Playing to Win in Badminton
Wednesday, August 1, 2012 at 12:15PM There's a recent controversy about players losing intentionally in Olympic badminton. A lot of people involved seem concerned that it's embarrassing for the sport. It its. It's embarrassing that some officals and spokesmen of the sport have so little understanding of Playing to Win that they think the players are at fault.
Playing Fake Matches
I have run many fighting game tournaments, and I have witnessed fake matches. I completely agree that fake matches make a mockery of the tournament. This is so important that one of the MOST IMPORTANT considerations when designing a set of tournament rules is to minimize the chances of fake matches occurring.
Forfeiting a match and playing a fake match are similar (in both cases, one side is losing on purpose), but not exactly the same. Forfeiting should be a natural right of any player in any tournament. A player should be able to forfeit for any reason or no reason, and this must be make explicitly clear in the rules. Further, it should be explicit that if a player (or team) wants to forfeit, then they should NOT play a fake match. Playing a fake match is about the worst possible thing for a competition because of the impact on spectators. If the rules make it clear that simply forfeiting is far preferable to playing a fake match and that forfeiting comes with no penalty, then the rules will have stomped out 90% to 100% of fake matches right from the start. It's just a lot more effort to play a fake match and there'd be no benefit over forfeiting.
That's not the whole solution though, not even close. That's just the failsafe you need in case there is any incentive to lose on purpose in the first place. It should be self-evident that if a tournament system ever gives players an incentive to lose, then it's a problematic tournament sytem.
Losing on Purpose
Let's look at some cases where you'd want to lose on purpose. First a few that don't have to do with the Olympic Badminton case, then the one that does. (If you only care about that, skip to the "Back to Our Story" section below.)
Let's start with two terms from game design: lame-duck and kingmaker. In a game with more than two players (or more than two teams), a "kingmaker" is someone who can, through his or her in-game actions, decide which OTHER player will win the game. The kingmaker is so far behind that he can't win, but he could deal a card (or whatever) to Alice or to Bob, which would determine the winner. This is considered really bad because you'd hope Alice or Bob would win off their own skills, not from some 3rd party's vote. "Lame-duck" (a term I use because I don't know what else to call it in game design) is the portion of a game where a certain player cannot possibly win anymore but somehow they are still stuck playing the game. Lame duck players are ripe to be kingmakers. When you don't have skin in the game anymore, so to speak, your potential to screw things up for others is pretty high. (Note that this is NOT what's going on the badminton case right now.)
Swiss. The kind of Swiss that at some point cuts to single elimination (for a more exciting finish) is full of lame ducks and kingmakers. In this format, you need a certain win/loss record to make that cut, but you can keep playing against more opponents even if you have a win/loss record that is *guaranteed* to NOT make the cut (lame duck). It's entirely possible that you will face someone who still has skin in the game: if they win they will make the cut to the top 8; if they lose, they won't. And you can decide that by forfeiting or not, with no effect on yourself, because you are definitely going to lose the tournament either way. Magic: the Gathering uses this format. You'd expect it would lead to shady situations because of all the lame duck / kingmaker stuff. And it does.
Round Robin. In this format every player (or team) plays every other player (or team). It has the very same problem as Swiss: lame ducks and kingmakers. You can be in lame duck situation yet determine the fate of your opponents. This is just ripe for their being under-the-table payoffs. Round Robin also has problems with the order that matches happen to occur in. If you have to play all your matches right at the start, you don't have the benefit of knowing the results of all the other (future) matches, so you don't know if you can get away with losing on purpose. But if your matches happen to be scheduled for later in the tournament, you do know the results of so many other matches that you can now do shady things. So all players don't even have equal access to the shady tactics, as it depends on the luck of scheduling.
Back to Our Story
And now we come to the actual problem with the Olympic badminton situation. There are "pools" of round robin play where the top 2 finishers from a pool advance to a single elimination bracket. Further, the system of seeding in the single elimination bracket is known ahead of time. This creates the situation where you could playing pool matches but *guaranteed* to make top 2 by your record. If you win, you will qualify and play team X. If you lose, you will also qualify, but you will play team Y. If you think you have an easier chance of beating team Y, you absolutely should lose on purpose. If you don't, you aren't playing to win, and you are kind of a bad competitor. You also happen to be playing in a tournament with absurdly bad rules.
