This is definitely off-topic, because the question isn’t “how do we personally chose time settings we like” it is “should 1s per move games be allowed for anyone at all or do they pervert the ranking system”.
1 Like
On that note, I also find arguments of “I personally find 2s easier than 1s” subjective personal preference.
The question isn’t which is easier, but whether it is possible to get expected results based on rank/rating from 1s games.
When 2 people both play a lot of 1s games, and one is rated higher than the other, do they proportionately win more games? or are the results significantly closer to random?
2 Likes
The other important part of the question is “is this measuring the right skill”?
Does it translate in any way to predicting those people’s wins in other settings?
I think that the assertion is that our overall ranking does predict our performance across the wide range of “normal” settings, and the assertion of this thread is that contributions from 1s games impact this negatively.
2 Likes
Yeah, I’m fine with that clarification… I just think the answer needs to come from analysis of the ranking system’s predictive powers with and without these games, rather than subjective views of individual users.
For me, 1s and 2s are both irrelevantly fast and I would lose 100% of games, therefore I choose not to play them. For others they can play at this speed, all power to them… but we need to prove objectively whether 1s games are a net positive or a net negative to the ranking system as a whole to justify removing it IMO. I would not use my inability to play that fast as evidence that nobody can.
1 Like
It’s a tough one though.
The problem is that you are saying “unless we measure this, in this specific way, we can’t make a change”.
But the way in which you propose to measure it sounds intractable.
Faced with a plausible hypothesis
“Based on my experience, I think it likely that 1s games are meaningless for rank, and we are damaging ourselves allowing it”
and an intractable proposal for measuring whether the hypothesis is true, what should we do?
One answer is “see if there is a concensus of judgement that actually anyone with experience will agree that 1s is not about Go skill, it’s about twitch”
If this, or a similar concensus could be reached, then the intractable experiment is no longer a blocker.
My experience (which is limited to only one completed game, I don’t think 9x9s should be counted here) suggests to me that the results are not comparable to those obtained with other time settings.
I don’t know how stone defender normally plays but I’m pretty sure I wouldn’t have won by 135.5 points in a game with more time.
In my opinion, how well you are trained to play fast matters a lot, maybe more than the actual rank.
I would like to volunteer if someone wants to do some tests 
1 Like
I hadn’t seen an element of “external validation” in this discussion?
The quest here is for internal consistency and meaningfulness I think.
My understanding is that we want and expect our ranks to predict the outcome of games with other ranked players, so we can do good matchmaking.
The assertion of this thread is that allowing 1s games to be ranked messes this up: people who are good at 1s games (it is asserted) are not at all well correlated with people who are good at other kinds of game.
Nothing about comparison with external ranking systems …
4 Likes
Whatever the outcome, restarts should be strictly prohibited in such settings, as most players cannot react fast enough to such an unexpected event. Even if they survive initially, they may end up in a pure click war—restart, pass, pass, restart, pass, pass—until someone’s finger wears out.
Perhaps the system should disallow such sequences across the board. That would take care of the recurring problem with the cheaters who stall by infinite restarting.
3 Likes
Sorry, but don’t agree with you.
Rather early in thread
I answered that question for myself by giving my preference with an example.
Maybe not the greatest and most important contribution to this thread, but I feel that I was justified to do so.
3 Likes
Is there a way to find all the completed games (excluding timeouts/cancelled ones) with 1 second per move and a negligible main time(or < 2.5 min. absolute time, probably, but excluding also resignation in this case) played so far? I know only the one I played, but if there are a lot (or enough), it can be enough to look at them to see if the rank in slower games correlates sufficiently with the result of the 1s/move games, I think.
3 Likes
1s seems like a fairly arbitrary boundary to me.
On which grounds was it even decided to include the 1s option, but not for example 0.5s, or 10 milliseconds? And why can’t I select less than 5s as initial time?
I suspect that the various lists of options were mainly determined by picking some nice round numbers that seemed about right, without actual analysis to validate those.
Why would a strong burden of proof be needed to exclude an option, while this was likely not needed to include the option in the first place?
4 Likes
Of course there are those of us who believe 1s is the One true time setting, and believe that all ratings that do not sufficiently correlate to the rank on 1s should be unranked…
5 Likes
I looked at the last 10 won games by the player mentioned in the first post. All these games were won by time while he was in a losing or even position.
4 Likes
This almost seems a self explanatory point if one already assumes that the game is being played too fast to read/think/evaluate.
That said I too wonder if one tried to do a similar analysis of ultra bullet games would one not find many games won on time, possibly from losing or even positions? Is that really a valid reason to discount a fast time setting?
Like it was already mentioned
I would be quite surprised to see resignations, it’s probably even impossible with the confirmation box. So it will likely end more often than not in a timeout unless it can reach scoring.
In any case, maybe it would be better to separate out some extreme settings into another category, maybe it would be better to make certain games unranked. I’m not sure though, and I’m not sure of the best way to come to such a decision?
Sitewide poll? Just ask anoek what he’d like, as I’m sure he’d need to agree on it?
Should one host a big sitewide tournament with such settings in order to gather data?
2 Likes
I think people should be allowed to have a rank for their 1s games. Why not? Someone else not liking it or feeling like it is too fast to play a reasonable game is not a good reason. It just shouldn’t be the same rank as for let’s say 1day per move games, since completely different skills are needed. It’s like putting your math and your english marks together in school. You should have marks for both - but no one would think of combining them both into one. 
3 Likes
I expect there to be more crossover of skill between different time settings and board sizes. I expect it to more similar to having a single mark for science versus having separate marks for biology, chemistry and physics (which I suppose some schools do and some don’t).
1 Like
It’s probably not difficult to obtain that form the DB dump someone published weeks ago.
I would be nice to do an statistic analysis on the games, similar to the one @anoek did to decide using overall ranks instead of specific ranks for matchmaking.
2 Likes
We had this question before.
It’s a question of judgement. If you aren’t willing to make a judgement, there are many decisions you will not be able to make, including this one.
Someone decided, based on their judgement, that 1s was the lower limit.
This thread questions that decision, and asks for a sensible judgement to be made based on the evidence available. The evidence available, and the arguments presented with it, would be the “grounds for making a change”.
1 Like
This sheet has restricted access.
Since Bugcat is already necroing this thread anyway, I thought it prudent to clarify - are we talking about removing 1s ranked altogether? What about with sufficient main time? As KoBa points out, 3m+1s (for 9x9) is standard on GoQuest, and seems perfectly reasonable for rated games. For that matter, I’d imagine that after a little practice, even 30s+1s Fischer or shorter would be reasonable for 9x9 rated games
3 Likes