Putting this here instead of a PasteBin because, what the heck, I’m paying for a website, I might as well use it.
Some thoughts on the Swiss rounds of the recent Ninja Gaiden II speedrunning tourney:
(A brief note that when I say “the tournament”, I mean “The Swiss portion of the tournament”. I’m aware we still have Top 8 to go as of this post, but this feedback concerns the Swiss portion)
tl;dr version: Swiss can definitely work for speedrun tournaments, even for longer games with some massaging. It would have to be done carefully, though. This one was pretty darn great, despite some issues.
First of all, I’m familiar with how Swiss works from my days playing the WWE Raw Deal card game. That said, I didn’t pop in on this particular tournament until midway through the Swiss rounds (So round 3 or 4).
On Swiss in general:
- It’s great for separating the cream of the crop from everyone else (This also works on the opposite end). It’s less good at reliably separating out the middle–you wind up with a bunch of people with roughly .500 records, due to the design of Swiss.
- It ensures the tournament more-or-less always remains exciting for the competitors, at least in theory. Besides the mish-mash in the middle where you typically have multiple people fighting for the final few “playoff” spots, the idea behind Swiss is that everyone continually gets matched up with people of roughly their skill level, as indicated by their records to that point. This works for the audience as well, as the matches should always be close (Sinister noted that they only had to switch featured races due to a blowout in one round).
- It does require a bit of time investment on the part of the competitors, especially if you want to try and get it all out of the way in a single day. As a side effect of how Swiss works, the tournament is effectively halted until an entire round completes. Contrast with a single or double-elimination tournament, where one “straggler” doesn’t slam the entire tournament to a halt, at least at first.
Now, some thoughts on this specific tournament, and whether Swiss can work for speedrunning tournaments in general:
- I think, as Sinister previously said, the answer to the question “Can Swiss work for speedrunning tournaments?” is a resounding “yes”. However, that doesn’t make it the automatic best choice across the board, as he notes.
- The main thing to consider is “Are you trying to get the entire Swiss round out of the way in a single day?” While the NGII tournament answered that question with “Yes”, I don’t think that has to be the answer. If the answer is “yes”, though, you’ve restricted yourself to games in the neighborhood of a half hour long, max. Doing a Swiss tournament over multiple days would be more difficult (Someone not getting their match done in a timely fashion/being removed mid-tournament has more of an effect than it would in a more “traditional” tournament, where you just remove the person and anyone in that person’s “path” effectively gets a bye. In Swiss, a removal kind of screws everyone who played the removed person due to how tiebreakers work, through no fault of the people who got matched up with the “straggler”), but not impossible if you have participants who are truly committed to getting games played–just run it like the Mystery Tournaments do, where you have a specific range of days to complete your match.
- Having your Swiss round spread out over multiple days would also help mitigate what is probably the biggest challenge for commentators–striking a balance between helping people unfamiliar with the game without being repetitive for those who tune in from the start. The shorter the game, and the more rounds your tournament is, the bigger this issue. I think Sinister and Duckfist did a fine job of handling this in this tournament specifically, by telling some of the “stories” of the featured racers.
- Sinister noted that the event lasted longer than he would have liked due to various issues. The main thing, to me, was actually that the tournament went 1 round too long. In most Swiss tournaments, the general rule is 2^x = N, where N is the number of participants rounded up to the next power of 2, and x is your number of rounds. So with 30 people, there should have been five rounds, not 6 (The idea being at the end, you’ll have one person with an X-0 record).
- Another potential reason that the event ran long from what I saw was the desire for multiple featured races in a round, which meant that some races didn’t start until after others were finished. While I don’t necessarily agree with the “Too many featured races” critique, if the primary goal of a future tournament is simply “End it as quickly as possible” as opposed to “Showcase as many runners of varying skills as possible”, having everyone start a round at the same time would be preferable, and doable.
- Going back to my point about Swiss separating the cream of the crop from everyone else, I’ll note that 7 of the top 8 seeds advanced to the Top 8, and the 8th person was still firmly in the upper echelon of entrants, seeded 11th out of about 30 people. I think this shows that the system does work–people claim to love a Cinderella story, but they generally get pretty annoyed with anyone but “their” Cinderella story making a deep run. In that sense, I think this portion of the tournament is a definite success.