I hope it's clear by now that tournament systems absolutely can have incentives to lose. And if you are holding such a high profile tournament as *the Olympics*, then I hope you'd deeply understand all this and design a system that minimizes or removes all incentives to lose, and adds in the failsafe of encouraged forfeit rather than fake matches if there was some overlooked edge case. It's LAUGHABLE to put even the tiniest amount of blame on the competitors who are playing to win here, when the tournament rules so clearly, so obviously, and so predictably have major problems. That is, you wouldn't need to even hold a tournament to detect this problem. You could just read the rules, see the clear and major flaws in them, then you'd want to direct your blame at the rules writers and correct the system.
It's doubly laughable to actually disqualify the players involved—how about disqualifying the judges? They don't seem capable of making competent decisions about tournament practices. Those who conspired to disqualify players for playing optimally inside a bad rules system are doing the sport a real disservice. Hearing about fake matches in badminton should make our opinion go down, but hearing about the sport's inability to see glaring problems in its own tournament structure should make our opinion go down an extra ten notches.
It's an embarrassing time for Olympic badminton. But not because some players lost on purpose—because someone created horrificly bad tournament rules and then tried to blame the competitors for playing to win.
Reader Comments (203)
@Sirlin, you're wrong and David S is right. Random seeding would have NOT changed the situation for the Chinese teams, at least, not enough to make winning a better option than losing. Read his first comment again.
@David S
"On the other hand, if they had stopped the tournament before the final games were played and played them all at the same time, China A would not have tanked the game."
That might hold true for this situation, but it can not be generalized. Suppose Chinese Team A can't get first of their group regardless of the outcome of the final match. Team Chinese B would still want to throw their game if that may allow them to avoid Team A in the bracket.
This is a interesting situation. Double elimination would allow Team A and B to win both gold and silver regardless of the initial setup, BUT it would still be advantageous to them not to play in the first rounds (so none of them has to go through the losers bracket). But honestly, I don't see any better solution.
All of the team sports in the Olympics use it (basketball, soccer, hockey, handball, volleyball, water polo, etc.) Major international soccer tournaments, including the World Cup and Euros, use it. The major U.S. professional sports leagues (MLB, NFL, NHL, and NBA) all have some elements of it, though mitigated by longer "pool play" seasons reducing both flukes and ability to position yourself strategically.
Individual sports generally use straight single-elimination. Badminton used to do that, but went to the pool play for doubles to increase the number of games (reducing from 16 teams to 8 now takes 24 matches instead of 8). Badminton singles have a pool-to-elimination format, but only the top place player from each pool advances--so there's no direct incentive to lose, though the kingmaker/lame-duck problem is still possible.
Problems with the pool-to-elimination format are relatively uncommon, but certainly not unheard of. International soccer has the final matches of any given pool played simultaneously to somewhat reduce the information issue dealing with intra-group shenanigans after one particular infamous incident, but cross-group issues (which is what happened here) are still possible -- for instance, the 2014 World Cup B pool final games begin three hours after the A pool games, so the B pool teams will know the #1 and #2 seeds from A pool when the B pool team plays.
In short, it's a design that creates some risk of this sort of problem happening, and the BWF certainly should have known of the risk. Presumably, they weighed that risk against the benefits of additional matches for showcasing the sport, a more forgiving earlier round for the top teams (after all, if the China B-team had met Denmark in the first round of the old design, the Chinese would be done rather than still in competition), and any other considerations they had.
Are there better tournament designs that would provide the benefits sought without the problems of this design? Certainly (as the above discussion here has convincingly shown). But it'll take more than just pointing at the BWF to reform the sports world and move it away from this format.
Another part of the problem with any single elimination tournament that ends in awarding 3 prizes for top 3 finishers is that single elimination is super, duper bad at determining anything other than the best entrant. (Consider true #1 plays true #2 in first round.) Regardless of the method of advancement and seeding from pool play, it seems like asking a single elimination tournament, even with a "bronze medal match" to determine the best 3 teams is nonsense.
A piq: I'm actually not wrong that it's insane to have completely known seeding in a case where a loss counts for nothing. That really and truly is infeasible and cannot work. If you are concerned about the different problem of teams helping each other, then sure playing matches at the same time is a good solution there. No possible solution ever in the world can have known seeding where losing counts for nothing though, so when I said that, I didn't say a wrong thing. To restate the fundamental point: the lowest hanging fruit is "don't make a tournament structure where losing is good for you. If you have passed that fairly simply test, then you can start to think about OTHER problems, like losing to help some other team."
