Adding Rapid and Slow rating categories. - Please vote !

I think splitting classic into rapid and standard is enough and also necessary. Leaving it in present shape mixes rapid and classic games in one rating zone. Creating too many ranges would cause problems with establish reliable ratings in each of them (how many long-term games can we play monthly ?).

@chgreg , I agree that with too many ratings, it becomes harder to get established ratings in all of them, especially for a 75m+ or 90m+ category. But the new ultrabullet did not create a problem for you, right? Neither you had to worry about getting a correspondence rating, right?

Then I was thinking here... Why are some people worrying about the creation of a very long time control category if they don't wish to play such long games ? Well I think it is the thought that they feel like they would not have an "official" rating if they didn't play in those longer ranges. I mean people can easily dismiss ultrabullet and correspondence as "chess variants", or non-standard rules, but feel uncomfortable about leaving slow/classical ratings blanked, as those would seem to be the "standard" chess category. That's one reason why I didn't use the "Standard" name in my voting poll. It would also be confusing since we already use "Standard" word to distinguish regular chess rules from our variants. But yet I understand the way you feel and the voting poll reflects it.

So by now it seems like we'll have to introduce only the Rapid category and observe how things evolve.

@EvilChess , I don't play bullet and I'm not interested in ultra/hyper/whatever bullet rating, but... I guess it's no problem to play 20 bullet games during 1 hour and 100-200 games per a day (or weekend). The same amount of games in classical would take weeks or even many months. One of proposals in the poll has the splitting point at 75 min, so for example LoneWolf 30+30 would become Slow games and Lichess45+45 would became Classical. I am interested in long games, but I can't see the sense of splitting them into 2 or more ranges.

@Linnemann , the fact that humans can't grasp more than 4 measures *at the same time* does not mean that the entire world must be limited to 4 measures. After all we already have 5 measures, if you consider Ultrabullet/Bullet/Blitz/Classical/Correspondence. Not to mention the variants.

But you have a valid point and I think the solution is to create some sort of "grouped ratings". So anything below 3 min would be grouped into your Bullet rating, but yet each game would be classified under one of it's 3 sub-categories named ultra, hyper and regular bullets. Similarly, Active is composed of Blitz, Semi-rapid and Rapid. And long is composed of Slow, Classical and Correspondence. By the end, we would have a just 4 main ratings, Bullet, Active, Long and Variants. Each one with it's many sub-ratings. Check this out:



The technical question is how to calculate the grouped rating. One idea is an pondered average based on players activeness in each rating category.

Going even further, each variant may have it's own sub-division. I know we don't have a large enough pool to divide in so many sub-categories, but it makes sense separate 30 sec games from correspondence games at least.

Looking at your diagram, Ultrabullet, Bullet, Blitz, Rapid, Classical and Correspondence is enough. In other words, split slow chess into rapid and classical. This also resembles the categories known from FIDE. But nine different ratings is too much.

Meanwhile im not even sure that it is necessary to split slow chess into rapid and classical, because 15 minutes on the internet is another time dimension than in OTB chess. It feels much more like classical chess. The current system is not that bad.

Dont forget the meaning of those ratings is to resemble the different styles of chess. And in my opinion only Ultra Bullet, Bullet, Blitz, long chess and correpondence can clearly be separated. However i have not played any 45 minute game here. This may be another rating category on its own.

@Linnemann , except for two 15 min games, you never play any longer than 10+0, which is understandable since you have a very fast brain (I can tell by your ratings). But for normal brains 15 min is still blunder zone, were a sacrifice may be worth only to add time pressure. While only in very long games like 45+45 we can play with max precision. And the only reason I don't play 45+45 online is because there isn't a category for that, so my efforts will be washed away in a glimpse.

I was thinking here, to make things simpler and not reinvent the wheel, maybe the grouped rating could be computed with Gliko2 too. In such case, each game would compute for each player two different ratings. Result will be acceptable.

In variants we are mixing up all time controls, even mixing ultrabullet with correspondence, so we can mix 30 min with correspondence too, and at the same time calculate separately 30-75, 75-300 and correspondence.

Anyone looking at the user's profile won't even have to look at the sub-categories, because the grouped range would already display a precise Gliko2 rating for the entire range.

The more i think the more i'm fine with the most clicked answer so far (8 mins <= Rapid < 30 mins). That would give me a large pool of opponents for time controls i want to play (up to 15+15). Yesterday i got no opponent within five minutes for 15+15, so i accepted an lobby offer with 15+10.

When 15+15 would be classical, i would play that and something like 10+5 to get a rapid rating. :)

Other people may play one or two games weekly with time controls like 30+30, 60+30, 45+45 when they have enough time for playing. But when they are shorter in time they maybe also want to play one or two lazy 10+5 instead of a couple of meaningless 3+0 blitz games. The two different rating pools make much sense for them.

I also think that the pool for players for a classical rating with 75 or even 90 minutes could be too small. But only 494 people played racing kings this week, compared to over 80.000 blitz and over 70.000 standard players: de.lichess.org/stat/rating/distribution/racingKings

Maybe that's no big issue. If such a rating categorie comes, i would simply have no rating there, like ultra-bullet, atomic, and so on.

@Karpfenkopf , yes that is the idea !

Voting poll is pointing towards a 30+ min split, confirming the result of previous voting polls. 78 votes is nothing compared to thousand of votes a blog link gets, but enough for a statistical sample.

Currently 30 min games aren't too popular, but I'm absolutely sure that this will change. Once a 30+ category is created many will be encouraged to play longer games. And once again Lichess will be on the edge of innovation.

When is the update expected with this new time controls?