I think lichess should creat a new rapid category ...starting from 15 minutes up
I agree that there should be a longer rating category, as well as an hourly tournament for that category.
@dstark, @Patrizsche: can you guys provide an explanation for why my suggested ratings adjustment would take a long time to run? As I said in post #50, I'm not at all convinced it would.
@furrykef : The main bottleneck I see in your plan is that you assume implicitly the games are already sorted in chronological order (of when each game finished), or that it takes negligibly little time to do so. This is absolutely not the case in computer science.
Sorting takes O(n log n) operations, where n = 215 million games in our worst case, and the O notation here is a fancy way of saying "for large n, the worst case is that it's roughly proportional to n log n", but that factor of proportionality could be huge or tiny. And in most practical cases, it's decidedly not tiny.
As a result, your sorting step is going to take, well, a long time. Extremely simplistically and optimistically, supposing that your assumed 60 hours for O(n) is realistic, O(n log n) would be 600 hours. And that's just the sorting step.
Once you've sorted the database in chronological order, you can then start simulating the rating changes, where estimate of 60 hours might then be applicable (optimistic even; the amount of memory you need to keep track of all the players' ratings and their RDs is going to lead to some slowdown, and on top of that you need to continuously compute the players' RDs on the fly.)
So barring some clever approach to the whole problem, it'll probably not be implementable in "reasonable" time.
Sorting takes O(n log n) operations, where n = 215 million games in our worst case, and the O notation here is a fancy way of saying "for large n, the worst case is that it's roughly proportional to n log n", but that factor of proportionality could be huge or tiny. And in most practical cases, it's decidedly not tiny.
As a result, your sorting step is going to take, well, a long time. Extremely simplistically and optimistically, supposing that your assumed 60 hours for O(n) is realistic, O(n log n) would be 600 hours. And that's just the sorting step.
Once you've sorted the database in chronological order, you can then start simulating the rating changes, where estimate of 60 hours might then be applicable (optimistic even; the amount of memory you need to keep track of all the players' ratings and their RDs is going to lead to some slowdown, and on top of that you need to continuously compute the players' RDs on the fly.)
So barring some clever approach to the whole problem, it'll probably not be implementable in "reasonable" time.
+1.
Besides wishing for tournaments where everyone plays the same number of games (swiss or round robin), this has always been the 2nd biggest flaw I noticed on this site. I have supported it as a patreon for a few months and I was hoping for a change like this.
Do the admins read these threads?
Besides wishing for tournaments where everyone plays the same number of games (swiss or round robin), this has always been the 2nd biggest flaw I noticed on this site. I have supported it as a patreon for a few months and I was hoping for a change like this.
Do the admins read these threads?
They do. My guess is they want to push hard on fast time control features and not encourage things that attract longer time control players because they feel like they can't control cheating in longer time controls. It's just about the only thing I've come up with that makes sense for why they are so hostile towards long time control stuff and so gung ho for every hare-brained fast time control idea that comes along. The removal of leader-boards for correspondence and complete second class citizenry for that type of game is good evidence for how they feel about it.
Long story short: don't hold your breath waiting for lichess to implement any features that benefit long time control games.
Long story short: don't hold your breath waiting for lichess to implement any features that benefit long time control games.
+ for Long time control tournaments , yes please.
still no word on it from the dev team?
No word
+1.
This topic has been archived and can no longer be replied to.