Andy: Yes, single elimination is something you do for time reasons. It's good in that it's not corruption prone, but bad for the reason you said. So it was kind of jarring to hear Shigeru say a sport went to double elimination "to show more matches" rather than "to be a better test of anything in the first place." I accept that single elimination is needed for time-reasons, but if three's any way to avoid it, you probably should. The exception, imo, is if the event in question is something that happens pretty frequently. For example, if we had weekly or bi-weekly Yomi tournaments, single elimination might actually be a good idea because it keeps time commitment low and if you got a bad draw, try again next week. If instead it was once every four years, that would suck though.
Really curious about this, hope it doesn't get buried. Some poker tournaments at casinos (CA) offer a bounty for knocking out any player, like a flat rate per person that's not part of the tournament pot, just as a bonus. Could something similar to a bounty curb losing on purpose? Great article, well explained, thanks!
China A losing to 'help' China B reach the finals is (in this case) indistinguishable from China A losing to get better opponents, since China B is a very good team and China A may well have thought of them as their greatest threat.
Secondly, to everyone who says it shouldn't matter because the best team should win: if China A were secretly the second or third best they might lose on purpose to avoid hitting the best team until finals. This would give them a silver or bronze even in a completely deterministic world where the best team wins for sure. This can be fixed by making the prizes winner-take-all (in the deterministic case), but that doesn't really work for the Olympics.
Thirdly, what do you guys think of double elimination until the Grand finals, then a single elimination match? Since you have to hold Winner's Finals before the Losers or Grand Finals, there's an incentive to win that match (and not face the risk of being eliminated in Loser's Finals; also you guarantee yourself at least silver). Obviously, there's no incentive to lose any match in any Loser's bracket event, since you would be eliminated. There's also an incentive to win every Winner's bracket game, since staying in Winner's and losing next round is never worse than moving to Loser's bracket now.
Fourthly, to address the concern of "what if too many people want to forfeit", that's a matter of making sure the prizes / social standing / whatever from the tournament are big enough to motivate players to actually play. It's an issue that's not related to the structure of the tournament or the order of matches.
Fifth, as magical super spackle over bad tournament rules, how about: players can forfeit any match for any reason (gambling and such aside). If both teams want to forfeit, then only in this special case will they play it out and the winner of that game will be declared the loser for the tournament structure. (In essence, it's assumed that teams want to win and will play for this right, otherwise if teams agree on the outcome it's assigned by their agreement, otherwise if both teams want to lose they play for that right).
Finally, I feel like it's 100% unreasonable for the organizers to disqualify players based on the style of their play. It's not really enforceable, it's a poorly-defined rule, and it sets the precedent that the game is played in a certain way. This has some chance of eliminating-by-rule otherwise valid strategies like the full-court press in basketball, or 6-man offense in hockey. The obvious question is: what can sports fans do about it? Boycott Olympic viewership? I don't think that will ever succeed as a social movement - not enough people will do it.
So yeah, all of the usual arguments notwithstanding, it occurs to me that the actual rules of play were unclear. That is, there seems to be an "integrity of the sport" consideration, which is basically a giant, assumed house rule when you get to that level of competition. As I see it, the main problem is that the determination is subjective, which leads to subjective enforcement.
Of course, the truly fascinating part is trying to decide whether "fair play" is enforceable. You can enforce it mechanically (through rules), or socially, but at the end of the day you don't get a medal for being nice.
And yes, I feel bad for the athletes; they aren't blameless, but I'm sure their trainers were complicit (if not coercive).
It's nice to see that with Shigeru there is actually someone taking part in this discussion who knows about sports tournaments. Before that it seemed like a very elitist group afraid of "violent and stupid sports fans".
Unlike in physically less demanding things (like games) in sports you need to take fatigue into consideration. Which makes random next opponent or choosing your next opponent impossible. At least if you want to have the tournament to be over in a reasonable time without too many days without any action.
Also, it is impossible to have all games taking place simultaneously as this would either require a huge venue or several ones to be used. Also, as spectatorship is a big part of sports, this would decrease the enjoyment of watching it - as you could not follow too many games simultaneously.
