Revisiting Automatch Time Settings: Data-Backed Proposal for New Automatch Settings on OGS

Feedback Suggestions
167

For ease of calculation, I take a typical 19x19 game to take 240 moves to scoring (which is maybe a 5-10% underestimate from reality?)

For OTB games I’d say that

  • a blitz game takes about 20 minutes max, which is about 5 seconds per move.

  • a rapid game takes about 1 hour max, which is about 15 seconds per move.

  • a somewhat casual amateur tournament game takes about 2.5 hours max, which is about 37.5 seconds per move.

But probably people prefer a bit faster games online, and also playing at a more regular pace without excessively long periods of waiting for the opponent’s move.

So for online games, I suppose a basic time of about 5% of the max game duration is acceptable, while it’s undesirable to enable players to bank more than about 10% of the max game duration.

So I’d propose something like:

  • blitz 1m + 6s (cap 2m)
  • rapid 2m + 12s (cap 4m)
  • live 5m + 30s (cap 10m)

Edit: changed to a middle ground for blitz and rapid increments

5 Likes
168

Preferably caps would not exist and we would use pure Fischer time.

1 Like
169

In that table, there are various percentiles. I think the “90th” column will be pretty close in spirit to a “max”, excepting the most extreme outliers.

3 Likes
170

That would also be my preference in most non-correspondence situations. I intended my statement as a minimum of what I’d prefer, rather than what I’d prefer in a perfect world

2 Likes
171
  1. I think normal 5m+3x30s is the best, and it should be the default value.
  2. Blitz 1m+1x10s.
  3. I don’t think 20m+3x30s need to be reserved, no one will play it at all. But if you really want to keep this option, then at least 5m+3x30s should be the default value.
  4. I am opposed to using fisher in live games. In live games you should only use byo-yomi, remove fisher.
  5. Correspondence games should only use fisher, delete byo-yomi.
  6. Delete handicap. If it must be retained, the default should be disabled.

Regarding 1:
I never played live games at OGS, only correspondence games.
If I’m going to play a live game, I’ll play it on FOX.
Time control is a factor, but there’s also the issue of matchmaking time and ranking mechanics.
Due to the insufficient number of OGS users, it is easy to encounter the same players.
Moreover, it is difficult to meet players of the same or similar rank.
The ranking mechanism will also have very large fluctuations. In just a few games, the rank may rise or fall by several kyu.
These are all reasons why I don’t want to play live games at OGS.
Of course, I also don’t like open challenges, invitations, and acceptances.

Regarding 2:
The shortest time setting in FOX is 15s, but I don’t think it’s fast enough.
I think if there’s one real reason why I can play live games at OGS, it’s that OGS can set game times in single digit seconds.

Regarding 3:
I think most players don’t need this kind of thinking time, except for dan players.
For players of this level, they will probably want to use open challenges, invitations, and acceptance methods instead of automatic matching.

Regarding 4:
To be precise, what I object to is the use of fisher for online live games.
It’s fine to use fisher in real life, but it’s easy to encounter evaders in online games.
The evader mentioned here does not interrupt the connection, he just needs to leave the computer without closing the game. Maybe he went out to eat, or he opened Youtube, who knows?
byo-yomi I only have to wait up to 90 seconds. And generally speaking, when the opponent escapes, there will only be one 30 seconds left.
If using fisher, how long do I have to wait for my opponent to do this? 5 minutes? 10 minutes? 15 minutes?
Please, I don’t want to play the patience game.

About 5:
I found that no one seemed to mention correspondence in this post. I was a little worried that the developers had forgotten this matter (if 4 was adopted), so I specifically mentioned it.

About 6:
I found no one talking about the handicap issue in this thread.
I strongly recommend against using handicap, I really don’t like it and I’m sure most people don’t like it either.
I’m not sure if the current handicap default is enabled or disabled, I was under the impression it was enabled, which caused quite a bit of confusion.
Many people will just cancel the game, this type of matchmaking is simply not efficient.
At the same time, I also suggest that the handicap default value for custom games should be set to None instead of Automatic.

1 Like
173

This heavily depends on your level and on board size though, I think most people have no issue (though it’s self-selecting in a way).
I’m surprised you have trouble at your level, but perhaps that’s because many DDK players mostly focus on 9x9.

The ranking mechanism will also have very large fluctuations. In just a few games, the rank may rise or fall by several kyu.

I had a look at your profile and you still have a significant uncertainty to your rank (+/- 1.7k), so it varies more by design than it does for other players with a more settled rank. That’s supposed to be a good thing!

3 Likes
175

IMG_2049
IMG_20492388×1668 484 KB

for those interested you can play around with this. Couple issues though, requires tsumego dragon account and an ogs account and you have to allow popups for the game to automatically open when the game starts.