- I’m a bit confused about the comment regarding seeding in subsequent rounds. A brief note on Swiss: Round 1 is seed-based. Round 2 then consists of 1-0 v. 1-0 matches, and 0-1 v. 0-1 matches. Round 3 continues in this pattern as reasonably as possible without repeating past matches (So 2-0s play 2-0s, 1-1s play 1-1s, 0-2s play 0-2s). This pattern continues for the remaining rounds. What I’m guessing is that within that framework, Sinister and the other organizers had the highest-seeded 1-0 play the lowest-seeded 1-0, and so on. I’m not sure how big of an issue this really is, particularly in light of the previous point.
Still, overall, I’d say it was a big success from a viewer standpoint. Basically every race was exciting, and I enjoyed the interviews too. I’d definitely like to see more tournaments use the format.
-EE
(EDITED to fix typos 5/6/16)
Loading tweets...
2 comments
Thanks for the comments EE, really helpful stuff! I would like to address some of the points you made/questions you had. Sorry they are not in order.
1) You are correct about the number of rounds, and I only stuck with six because we had more participants and many dropped last minute. I did not want to change expectations on such short notice for the entrants who had been told 6 rounds since the beginning pretty much. I don’t really think an extra round hurts in determining top 8. Interestingly enough the top 8 after 5 rounds would have been the same, but seeded differently. However, the sixth round did give nine of the middling competitors a shot at the top 8, which I think is more fair to them. If round 6 had played out differently we could have had a dramatically different top 8.
2) The Swiss was seeded in subsequent rounds which hurt middle/lower ranked players, and is something I would probably get rid of going forward. Randomly pairing 2-0s vs. other 2-0s seems more fair by not necessarily punishing “Cinderella players” as harshly for doing well. Fastatcc was the player that was hurt by this the most.
3) We had a different tiebreaker system that was based on individual performance (sum of total times over six rounds) to avoid other poor tiebreaking scenarios that can occur with Swiss.
Hopefully that clears up some stuff, but I agree with most of your comments, including that longer games up to 30 minutes can work. In those cases you would obviously never have more than 1 feature match per round. You could try and do Swiss across multiple days and I thought about this but if people can’t show for the second day for some reason that really messes things up a bit.
I have not thrown this out there yet, but what are some games that people would like to see Swiss tournaments for? I will get the list started: Chip N Dale: Rescue Rangers. Super easy to learn and get a decent time, already has a big enough community to support without a ton of new players.
Author
And thanks for the response! It’s cool that it wasn’t in order, since I don’t think my original post had much of a coherent order to it anyway.
1. That makes a lot of sense. If people were told from the start it would be 6 rounds (And you thought you would have the entrants to support it), then yes, keeping it at 6 rounds as opposed to changing it last minute is probably the best solution. And as you mentioned, it ultimately didn’t make a difference in determining which 8 players would advance. It is true that the extra round did give those on the outside-looking-in an extra shot at cracking the top 8, though I would say that’s the glass-half-full way of looking at it. A more cynical person (Such as myself) would argue the flipside of that, which is that the extra round potentially screws those on the inside-looking-out after 5 rounds out of their rightfully-earned playoff spot by forcing them to win an extra match (And I would argue that hard if people hadn’t been told from the beginning it would be 6 rounds).
2, 3. I see. I’ll be honest, the “Raw Deal days” I mentioned in the original post were ten-plus years ago. As such, I don’t remember if there was any kind of institutional memory that carried over results from tournament to tournament, such as some kind of ELO rating, that was used to determine post-first-round pairs, or if they were just random (Within the Swiss parameters). For a one-off tournament, doing it randomly is probably best. Also, going by lowest average time for a tiebreaker is a great idea that also reduces the effect of a potential tournament dropout for the people who would have played that person.
Also, I’m glad you found the comments helpful. From your Pastebin, I got the impression that your initial thought behind doing this was “I want to try and get most of an event done in one day, how can I do that?” and arrived at Swiss as a potential solution, which I then took and wondered if it would be applied in other contexts/multi-day events/etc. Because of that, I wasn’t sure how useful my observations would actually be.
As for other games I’d like to see in this format, I don’t know if it’s because I have “Sinister1 on the brain” or what, but my first thought was actually Batman on NES. Similar length to NGII (Probably a bit longer on average), and the steep penalties for an unintentional death mean even a seemingly out-of-hand race can turn around in a hurry with one mistake. Other potential ideas, building off of your Rescue Rangers suggestion, are Ducktales 1 and 2.