Double elimination is not a good solution due to the flaw that it might require two finals and thus is horrible from a logistical point of view, especially for sports where the required recovery time after a match is high (like soccer).
Mober's idea is very interesting. Essentially, rather than tying your choice of next opponent to whether you "won", there are two 'equal' paths at each level of the tournament - say, red and blue - and the reward for winning a match is to choose which path you're going to take.
In fact, I'm thinking it implies a tournament where, instead of a pair of matches, the basic unit is a set of _three_ matches - call them A, B and C. From round to round, the players in each match get redistributed - i.e. one of the players from match A will play against one from match B, and the other will play against someone from match C. The winner of match A gets to choose how that distribution happens.
(He doesn't get to choose a specific player, just a match. e.g. suppose the winner from A chooses B, the winner from B chooses C, and the winner from C chooses B. Outcome: the winners from B and C play each other; the winner from A plays the loser from B; and the losers from A and C play each other.)
Alan Au: "fair play", by which I assume you mean good sportsmanship, is absolutely not enforceable in all cases. One of the chapters in Sirlin's book talks about banning things in games; the ban has to be discrete, enforceable, and warranted. So for example, banning camping in an FPS is not enforceable because if you prevent players from staying in the same place for 3 minutes, they'll just stay there for 2 minutes 59 seconds. If you lower it to 2:30, then they'll camp for 2:29, and so on and so forth.
In the case of the badminton players, all they had to do was sell the loss better. Maybe play the game straight until the score is 18-18, then throw the last three points, or maybe throw every 4th point. How do you detect that or punish it mechanically (via rules)?
While it is true that the tournament system is badly designed and needed to be fixed, the officials basically had no choice but to to disqualify these players. Imagine that the official do nothing at all, and one of the teams who played fake game managed to win the gold medal later. Then the following conversation may be heard on the street:
Person A : What do you know about the gold medal winner of the Olympic badminton game?
Person B : I don't play badminton, but I heard that they win the gold medal by playing fake game.
That would be a great blow to the reputation and image of badminton as a professional sports. The officials had to do something to tell the public that they are at least trying to correct the problem. And saying that they are going to improve the tournament system by some or other methods would not produce the same forceful impression to the public as that produced by disqualifying the players.
I'm rather upset that the teams were punished in this manner. Anyone stating that losing on purpose was "Olympic spirit" is being foolish since the goal in the Olympics to win medals for your country. Virtually every new network summarizes the Olympic results by doing a medal count by country.
A team that isn't confident about beating a specific team should absolutely lose on purpose if it increases the chances of them getting a silver or bronze. The second best or third best teams win nothing if they face the best team and are eliminated too soon. It's totally in the spirit of the Olympics to try and win medals.
"In the case of the badminton players, all they had to do was sell the loss better. Maybe play the game straight until the score is 18-18, then throw the last three points, or maybe throw every 4th point. How do you detect that or punish it mechanically (via rules)?" - Lettucemode
This doesn't work unless one team wants to throw the match and the other does not. If both teams wish to lose (like China A did), then an agreement to play to 18-18 then throw the match will be broken by one team playing to 17-18 and throwing it (to guarantee the loss). The 'best' (only) way to throw a match both teams want to lose is to score as few points as possible and allow the opponent to score at every opportunity.
Yeah, this story was pretty ridiculous. A couple other random things I heard.
First, I heard some badminton federation official stating that the reason for these ridiculous rules was "We wanted all teams to be guaranteed to play at least 3 matches because it would be awful if someone travelled all the way to the Olympics and barely even got to compete." So, apparently their solution was to add in a bunch of fake matches where the teams had more incentive to simply save avoid fatiguing themselves than bother competing anyway. Great move there.
Second, I heard it quoted that the Chinese team that was disqualified are planning to quit the sport now. At least the players understand what a farce this was, and have determined it's not worth their time to stay involved with an organization that doesn't know how to organize anything.
The really stupid part is that now we have teams accusing other teams of "losing on purpose" in order to gain a better position in the tournament:
http://www1.skysports.com/news/12040/7955498/
Imagine if a team accused another team of winning on purpose in order to do better in the tournament. You'd think that's the stupidest thing you've ever heard. Now raise the wtf factor 10x, and we get something like this.
