June 23, 2012
Four Man Round Robin is a Waste of Time

If you watched the first day of Red Bull Battlegrounds, you’ll remember a third stream was added on short notice towards the end of the first day in order for all the group stage games to be aired. Games went longer than expected, and the decision was made to start playing more games in tandem than the original plan called for and broadcast the replays after they finished.

Some might call this unavoidable. Starcraft does not have a pre-set time limit, and games can end in just a few minutes or go as long as an hour. When the tournament calls for each match to be a best-of series then not only is the game duration variable but the number of games played is as well. This makes scheduling and organizing a tournament an unpredictable nightmare. This isn’t good for anyone. Fans don’t know when to tune in to see their favorite player, players in the event aren’t sure when their next match will be, and tournament administrators only have so much hair to pull out from stress. It’s easy to chalk up a debacle like this to bad luck. But is it just bad luck, or does a bad format share some of the blame?

The Red Bull Battlegrounds tournament used a structure in the first phase of the tournament which was simply wasteful. The idea of splitting players into four man groups and playing a round robin with the top two players advancing has a history with Starcraft dating back to the OSL. However it has been mindlessly misapplied in the western SC2 tournament scene. Four man round robin is almost completely inferior to a simpler, quicker, common alternative. This is something GSL realized over a year ago when they replaced the system’s use in Code S’s group stage.

GSL switched their group stage to a double elimination bracket or short swiss style tournament. Two pairs of players face off, creating two winners (1-0) and two losers (0-1). The two winners play to decide who advances (2-0) and who plays the consolation match (1-1), while the two losers play to decide who else enters the consolation match (1-1) and who is eliminated in fourth place (0-2). The winner of the consolation match advances in second place (2-1), while the loser is eliminated in third (1-2). Five matches are played in this method, one less than the six required to settle a round robin with the same number of players.

Some in favor of round robin argue that the extra match offers additional games to broadcast for the fans, but this does not strike me as a benefit. We live in a time when there is an overabundance of Starcraft content. There are multiple major organizations putting out content multiple times weekly, along with live streams, minor tournaments and showmatches each day, larger tournaments each weekend, talk shows, and so forth. There is simply more Starcraft related media than any one person could reasonably expect to consume.

It stands to reason that competition organizers should not to try and squeeze every possible minute of Starcraft footage out of each tournament but instead to make every minute as relevant and exciting as possible. Since these round robin groups take place at the earlier stages of the tournament, no tension has been built and the prize is far off in the distance. Saving time by settling the earlier rounds with less games affords the organizer the possibility to increase the number of games later in the tournament when there is more at stake. The DreamHack Summer Open had 56 groups play out before entering the RO16. Playing five series instead of six would skim 56 BO3 series off the schedule. The time saved would allow DreamHack to host longer series during the climactic stages of the tournament instead of using BO3 all the way until the finals, with plenty of time to spare!

A proponent of this group play system would claim that I’m in favor of sacrificing the integrity of the tournament in order to save time, but this isn’t the case. In this particular application of round robin, losing two games guarantees elimination before the third game is even played most of the time. To advance after two losses one must not only win his third game, but also hope for the right sequence of results to take place among the other players in the group to create a three way tie. Oftentimes it’s clear this isn’t even possible before the third match has even begun. Furthermore, since most tournaments ignorantly use criteria to break ties rather than extra games (see another article, Break Ties, Not Hearts), advancing after two losses is even less likely to be possible before the last match is played, let alone actually happen. This suggests that the double elimination method saves time with little loss in accuracy.

In a round robin group the meaning behind the matches is vague. It’s clear that losses are bad and wins are good, but precisely which matches or games are the most significant are usually not apparent until the moment has already past. There is little excitement or suspense if the match seems to lack a clear and direct consequence. In the double elimination group, the first loss is recognized as a setback, but not a catastrophe. Two things are made clear: the next game or two games must be won, and neither will take place against the best player in the group.

In a round-robin group, this isn’t necessarily the case. Imagine a group where two players are closely matched (B1,B2), another is better and will always win (A) and one is worse and will always lose (C). Regardless of which format, we can expect that A will advance and C will be eliminated. In a round-robin format, B1 and B2 paired in their first or second match would obscure (to each other and the viewers) the fact that the match between them is the key to continuing in the tournament. In the double-elimination format, these two players will always meet in the consolation match to decide second place, regardless of the pairings. Better still, being paired together in the first match will mean the consolation match takes place after they’ve already played each other and can better adapt to one another. Considering this, one could argue that double elimination is more accurate than round robin when it comes to determining the rightful second and third places.

Double elimination groups will never creates a situation where a meaningless or lopsided match can be played. Every match takes place between two players on equal footing with the same at stake. Round-robin groups can create circumstances where one player will advance and his opponent will be eliminated, regardless of the outcome of their match. Neither player has much incentive to perform at their best and make a game worth watching. Ethics and obligation could motivate the players to go out and play anyway, but why even bother? And who could expect a forlorn player, his hopes demolished only moments earlier with the knowledge that his tournament has ended, to go out and play well even if he wanted to? An even worse circumstance is when the eliminated player is matched up against a player whose chances for advancing is still up in the air — this gives the player in contention a clear advantage that the other players in the group did not have.

These lopsided scenarios not only degrade the accuracy of a result from a round-robin group, but also provide the opportunity for nefarious players to collude. If in a match between two players there is an outcome where one gains much and the other loses nothing, the players can guarantee that outcome and fix the match. While there is no format that can prevent players from match fixing if they so choose, a round-robin format actually incentivizes it by creating opportunities for a player to enter a match knowing that he can take a dive and still be guaranteed to continue. This is in contrast to other formats, where choosing to lose a match would involve sacrifice. Perhaps even a greater problem than the potential for cheating is that creating these opportunities could cast suspicion on any player involved whether or not they engaged in foul play as long as motive could be argued. Since proving whether or not a match was thrown is virtually impossible, not only is there no way to catch prospective cheaters but also no way to clear an innocent man’s name.

OGN’s use of four man round robin used single games, not BO3 series, to settle matches. The likely intent behind this structure was that playing single games against multiple opponents was an alternative to a multiple game series against a single opponent. Furthermore, the fact that matches were individual games and took place amongst a highly skilled pool of players greatly increases the likelihood of a tie occuring, in which case OSL always held sudden death matches. While I would still prefer double elimination (so did MBCGame, RIP), OGN’s application has merit considering its place in a league spanning several months rather than a single weekend and that a single group would play out over several weeks rather than a couple hours.

The method like the one used at DreamHack and Red Bull Battlegrounds is unnecessarily excessive to the point of madness. Whether it is stemming from a misguided attempt for an accurate tournament, a desire to offer players more competition time, or that simply the longer a stream stays live the greater the number of total views, in the end it just creates a poorer product.

Is four man group play with double elimination indisputably the optimal format? Certainly not. However, it would be trivial for any tournament used to or partial to four man groups to choose the double elimination method instead of round-robin. Doing so would do wonders for reducing the length and increasing the quality of the early stages of a competition.