The automatic ban system in place is crap. It warns everyone consistently, however it doesn't detour the guilty parties. And it has potential to ban innocent people. I get a bunch of warnings due to losing my train of thought and spend 10 seconds to try to figure out that I can't figure it out. Yeah I have not been banned really except for I think I did once for resigning early due to playing at work on my break and I had to leave. I accepted that.. However I have been watching this and it really hasn't stopped abusers from disconnecting in lost positions. Which is the point of the auto system correct? Why not put in sharper rules like detecting people that commit the actual crimes and do more severe things like a 6 month rating ban or something. I had one just now who dropped his queen like on move 5 and obviously left.
It's just a little annoying that the abuse still exists, and innocent people feel like they can get sanctioned for things outside their range of control. End my vent I guess. Probably won't go past here.
Thanks for listening.
> Why not put in sharper rules like detecting people that commit the actual crimes and do more severe things
This is being worked on. The main challenge is distinguishing abusers from people with an unstable connection, and that's something we must get right before making harsher punishments. But that, too, is being worked on.
> It's just a little annoying that the abuse still exists
The abuse will never cease to exist, we can only mitigate it as much as we can. There are people that have collected over 30 automatic bans (and the ban duration expands the more you get banned and can go to weeks) and still didn't get bored somehow... (Note that if they were to circumvent their bans with alternate accounts, we ban them entirely. On that note, your suggestion of banning them for 6 months would only lead to them creating more accounts. We could ban those, but in the meantime, the abuse still will not cease to exist, and there exists no single technical solution that can guarantee to keep someone out of a website.)
I would like to end with a statistic that hopefully restores some faith in our current (albeit admittedly imperfect) system: if we consider all automatic bans we hand out and sum their durations, the result is **4.3 years of automatic bans per day**.
@ProgramFOXGlad to hear this is being worked on. Stronger automatic bans for players who are making a habit of abandoning losing positions is one of the most-needed enhancements on Lichess right now. This kind of behavior is growing out of control and it's damaging the atmosphere around here in a big way.
Thanks.
Well thanks for the answer. It made me feel a little better.
I am not an expert but what about investing some energy in a self-learning AI like Irwin? Which regards all the patterns of notorious quitters?
So instead of coding the criteria yourself let the AI develop the „fingerprint“.
An idea could be to have people verifying their identity with their phone-number or a google-account. That way it's going to be difficult to systematically create new accounts.
For those of you that like me have had enough of people abandoning games in lost positions I suggest you to play tournaments. Those that joins tournaments usually wants to play many games and do only very seldom abandon games.
@Sarg0n We would first have to build a solid dataset for quitters with all possible parameters, which we at the moment do not have in that format and could also not easily extract. But a neural network should also be on top of a simpler system (and not a replacement) because of the scale: having the neural network look at *all* games by *all* players would require way too much computing time. And before I want to start thinking about the practical execution of such an AI, I'd first like to get the simple system do a better job :)
@Youcandothis Google accounts can be easily newly created (and would not stop people who are already bored enough to create new Lichess accounts for abusing). Phone numbers would be efficient but would also stop a lot of fair players from joining ("why does this chess site want my phone number? ew. guess I'll play somewhere else, other sites don't need it", which is a very understandable way of thinking).
@ProgramFOX For those that don't deliberately try to abuse Lichess, I think making the account creation process a bit more difficult could make them more willing to learn resign instead of first making a new google account and then making a new lichess account.
Another idea for making the account making process a bit more difficult is to have maybe a 3 day waiting period from making the account until it's possible to use it. Of course, some people could start making accounts already before they are banned, but it would anyway be easier to learn to resign.
Ah ja. I have seen that in a German forum (Uhrform - watches form). You have to write a couple of words when registering and after 1-2 days a mod gives permission to use.
I just don't want to be adding too many mechanisms that make everything harder for our legitimate users, which is still the largest portion of our userbase. Letting people wait three days will annoy plenty of people and will indeed just lead to abusers creating accounts in advance.
Google verification would mean that we require all of our users to have a Google account. Most will, some won't, and requiring it would not be in the free software spirit of Lichess.
