@Linspiring > The suggestion was other provisionals , I might be crazy but that's actually how I think it used to be done.
Ah, now you've reminded me :) The problem with that is that if two 1500? players play each other, then afterwards you still don't know how strong they are.
>Ah, now you've reminded me :) The problem with that is that if two 1500? players play each other, then afterwards you still don't know how strong they are.
Hmm good point to an extent. But I think if you still add points to the winner and take away from the loser and the provisionals keep on playing each other numerous times, as long as that pool is an average cross section of players then people will still be getting closer to their true rating. The fact that there is the double uncertainty of both players, simply means the confidence interval shrinks at a slower rate. So the upshot would be the provisionals have to play each other a few extra games than they would if they got to play normal accounts.
<Comment deleted by user>
Horrible suggestion, the computer only plays casual and doesn't have meaningful rating numbers
I find the Glicko system is good but is a bit too extreme on the first few games. If you win your first 5 games you start as 2400 + and if you lose your first 5 games you start as a - 1100. Even if you are good it can take a very long time to build your rating after the provisional rating if you have a bad start.
It doesn't take that long if you're really that good and have a bad start. If you're good, it's somewhat unlikely to start with 5 losses in a row.
<Comment deleted by user>
How do you want to estimate someone's true rating based on just two games?
hfggj OK NO
perdon tenia el bloq mayus
como decia
baja
no hagan caso a lo primero que escribi
<Comment deleted by user>