I disagree with your opinion. You also don't provide any suggestions as how this could be fixed. Any solution that guaranteed fairness amongst players of differing strengths would need to be so complex as to be unworkable in reasonable timeframes.
But where I disagree is your opinion on playing to win. It is not that straightforward. Do you want to play to win, or do you want to be the best? It's true that more and more people see winning as the end goal these days, these people are short-sighted. Those who want to be the best don't throw matches or take short-cuts (and if you don't know why, you'll never be counted among them).
@Sirlin: Great post, I agree that it is solely the onus of rulemakers to ensure that their system creates a fair and enjoyable experience. If you don't want people to throw games, you'd better make sure it's not the optimal strategy for winning or specifically prohibit it. Wishy-washy "spirit of the game" garbage is nothing more than an excuse for the organizers to re-write the rules halfway through the tournament, which makes a mockery of the competition. They might as well disqualify a Volleyball team for winning by spiking the ball.
@Thanatos: "If you're a good competitor - you are always playing to win, no matter if it's a team that's all but out of the tournament, whether YOUR team is already out of the tournament, or if it's just determining seedings. If you don't think you can beat each and every other team in the tournament you have no business winning it. "
Maybe this holds true in your fairytale ideals, but real competitors have limited stamina to spend over a period of time. Even if the rules are such that losing is never optimal, going balls to the wall in every match is not playing to win the tournament. Let's say that Teams A and B are the best in the world, and A will doubtlessly beat B as long as their players have equal stamina heading into the finals. If both teams can get to the finals by playing at 50% capacity but Team A decides to give 100% to every game while Team B puts in the minimum effort, Team A has sabotaged their chances of winning even though they could beat every other team. That's not competing, that's showboating, and any "competitor" stupid enough to do that will be punished for it unless they are absolutely light years ahead of their competition.
Do you know of anything that can be done to eliminate fake matches due to a pot splitting agreement?
I have watched some fighting game tournaments where a finals match that should've been good was instead boring because the final two players agreed to split the pot and consequently did not care about the outcome.
The Olympics solves this issue quite neatly. A gold medal cannot be split. Perhaps a payout method less immediate than cash such as a check would make it inconvenient enough for the players to split the pot that they would not do it. EVO points help as well but not all tournaments give EVO points.
I wonder if a tournament system in which the players could pick their seeds after Round Robin / Pools would be appropriate. Each team would get to select their seed in the rank order of RR play based on Win - Loss, point differentials or what other bajillion metrics you decide on. You would still have the problem of kingmakers in rare edge cases.
In general you'd expect the seeds to come down to the traditional pairs of 1 - 8, 2 - 7, 3 - 6, 4 - 5. However in the edge cases of a pair being from the same country a team could intentionally pick a lower seed. Sounds odd, but you'd be giving people the option of playing a harder seed and every incentive possible to be win all your games in RR for that option.
@Sirlin
Your forfeiture stance is a non starter. I understand why it is good for the tournament from a game design stand point, but in a tournament why spectators and advertisers pay large sums of money to have access to these games (e.g. the Olympics). The existence of the tournament depends on them. For example, if forfeits were allowed then the last day of pool play may have a large number of forfeitures; something that would be unacceptable to spectators. Also, on a semi-related note, even though a couple of people brought it up, you haven't said what would happen if both teams wanted to forfeit.
Your idea of randomizing the seeding after pool play does a couple of things. First, you are right it does eliminate incentives for losing. However, it also eliminates incentives for winning. Like someone else mentioned, this situation is similar to the SUMO wresting described in Freakonomics, and is, thus, ripe for bribery and payoffs.
Your idea of hiding the algorithm of the seeding would not work very well because of the limited variability that could be used. For soccer (something I am a bit more familiar with) you only have 3 real variables to use; wins, draws, (loses are constrained by the previous 2 variables) and net positive goals (the number of goals you have scored minus the number your opponents have scored). With such limited variables, algorithms would rarely created differences or only create a small number of differences between them.
The best solution offered so far (although not perfect) is to let the top seeds pick their opponent each round. This not only eliminates an incentive to lose, it keeps incentives for winning and even more, winning as by as much as possible (since that would factor into rankings). For example, selection for getting out of pool play would be the same as before, but then for rankings you could use the net positive points. That way, in the deterministic world where the 2 China teams are 1st and 3rd best, there would never be a reason for one team to tank and be in the lower half of the rankings.