This creates a custom challenge with the click of a button. It doesn’t use automatch because…this entire forum post…lol too hard to find the best times. Uses Chinese rules and fischer time.

1 Like
176

Looks great! Easy to follow and convenient to proceed to the game. The only two things I would improve are:

  • decrease the number of game options (two types of games for both 9x9 and 13x13, and three types of games for 19x19) to not dillute the player pool
  • show the number of people that you could match against on the buttons; either number of players with similar rank or number of players who are looking for that game type right now
177

That’s too fast.

And I’m not buying the argument that “open challenges are an option”. I’ve been playing on lichess.org (and on freechess.org and similar before) for long enough that I know exactly what will happen: the option will be there in the UI but approximately nobody will use it.

IMO (and I know this is not the most prevalent opinion) this is a situation where site admins should take a small risk, and lead instead of follow. I.e. give a small bump to those of us that prefer a slow game, rather than leaving it up to a 2 wolves vs. 1 sheep vote.

2 Likes
178

Are you saying that we should make the default something that is favoured by a minority? But

What would you suggestion be? Do you like any of the others already put forward?

1 Like
179

I think 20+20 or maybe 20+30 for classical would make me try it. 20+20 was a common proposal in the thread. And I would make the cap much smaller, maybe just the initial 20 min. I feel that the whole purpose for increments should not be allowing to bank time, but rather preventing mad time scrambles.

2 Likes
180

This seems significantly longer than current “normal” setting, no? To each their own, and I may be wrong, but I doubt such a slow setting would be very popular on automatch.

181

I wrote “would make me try it” precisely because I don’t play live now, the default settings being too fast for me.

Maybe another way to look at it is: you can either try to attract all the players who now enjoy playing on the other live sites, or you can try to attract us turtles :stuck_out_tongue:

But also note that I am not alone in the thread proposing these settings. And I don’t expect that “classical” would be the most popular choice.

–
Ian

1 Like
182

One thing FICS did right was allowing you to have multiple open chanllenges (“seeks” in their lingo) at the same time, and then when you got picked up on one it auto-cancelled the rest. So you could propose both a “popular” and “unpopular” time controls at once, and you didn’t exclude yourself from the pool either way.

lichess unfortunately didn’t replicate that (at least not for free accounts), but at least both the “quick” and “custom” challenges are both visible in their “lobby” interface, so again there is some degree of flexibility – you can make an “unpopular” proposal but then pick a different one from the lobby if you’re too long to get matched.

–
Ian

1 Like
183

Under the hood that’s what we do too for the various board size settings, though I’m getting the impression from some that they’d prefer singular buttons for simplicity. How did the FICS show those options to the user?

1 Like
184

I think we should keep the multi-queue that we already have as it follows the theme of maximising player pool, and as per the polls, introduce a rapid time option and change all auto-match to just Fischer.

185

I’m not so sure about this. In my experience, when I search for all 3 board sizes, I end up with 9x9 about 99% of the time.

I strongly agree with @spaceraven’s point. Most players already know what board size they want to play, so the multi-size option seems unnecessary. Plus, it complicates the UI design. It’s hard to clearly show which time settings you’ll be playing with when multiple board sizes are selected - you can’t fit all that info on a single button.

We should simplify things and focus on making the interface more user-friendly. I think maybe we can look up more into the idea of having multiple open automatch searches. But I’m not sure how to make it beyond obvious to the user with this approach that the game is being searched for. I have a few blurry ideas, though

4 Likes
186

FICS heyday was in the 1990s, so the main UI was a terminal window. You typed a command like maybe “ownseek” and it spit back a line each with the parameters of each seek. Graphical shells like xboard and winboard paired that with a graphical chessboard, but that was it.

You could even play a game entirely within the terminal interface. By default, the server printed an ASCII art picture of the board after each move. And for a few years, before Linux X11 was solid, that’s what I did :stuck_out_tongue:

–
Ian

4 Likes
187

Can’t we agree on keeping things separate? Because of the byo vs Fischer debate, this thread is leading to no change at all, when either one would be tremendously healthy for the platform.

Just don’t change to Fischer for now, there’ll be better data later once usage of fewer and better defaults consolidates.

1 Like
188

I believe the Byo vs. Fischer debate has (hopefully) passed us. The more recent discussion is “what are the equivalent time parameters for Blitz/Rapid/Live Fischer?” (See anoek’s comment) which is important either way because the Automatcher currently supports both Byo and Fischer.

Was a concensus reached on that? If not, maybe it would help if someone plotted all the proposed settings (x=increment, y=main or max) and then we can throw a dart at the graph :smile:

4 Likes