using the almost free ladder calls to decide where to spend the expensive match calls is a really neat trick, 38 to 62 useful calls out of 100 is a clear win. now that you re-read every 2 hours instead of 30 minutes, do you lose many players who play a few games and then drop division in between?
Thanks for the question! I just checked the data as it's a very interesting point. When we re-read the ladder in 2 hours - about 84% players are still on the same page and 16% are moved to the other division. We're storing their games per-player, not per-page, so they can pop up on other pages we read, but this depends on luck as we're not reading all pages. But even if we're loosing those players - we still don't loose games because after each pass most servers still have 15–40k players waiting with new games, far more than our match API limit can fetch, so losing some just means picking someone else
84% still on the same page after 2 hours is higher than i'd have guessed, so most of a full re-read is just confirming nothing moved. are the 16% that move mostly near the top of the ladder or spread all over?
Praneet Brar
Building cool things like launch platforms
using the almost free ladder calls to decide where to spend the expensive match calls is a really neat trick, 38 to 62 useful calls out of 100 is a clear win. now that you re-read every 2 hours instead of 30 minutes, do you lose many players who play a few games and then drop division in between?