The introduction of swiss and round robin for tournaments on lichess

Just a few examples then: (adapted from a post of Poopypant)
- Prioritizing top half versus bottom half of scoring group
- Never allow 3 games with the same color in a row for a player (Except in the last round)
- Doing your best not to allow imbalances of 2 or more between colors
- Never having a player play the same player twice if the number rounds is lower than half the participants
- Prioritizing playing those who you've played the least after that point
- Prioritizing higher ranks/seeds getting their color due in a section
- Prioritizing alternating colors
- Balancing distribution of byes
- Restricting how many scoring groups an individual player should job, and balancing that against splitting another group.
- Determining the best player to drop from score groups, favoring the bottom seed unless otherwise impossible to pair
etc.

Full rules are spread over more than 10 pages: handbook.fide.com/chapter/C0403

Not as trivial as it sounds

#131 #135

@Toadofsky - Thanks for the insights here

I don't want anyone to have the impression that comments here are unappreciative of the efforts that the development team and community put in and these last two months have been about exponential change in so many different ways.

As #135 says better communication strategy would help here - I mentioned the idea of a roadmap in the previous thread - some idea of where the dev community is going will focus of these discussions and help inform future direction e.g. what's planned in the next three months, current plans for 3-6 months hence, ideas beyond that

I come from a development background and understand your sentiments about good code.

I think the key thing that has changed compared to previous efforts at swiss oturnaments is that chess clubs are coming online. I don't know what proportion of the recent exponential growth in numbers derives from OTB chess club members but I can see the number of chess clubs on line in England has increased significantly (englishchessonline.org.uk/english-club-finder-map/).

In England the English Chess Federation has basically pointed clubs in the direction of LiChess and Chesscom. I went to LiChess first to set up our club because I liked the open source not-for-profit ethos and there many, many things about the site that I like - e.g. the UX is very good, arenas are fun (and I am not suggesting they disappear), we can do team battles.

I am looking at this from the perspective of how best to run our physical chess club online. As a physical chess club - we are used to the discipline of meeting on a given day and at a given time. Internally we run a round-robin club championship over the season, we run quickplay swiss tournaments. We participate in leagues against other local teams. Sometimes we have a simul display (e.g. typically once a year) and we do events as a group like 'Test your chess' where one strong player leads us through a game and we discuss together.

So my perspective on this has been driven by trying to see how much of what we have provided as a real life physical club can move online and also what new opportunities does an online presence give to us.

Swiss tournaments was just the first request that jumped out of the process...

I think there are probably enough comments from me on this for now - so I will leave it there.

If and when someting decides to take the idea forward I am happy to get involved and contribute.

So to summarize:

1) There seems to be a significant demand for swiss tournaments. First of all by players who prefer longer time controls, secondly by chess clubs that are registering en masse due to the Corona crisis.

2) It's definitely harder to implement than the arena format. Which means that efforts to implement it shouldn't take the task lightly. It needs to be well planned / organized (no implementation is preferred over badly structured code), and one needs to think carefully about various aspects, some of which have been summarized in post #141.

A reason to implement it fairly quickly may be that there is a risk of chess clubs choosing other platforms than Lichess once they find out they're not really getting what they want here.

Reasons to take a slower approach are obvious. As stated above: no implementation is better than a bad implementation.

Reasons not to implement it.... I really can't think of any obvious ones, other than a risk of time investment that doesn't quite result in a complete implementation with quality code. I'm not quite convinced by this reason, I'm sure it is at least possible to do it well. (There were very good implementations on other platforms 25 years ago.)

Am I missing something?

@Solal35
Wow, i suddenly have a lot more respect for the sites that do handle it!

Ok, I can understand if lichess servers just cannot handle it. (How much do you need to make it possible?)

Alternative: I for one, would be more than happy to accept the shortfallings of the Arena pairing system, if 2 things can be added:
1) Ability to disable streak bonus for self-created tournaments
2) Ability for delaying the pairing until a round is complete, for self-created tournaments

Is this more attainable?
Would any one else be interested?

