it's important. And just when isn't it important? (actually, sometimes)
Do you have the right error reporting processes in place? The team lodged an appeal, and clearly that didn't go anywhere. Why not? The entire bout could surely have been replayed and the timing checked. The result, in truth, should never have been in doubt, and could easily have been overturned.
Do you have the right corrective measures in place? After reviewing the bout, surely it should be straight-forward to correct the result. When you have a mistake reported, can you roll-back and make it right?
And finally (for me - perhaps you have more points to make), always, always consider the individual. This poor woman was left distraught on the piste, unable to leave, in full view of the audience, in tears, her lifetime's work hanging in the balance, out of her control.
So let me ask - how well do you look after your users?