Option 1: count all timeout, but the community doesn’t prefer it, so option 2
Option 2: has a bot like Leelazero checking the win rate if it’s, say over 80% of 1 side, the winning side is counted as a win. I am not sure of handicap games though because it may be 90% for the lower kyu but he still have good chance to lose it.
Interesting - we might almost be verging on being able to ask a bot, automatically, what the outcome should have been!
I still honestly think it is not really about intentiaonal cheaters, but rather about people feeling cheated. Still a problem nonetheless, yes, just for the precision’s sake:)
Possible, I guess? As a “more acceptable” compromise. I was thinking about that too at one point (though I was thinking 2 or 3 at max 
), I will try asking the devs.
I do not think that is realistically aplicable. Esp. correspondence players take longer to rank up and are potentially still getting better, so it is far from unthinkable that 50/50 would not be the case. And it would be even more confusing for players to explain and hard to keep track of.
Honestly unthinkable for me timeouts are too common for that. And if it were up to me I would still probably mark most of them as undecided (also thought about lower on). We are however thinking about a feature where reported games were paused until decided. Not yet fully thought out, but guess that could help with this issue too, if people managed to report before the “surely won” game timed out.
The auto vacation I really like as well (although not mostly as a solution to these problems). I think it is even somewhere on the todo list, just not really too urgent at the moment.
I am not sure if you are not largely underestimating the hardware requirements (and thus potential intentional attacks) for running lizzie on potentially multiple games at once (esp. keeping in mind that OGS is a free service).
Similarily (although this is partly just for curiosity’s sake) I am also not sure relying on bots is that good of an idea. Bots think differently about the winrate than us. For them it can be the one magic variation that is 90% win, but humans are not that precise. In the later parts of games maybe, but saying with any kind of certainty “white would have won this game” even 100 moves in, I would have felt like lying I find it more precise to say we can’t know who would have won.
Just to ilustrate what I mean. This is from a rather higher ranked game. (about my level) and you can see the winrate jumping from below 20% to above 80% in like four moves. We are just not anywhere close to bot level to be able to have it call games for us (in my amateur opinion of course). Bots just see the winning chance differently as they are always considering the best imaginable line of play, which is unachievable by us.
Perhaps it is silly of me to wade into this mess, but here are my two cents.
First - Do NOT use bots to judge results of games that are timed out. This is about the worst of all possible solutions. Human beings play games together, and unless you can arrange to make a 15kyu bot play a 10kyu bot (and play a large number of games from the resulting position) then the timed out game from a random 15 and 10 kyu pair might go the wrong way. This is such an unwieldy solution as to be impractical - and also I think unjust.
Second - Since all movements are tracked, for the case of isolated strategic timeouts, why not let an initial timed out game count as a loss for the timed out player, and then count subsequent timed out games as losses if, after the first timed out game and BEFORE the second, there is ANY interaction with OGS (login or movement made in any game).
Third - The assertion by Wulfenia and others is that people are strategically timing out of a large number of lost games somehow to avoid their rating falling. I sincerely doubt this happens very frequently - if it does, and if the pattern of behavior can be investigated when suspected and shown to be likely - then manually enter all the losses and wins accordingly. Perhaps in the weeks of discussion I have missed that this isn’t possible (maybe it is). But really - again - I doubt that this happens often at all (and honestly I would just accept it as a small flaw in the system). I think when a person loses a large number of consecutive games by time out, it is accidental - it has happened to me several times when I’ve been too busy at work and forgotten some games… and because this is the case the vast majority of the time when multiple games are timed out, the present system is MUCH preferable. To lose rank on every timed out game would be very harsh on those who simply forget correspondence games that are close on time.
Fourth- as an alternative to the Third - because I do think this can easily be shown if it’s happening - make this an investigation that is instigated by a person who won a timed out game and who suspects this happened. If it is shown that a person cheated by intentional mass timeouts, then their account can be suspended - or they can be given a manual deduction of rank for all games lost in this way. If they do it again, ban them permanently.
This isn’t rocket science… nor do I have any belief that there are very many of our members who are being so careful about timeouts that they can somehow arrange consecutive timeouts without any interaction with the system, to avoid losses.
And finally - really… should our OGS ranking really mean so much to us to make a big deal out of what may be a tiny handful of missed wins in hundreds of games?
There are proposals for this sort of thing. I’d love to see that happen (though perhaps not exactly as you’ve stated - I think it needs to be more slated towards counting them as losses if you come back within a certain time after the last timeout).
That is not currently possible. It could conceivably be made possible. But why have this manual messing around and reporting load and moderating load when #1 does the job.
Why? If you day dream when you are playing live and time out, you lose. The same contract applies to correspondence games.
We already do that. It is a complete pain in the butt, and rather ineffective, due to the workload involved.
It tends to mean that much to people who have invested effort into specific games only to have them annulled.
It’s possibly not only about the rank itself but the natural justice - the feeling that cheating should not be allowed.
GaJ
This topic was automatically closed 91 days after the last reply. New replies are no longer allowed.