But maybe the comparison should be with the current system?
If you are a new 1k arriving here, at the moment you start at 12k, whereas in the new system you either start at 10k (or maybe 12k as per voting results) if you pick “intermediate” so there’s no loss.
And there’s a chance that at least some 1k newcomers might be confident enough to pick “advanced” so arriving as 2-3k, which is a big improvement.
So as I see it there’s no downside and a chance of a big benefit.
And less is usually more when it comes to words, especially on mobile!
I’m not anoek but I have to say that I really like this proposal. I especially like the logical, rational and constructive way you have put it forward, taking genuine and calm account of objections, questions and arguments and evolving it while staying clear about scope at all times.
If every suggestion for change was approached like this the world would certainly be a better place!
We’d be coming full circle, but that’s not necessarily a bad thing. When we first launched we didn’t have a ? at all, it was just the rank. Then at some point the community proposed we stick a ? at the end of the rank for people who have played less than X games (I think we had both the number and ? at least, but it’s been probably a decade and my memory is fuzzy starting this morning). Then I think it was with Glicko2 we now had the concept of rating deviation and so we switched to a simple ? when the deviation was high, indicating we really didn’t have a good idea of what their rank was.
The thing I like about the ? system is that as the new player with a ?, when you play a few games you don’t see your rank adjusting by ±5 or more ranks at a time, followed by the much longer process of obtaining or losing rank when you’re deviation is a lot lower. For a beginner that doesn’t know what the ranks really mean yet, I think it just adds some noise to their understanding of ranks, and to those coming from another server, it might demoralize them abnormally if they see their rank jump to be several ranks higher or lower than they think they should be.
I can understand the appeal of an opponent to have a rough idea of what the system currently thinks their opponent is rated at, at the same time I would hate for folks to use that as a reason to decide not to play a game with their opponent because that value is expected to be pretty far off while they’re getting their initial rank settled, so while it adds information, it’s bad information.
One more possibility that comes to mind would be to use ? when the RD is really high, say > 250, then switch to 21k? when their RD is say between 120 - 250, then drop the ? when their RD is less than 120 (Note, those RD numbers are just pulled out of thin air, if we were to explore that more we’d want to probably do some analysis of what values to use).
(Starting to feel like maybe changes to ? should be in a separate thread, maybe… but…)
A follow-up: if we dropped ? entirely, what would happen to the concept of a “provisional player”?
For example, many tournaments disallow provisional players. Would that still be a thing?
(Note: the automatic site-wide tournaments don’t allow provisional players… I think this is really odd, because there are no rank restrictions, and the automatic tournaments are one of the easy ways to find games against humans…)
Oh, I assumed that was the current system (just based on anecdotal observation)… if we haven’t tried that yet, probably makes sense to try.
I also don’t understand this restriction. What’s the problem with provisional players when it’s not a handicap or McMahon tournament? You’re getting assigned to random opponents anyway.
In fact I think we should encourage new players to jump into live tournaments.
[13:12]MKho125: Hi there, is it possible to set rank for new account now?
[13:13]MKho125: I just helped my son to create an account, but this new account is set at 10k while my son is a complete beginner (I just taught him the rule)
[13:14]MKho125: currently he cannot play automatch with people of his rank, and I don’t really want to make him lose a dozen games just to get a correct rank (it might make him lost interest with the game)
[13:17]dangermousse: play the first few games for him?
I wanted to share that this has indeed been a discouraging experience for me. I’ve been playing quite a bit every day for a month (almost entirely correspondence games) and only recently feel I’m being matched fairly. It felt like an early, undeserved win by timeout led the system to overrank me for longer than necessary.
I’m a new player who, like many, bounced off go several times before it finally stuck — but each time I could feel it was special, that I would come back to it. As a result, and I think this is quite common, I already knew the rules and the very basics of shape and strategy. This compounded the frustration because I didn’t know where I should expect to level out — 15k? Nope. 20k? Nope. All the way down to 25k? Yep.
I’m tenacious, it’s fine… but imagine how many more go players there might be if the new player experience over the first week or two was the pleasant feeling of improving from 25k to 23k, rather than a rank freefall completely hiding their initial, stumbling progress.
I’m sorry you had that experience. It’s too bad - I think that very first game had an outsized effect too. Black was more than 30pt ahead, but timed out. The rating system saw this and thought surely you must be at least as good as 12k.
About 1st point: I don’t claim that it’s everyone’s view, but I personally don’t really care(cared) about loosing many games as a beginner. I think it’s part of the process of improvement which I’m ready to withstand to become better happily.
As of your proposal, don’t you think the same thing would happen as at fox when you do reverse sandbagging to get to play against stronger opponents to improve? It may happen that you’ll make “advanced” tier having no value because anyone can be one. There will no reliability in strength representation whatsoever of chosen highest rank you’ll choose or below it.
In regards to 2nd point: I personally would not play a ranked game against someone 9+ stones weaker. What’s the value of doing it anyways? I get, relatively speaking, 0,2 points for winning in case I’m the stronger side or I loose 100 rank points because of some silly overlook (maybe because I don’t treat this game seriously because of how big skill gap is) or connection loss for example, or person might be stalling in which case I’m bound to continue playing because of how much I’ll loose if I don’t continue(I would very likely resign in normal case and find another game or go about doing some other stuff). So many risks, no adequately comparable reward. Why play game you cannot win by design???
.
My personal request is to make it impossible to artificially increase your rank by grinding on weaker bots. Maybe make it that when you play against bots – against bots only, not human players – you gain 0 rank points if your rank and bot’s rank are the same. Make it so bot’s rank cannot drop or grow. In this case you’ll have to define bot’s rank manually. It’s a nuisance, but I think it might worth it.
If you plan to make self assigned ranks or proficiency categories just make them below 12k not above.
To clarify, I’m not against adding possibly to play against 9+stones stronger. Sure add it.
I’m just ust saying there’s not much of difference between gaining 0.05 rank points and losing 0.05 points and playing unranked. I would go as far as to round it up and say that they are the same.
Its possible to play ranked vs anyone now, but people still often cancel games vs [?] accounts.
If it would be possible to choose starting rank, it should be shown before “?”
[12k?] probably will be cancelled less often than [?], so less people will leave OGS before getting clear rank
Also people dislike that they can’t use rank restriction against [?] accounts. If someone created custom game with 16k-11k restriction, it makes sense to allow [12k?] accounts to accept that game and do not allow [22k?] and [2k?] to accept that game. If these new accounts would not be showed as just [?], user that created game is more likely to believe that opponent is within intended range and would not cancel.
