When a new user plays their first couple games they will usually lose a few hundred rating points. Losing doesn't feel good. Couple that with assigning a number to their defeat and people will be hesitant to keep playing.
I think each new account should play several games until their rating has a smaller confidence interval before they are shown that rating. That way when they see the repercussions of a loss a new player will only see a couple points lost.
Thanks for reading. I love this website and only want more people to enjoy it.
Ill try tomorrow to think about some idea(Im tired now), It seems to me your idea is accurate but It should be implemented in a different way like creating a beginner level perhaps
a bit like videogames where you have a hero that goes up levels and once that is acomplished then they have the normal rating, the trick is not to compare beginners directly with normal players so that they look stronger of what they really are or rather that they dont get discouraged so early because Its very easy to compare oneself with another number which has months or years of experience behind.
So I would do that I would create a transition phase before reaching the elo and in that phase there would be only winners like going up the hill, with no downs and then once they are good enough then you have the valleys in the graph, the ups and downs the swings.
You could implement for example something like a limit,
or like challengues, points rewards, to beat several times a 1500 player to end up the beginner phase or whatever, a minimum games played or a tactical ability puzzle and once they obtain all their points( prizes) then they start the elo system.
So that you have two systems the first one is always growing, always learning and this is very true in the beginner phase and then It starts the elo, (the swings) I mean not just in the games but in your score.
Because all you can do is change the score system, the facts are there, if one loses all the games He knows it
( begginners will have games swings but no points swings), but if instead of telling him that he is losing you only add points to him when He wins, He will not see it as losing, He will see it as waiting for a win, just my two cent.
Of course You can do what you say, put an absolute approach like the minimum number of games before the elo appears but thats not a good idea because there are some players that are strong from the beginning and to others It will require maybe much much longer, since Its not the same being a beginner in lichess than being a beginner in chess,
Of course maybe its too much work for a free site
but its not probably not that difficult, just think something like a presoftware and then a later software, bios and windows like loading
I am tired, Its just some ideas off the top of my head.
I think this is an interesting idea. I think I have a way that this would actually pair with someone else's thought a few weeks ago about how new players should somehow be indicated due to the frustration of playing against a player at 1500 and then losing a ton of points because in reality they are much stronger.
Is there a way we could have it set a player's rating to display only as 'provisional', continue to track their rating under the hood, but treat their opponent as if they have an equal rating for the sake of calculations? I feel as though perhaps this is the only major flaw of Glicko-2, in that an established 2400 rated player playing against a GM in their in their first match on the site is going to lose an absurd number of rating points as the system treats that GM as a 1500 player with low confidence. The volatility therefore rectifies some of that, in that the established 2400 isn't going to lose as many points as if they were playing an established 1500...but it will still be a huge point drop.
Granted, why the 2400 is playing a rated game against a 1500 they don't know is questionable anyway, but I wonder how such a suggestion as the one I propose would negatively impact the Glicko-2 calculations. Would it somehow break the system? I can't seem to see how.
If they get discouraged only because their rating is low (which is a consequence of them being bad at chess because they are beginners), they are not very logical and should solve a few emotional issues before being able to take a loss like a woman/man.
Rating ain't nothing but a number, babe.
what is the problem in losing rating points? rating will just equal your level of play. it is not going to stop people from playing. actually people play with very low ratings on each and every site and finding opponents that suit you is more fun for some, then playing some 2300+ player when they start playing. most people will not reach 1800+ without a lot of experience.
the question is: do people really quit because they lose rating? does lichess have not enough players? would anyone benefit from these people that supposedly drop out when losing 3 games of chess in a row against stronger opponents? after all maybe they even enjoy playing against stronger opponents - when I started out I liked to play strong players to learn new tricks. I think the idea is ok, but the benefits are questionable. I just do not see people interested in playing chess dropping out because of rating. I think the site here is very well designed, and would rather see features like an analysis module to analyse your games with the computer by trying out variations with a built in engine. i think the feature to analyse games is the biggest advantage of this site plus the ease of use in comparison to the old way of logging into some old fashioned ICC / FICS server and having to download one of 20+ interfaces etc.
I would rather see something like a beginners chess tutorial or a chess tutorial creator, so people can publish analysis with video / audio on this site to start of people then a nondisclosed rating for a few games. after all rating is not that important - your playing skill and fun is.
I think the point is that in the ELO system and in the Glicko system, when you are provisional you don't lose and gain nearly as many points as you do in the Glicko-2 system. Glicko had the Ratings Deviation system, which indicated provisional players by their extremely high RD. Eventually, yes, your rating would end up where it should be, and your RD would be extremely low. However, Glicko-2 builds this into the system itself, and so generally the first match you lose is going to drop you by anywhere from 100-200 points immediately. This can be quite shocking for a new player, and especially shocking for a new player on this site who has played other rating systems.
Does that initial shock drive people away? Doubtful. However, I think having a provisional period would not only alleviate that initial shock of watching your rating tank by hundreds of points, as well as protect other players from misleading initial ratings. It would also likely mean that many computer cheaters would never have a rating displayed because they'd be caught while still under provisional, which would give cheaters less incentive to bother since they'd never see a high rating.
Maybe it would be another option if people after registration would need to solve three easy puzzles. If you do not manage to get at least one right, you should be placed at 1200 rating max from the start. This would ensure people would get placed closer to their ranking and make the start less frustrating. Another option would be to let people choose their starting rating themselves by proving three options: beginner - amateur - club player - beginner should start at around 1200 - amateur at 1500 and club player at 1800.
Huh...that's an interesting idea Chilly. I wonder though if that would just make it easier for sandbaggers?
Some people are saying they want other features prioritized. Others want a new initial rating system or training system.
My top post is a compromise of the two. It would be very easy to implement. Simply don't show a new player their rating until it stabilizes. It could say "provisional" as static_shadow said.
i dont think it would encourage sandbaggers. sandbaggers would get exactly the same starting elo as they get now. and mods ensure sandbagging accounts will be deleted anyway.
the puzzle before registering would for one adress automated bot registration and at the same time offer beginners a starting rating that would be less frustrating. you could use very simple chess puzzles that any beginner above 1400 would solve and therefore ensure total noobs get a very noobish rating to ensure they will face noobs when playing their first games. experienced players will not object losing a few of their first games before settling around 1400-1800 rating. and good players will have enough prior online chess experience to know that winning a few games until they reach their rating at 2400 points is completely normal (and fun btw).
the provisional thing would also work, but this would only prove useful for chess newbies and not have any advantage for experienced chess players - i would for one rather see exactly how my rating develops in my first games.
i would say that at least 80% of the players here have enough chess experience to not being annoyed by losing rating at the start and for those you are concerned about, the above way would ensure a rating closer to their actual playing level.
