The quality of the time you spend on chess matters too, not just the amount of time you play. The hard reality is that playing short time controls online won't improve your chess understanding beyond a certain level. Playing blitz makes you better at blitz, but you don't practice calculating or analyzing positions deeply in high pressure situations except over the board with the clock ticking down. Online ratings give you an approximation of your strength, but over the board chess is the real test of your mettle. Someone rated 2400 in blitz on lichess could be anywhere from 1700-2300 strength.
What kind of a weird approach to chess mastery is it, to use statistics to derive guidelines from data. The weird part: the crucial info is already lacking in the raw data itself, like: quality of the time spent, as @joddle suggests, or motivation driving the project, not to mention expertise of the opponents, state of mind of the fighters, plus many more dimensions that humans usually summarize as "raw talent". Also, some pieces of guidance a.k.a. lessons taught, can make a huge difference, just as age does.
I suggest, you annihilate the entire concept of "getting good" by some measures "automatically". And during my life, i have seen young people improving fast, while others almost failed completely. And the sick motivation, Hiraku had in his days, is really nothing to be desired at all. In a way, life put him in a spot, that was so utterly unbearable, that getting into chess looked to him like his only way out.
There is so much more to say concerning your topic, like the concept of 10.000 hours, anyone needs to put into an activity, be it professional or not, before reaching mastery - no matter what - . But even that idea is only as useful as a finger pointing at the moon. If you look at the finger, you will miss the moon!
But to increase playing strength quickly, i suggest you surround yourself with strong players and take lessons from time to time. AFAIK, that'll put you on seven league boots! (Lots of fun!)
I suggest, you annihilate the entire concept of "getting good" by some measures "automatically". And during my life, i have seen young people improving fast, while others almost failed completely. And the sick motivation, Hiraku had in his days, is really nothing to be desired at all. In a way, life put him in a spot, that was so utterly unbearable, that getting into chess looked to him like his only way out.
There is so much more to say concerning your topic, like the concept of 10.000 hours, anyone needs to put into an activity, be it professional or not, before reaching mastery - no matter what - . But even that idea is only as useful as a finger pointing at the moon. If you look at the finger, you will miss the moon!
But to increase playing strength quickly, i suggest you surround yourself with strong players and take lessons from time to time. AFAIK, that'll put you on seven league boots! (Lots of fun!)
Liked this
Interesting
Data scientist and longterm player here. I may be an outlier because after 5-6 year being stale in the 1700-1800s I jumped to 2000+ rapid in the last 2 years. Anyways I think there's something missing in this analysis: lichess had an huge surge of popularity in the last years, and I don't think a rating of -say- 1700 from 6 years ago is the same as having a 1700 rating today.
I think further studies are needed to investigate how lichess rating reflected playerbase strenght over time. But seeing how most longterm players had a decrease in rating, and discarding the hypotesis they ALL regressed in raw chess skills, it's likely that the level of lichess (and online chess in general) went up over the years, and a 1700 today is stronger than a 1700 years ago.
I think further studies are needed to investigate how lichess rating reflected playerbase strenght over time. But seeing how most longterm players had a decrease in rating, and discarding the hypotesis they ALL regressed in raw chess skills, it's likely that the level of lichess (and online chess in general) went up over the years, and a 1700 today is stronger than a 1700 years ago.
Wow, nice article. I know it's been already pointed out in the article that those results may just because of the data points collected but I do think studying along with playing a handful of games on daily basis is the way to go, well at least for me and I like this way.
Let's hope I can keep up and continue without any more breaks now. After all, all of us are not GM. Hikaru Nakamura xD
Let's hope I can keep up and continue without any more breaks now. After all, all of us are not GM. Hikaru Nakamura xD
Very nice and interesting work. I like how you think about the limitations of the available data. And I find it kind of funny that the result is, playing a few games on a more or less daily base could help to improve. Maybe beside all the limitations you already mentioned this result could be robust. I think regular playing will, in many turns, come hand in hand with doing also puzzles, watch youtube videos, read books etc. Props for using R :)
@chris-tian said in #115:
> 2000 lichess what? Classic? Or hyper bullet? Big difference. And even then...older guys are slower in bullet, but will crush you otb anyways.
2400 rapid and I've like 1420 otb and I suck at <3 min time controls
> 2000 lichess what? Classic? Or hyper bullet? Big difference. And even then...older guys are slower in bullet, but will crush you otb anyways.
2400 rapid and I've like 1420 otb and I suck at <3 min time controls
@krobbi cool to see a fellow data scientist here! i admire the work the author put into their study and i didn't want to be too nit picky about the author's analysis but there's lots of issues using online ratings to estimate strength. as you said, inflation is probably the biggest issue that needs to be accounted for. i don't think you're an outlier in that regard: i've seen my blitz and rapid ratings increase a few hundred points over the last 2 years, but i have been retired from otb chess for 2 years and haven't been making any serious attempt to improve in chess like i used to when i played competitively.
online ratings are also extremely variable as different players play different time controls under the same rating category. a 2400 rating achieved playing 5+4 blitz is different than a 2400 rating achieved playing 3+0 blitz where you can flag your opponent in losing positions. people can also select their opponents here which biases ratings. although i doubt this will happen, lichess could introduce a new rating based on automatic pairing of players at a specific time control, like the old chess server ICC has. for example, a 3+2 pool blitz rating would be interesting and would standardize blitz ratings a bit.
online ratings are also extremely variable as different players play different time controls under the same rating category. a 2400 rating achieved playing 5+4 blitz is different than a 2400 rating achieved playing 3+0 blitz where you can flag your opponent in losing positions. people can also select their opponents here which biases ratings. although i doubt this will happen, lichess could introduce a new rating based on automatic pairing of players at a specific time control, like the old chess server ICC has. for example, a 3+2 pool blitz rating would be interesting and would standardize blitz ratings a bit.
Long time depends on your mind
How much u use
How much u use
This topic has been archived and can no longer be replied to.