Sandbagging is a big problem at low rank. Letting people choose a rank on a browser-based server where making an account is so easy… Sandbaggers will be making new accounts everyday. Banning them will have almost 0 effect. Consider the hoops they go through to lower their rank (throwing games on move 3, playing against bots… and other detectable methods they use); they won’t have to do any of those things anymore. As soon as they are unhappy with their high rank, they’ll be able to make a new account with a new rank. Letting people choose their rank might lead to fewer complaints on the surface or fool beginners into thinking they can get matched with people their own level, but the number of players who get sandbagged and experience unpleasant games would increase and doing anything about such cases would become more difficult.
OP’s issue… You can currently set a range of rank you want to play against. OGS lets you play -9/+9 rank players in ranked games and choose any range within those limits when you’re creating a game. Even if the server assigns you 13k by default, in theory, you should be able to limit the range of opponents to between 20k and 22k for the first game, for example and stronger or weaker players won’t be able to join your game… Then, if you lose the first game and derank to 15k, you can set the rank 22k to 24k. This lets you start playing complete beginners in 3 to 5 games. I’m not sure why this didn’t work for you. Perhaps I should make a new account and test it. If anyone knows how this system works for provisional accounts, clarification would be welcome.
However, even if this doesn’t work, I would argue against the necessity of a complete beginner being able to dive into ranked games… Someone who just learned the rules of go could be satisfied with being able to create an unranked game. “Ranked”, as can be told by its name, is about competetion. Losing 3-4 games shouldn’t be that much of a bother to someone who wants to go into that battlefield, especially if it ensures a system with more integrity overall.
On the one hand if people played normally possibly, because the high deviation (I don’t think the volatility changes too much in practice) is supposed to mean that settled players ranks shouldn’t be affected too much, while the new player corrects fast right? That said, people don’t always play normally, sometimes they do weird things like pass in the middle of the game or fiddle with scoring, possibly they’re a new player or they’re a troll. It likely causes a mess, like when an experienced player resigns against a beginner to just start a new game. I think all those things need to be manually annulled but it’s not clear to me it won’t have knock on effects.
The big downside as you mention is it’s easy for me to keep creating 25kyu accounts to thrash beginners which won’t make their life any better.
I obviously don’t have the info on that, so it’d be better to hear from a mod. However, in the last few days there’s even 5d accounts causing trouble, messing with the scoring phase and just passing for 100s of moves, so I wouldn’t rule out that there’s “enough” trolls with nothing better to do.
I think this was the proposal linked to on the github issue.
Letting sandbaggers and beginners both start at 10k/13k (default initial rating) results in the same poor experience for beginners as letting them both start at 30k (self-declared rating). But allowing beginners to start at 30k does at least give a better experience to beginners who happen to get real beginner opponents (besides 30k sandbaggers).
Previously, new players did set their own rank. Glicko, as I understand it, was instituted primarily to get away from that because the old system generated a huge number of ‘please reset my rank’ requests, which became untenable.
Rank resets were very common under the old system (pre-Glicko), from what I have been told, which is not surprising since people were frequently setting an inaccurate rank for themselves. Ranks can’t be reset under Glicko–again, so I have been told.
BTW: the new provisional rank is 6k since the recent rank revision.
No you are not asking a estimated rank. You ask some facts and then the system assign a starting rank.
A player will have less chance to make a mistake by saying he s a beginner or not as by estimating his ranking.
Thanks for speaking to the sandbagging issue. However a small correction is in order. Alt sandbagging, which is overwhelmingly the most common form, has always happened, happens now, and will always happen unless account creation is limited. Even that isn’t an ideal solution in the era of VPNs, but it would help. Starting players at 25k or 30k gives alt sandbaggers a vast range to play in before they have to create another account. Creating and ranking up accounts becomes obvious much more quickly under the current system.
At least there is an explanation for not being able to set a custom ranked game with less than 15k limit.
And if so, this is surely a sign that humble ranks are not working as they should or at all.
Surely if the humble rank system were working this would help with at least part of the problem since custom games could be created with an appropriate rank range (or close) and auto match would presumably make somewhat better matches. Not ideal but better than now I think.
Here’s an idea. Perhaps provisional players could be given the exclusive privilege of restricting rank by a wider range, say 19 stones. Then they could play 25k beginners. This prioritizes the welfare of beginners over the sandbagging problem, but that seems to be the consensus here.
Yes, I feel the current implementation deteriorates the user experience of all beginners on OGS to mitigate the sandbagging problem somewhat.
Instead of coming up with patches to work around a provisional rating that is much too high for beginners, I feel that the proper solution is to improve that provisonal rating for beginners one way or another and try to find other solutions for the sandbagging problem.
In general problem solving, I think distinct problems like these (reduce sandbagging vs allowing beginners to find suitable opponents) need to be investigated and solved separately.
Once you try to solve multiple distinct problems with one solution, it introduces a coupling that makes it difficult to optimize the solution and you often end up with a solution that is poor for each problem you’re trying to solve.
In some lucky situation, you might discover a solution that fixes multiple problems at once, but it should not be a premisse that such an ideal solution even exists.