EDIT:
hang on... Do you mean Arena pairing does not do any of these? Is it just simply random pairing?

@xK4LIBUR

"How much do you need to make it possible?"

This is an excellent question.

@thibault

Regarding post #71

"There's no way to rank 38 players with 3 games, with arena, swiss or any system. Time controls and tournament duration must be set properly."

I think the problem runs deeper than that. If you're interested in organizing a tournament with a longer time control, say 10+5, then there really is no way at all to organize it in an arena format that makes any sense. Which makes it virtually impossible to have real rapid or classical tournaments on Lichess. The best you can do is limit the number of players and make the tournament long enough to rank those players, but in that case the tournament will be won by those who treat it as a blitz tournament.

Actually using your time is counter-productive, which - to me - is the core problem with arena. It works great for speed chess, but that's the only setup in which it does work well. Not all of us are interested in speed chess. OTB clubs are typically interested in slower time controls with relatively small groups of players. You can't have rapid or classical tournaments with a system that effectively rewards not using your time.

#143 This is another problem http://xyproblem.info - not that there aren't reasons in favor of a Swiss system, but popular demand isn't a persuasive argument.

As for "communication strategy" referenced earlier... my comments are written from my perspective, not from Lichess' perspective. Reasonable minds can differ and Lichess did try Swiss once already.

On a lighter note I reference Lunduke's first presentation, my point being that what *could* be persuasive is a demonstration (i.e. the finished product) which convincingly proves 1. the code is heavily tested 2. the code can be understood & maintained. Producing such a demonstration requires development effort:
youtu.be/0pXjk46kTZU

@Toadofsky

"but popular demand isn't a persuasive argument."

Why wouldn't that be persuasive? Without popular demand, lichess wouldn't exist.

The potentially broader question would be: what's lichess' target audience? Speed chess players, or chess players in general? I think it's the latter, or at least.. it should be. And if it is: hell yeah, popular demand is a persuasive argument.

PS: this xy problem you refer to, although interesting from a theoretical point of view, is about users wanting a solution that doesn't actually solve their problem. That doesn't apply here at all. In fact, it almost feels like the opposite is happening here. Those pleading against different tournament formats saying that either there is no problem to begin with, "you say you want different formats, but really... you don't", and offering solutions that actually don't solve the problem, like trying to organize rapid and classical tournaments in an arena format and/or 'do it yourself'. In my perception: those people don't take the users who are asking for swiss tournaments seriously. They're just looking for reasons to ignore their very real problem.

Sorry, my language and possibly reasoning was unclear:

Popular demand isn't the same thing as an argument. An argument would be something more like (all of the following):
* Here is a definition of the problem
* Here is a program demonstrating a solution for that definition
* Here is a set of tests demonstrating that the program works
* Here is a set of standards demonstrating that the program and tests are written in alignment with best practices
* Here is a set of programs written the same way, with evidence that they are being maintained over time or that upgrading is possible

This is the video I was thinking of but couldn't remember at the time... Tom Scott argues that software erodes quickly:
youtu.be/BxV14h0kFs0?t=220

@Toadofsky #149 And that's exactly why this is so difficult. As team leaders, we can perfectly describe the problem. We can even describe the consequences, like walking to another site.
(To give another example as to how things escalated with our team: there are two players who hate playing without purpose against someone rated >700+ higher. So last tournament, one of them quickly resigned 4(!)/10 games after 0-2 moves to get another game. 1/ That influenced the outcome of the tournament, and 2/ luckily, Lichess accused him of sandbagging. He came to me not having a clue what they were talking about. For him, the fun has gone.)

Now, as regular chess users and possibly tournament directors, most of us won't be able to provide a program, let alone demonstrating that it works large scale, let alone proving it will be there to last. I think Sevilla has open source code and a working Swiss system, but I don't know the details. Speaking for myself, I can only contribute literally, so I'm a patron since I asked for this. And I can test as much as you would like. Probably I'm not alone. But we can only ask with a big group for Swiss or round-robin without creating it, I'm sorry.