Does OGS have any suggestions about what to do about sandbaggers, or is it best just to ignore them?
I think new rating system solves sandbaggers to good degree
It really does. Dwyrin’s SandyBaggins account jumped from 7k to 1k. I don’t mind sandbaggers per se, but I do mind trolls. I’ve played sandbaggers in the past that kept laughing and cursing at me in the chat for my bad moves. That really pissed me off.
What is good about the other kind of sandbagger (the educated/nice one) is that they are stronger players dedicating their time to you, so you might as well try to learn something from the game! Any way, you can always report them to moderators.
How does Glicko achieve this where ELO does not?
(I’m interested in what is different about the systems that delivers this result).
Glicko2 has a measure of both rank confidence and passage of time, whereas Elo does not:
- On Elo, a brand new 10k player is the same as an old and well settled 10k player; but Glicko2 trusts the latter far more strongly.
- On Elo, a 10k player that just lost against a 15k player is the same as a 10k player with a consistent performance; but Glicko2 trusts the latter far more strongly.
- On Elo, a 10k player that plays every day is the same as a 10k player that hasn’t played for months; but Glicko2 trusts the former far more strongly.
Somehow I have a sense that this makes sense why it might flush out sandbaggers, but I can’t really see it clearly.
Could you elaborate on the effect of these things as far as sandbagging goes?
For example, as rafaelcp said, SandyBaggins is a well known and honoured sand bagger - and magically his rating jumped in the way you might expect. Why was that?
(PS - now that we are talking in the open about SandyBaggins let me say that if anyone complained about that particular sand bagger I hope that OGS would shut that complaint down in a flash - there are some sandbaggers that we all benefit from!)
In this case, I think it’s simply due to how Glicko-2 ratings increase faster than ELO with win streaks, though I’m sure the huge rating jump achieved by winning the first game (because of really high initial deviation) also plays a role.
I’m not exactly sure why that’s the case (maybe rating deviation stays relatively high until the streak breaks?), but since SandyBaggins has a perfect 23/23 win rate, even with relatively weak opponents, the account increases in rating fast.
Though, the “magical rating jump” as described might be more of a case of the Glicko-2 rating being retroactively applied to all historical games, instead of the new system being seeded with the old ratings, so the new rating consists of all rated games calculated under the new system instead of just new games if the latter were expected.