Time settings?

[Kaya.gs]

As it was stated before, free games have much wider time settings, so its not preventing people from playing long games.

Its a given that there has to be some sort of upper-limit.

In the same way a 1 minute absolute time should not be rated, a 9999 absolute time rated game should not get started. Its guaranteed to end in an upset game.

I disagree, except in a vague theoretical sense. It is up to individuals to decide what is appropriate and what is not. No-one has a problem with this in chess (where time settings like 1 minute absolute time are not uncommon), and I don't see why it's up to the server to restrict it.

Again, I have seen plenty of evidence to the contrary, but not evidence supporting this view.

I can assure you no man alive ever played a 9999 absolute time live rated game. He would have to be awake for 7 days straight without sleeping.

A (midly)less extreme case is 38x38 games.
Both players could agree to play such a game in a rated fashion, but i havent seen anyone finishing one of those games ever yet.

Yahoo with live player replacement,FoD with absolutes, and KGS with its escaper system, they all require agreement. Yet the upsets still happen.
As i stated, agreement does not mean precision or a good experience.

Also its to note that we will support "turn-based"/deferred go eventually, and those will be rated. In principle they can be played as live as any other game, and wait as long as the players want. It wouldn't be a problem in that case, because neither player is forced to be online and present all the time, as with regular live matches.

Are the same ratings used for both turn based and real time?

Maybe. Depends on popular demand, and the player base.

Having a middling allowed time also does not guarantee anything, which is precisely my point of the objection. 45 mins + 40/5 is not a particularly slow game, if we are to have a floor than I definitely feel this is too low. It's _probably_ not enough to stop me playing on the server, but it is a fairly strong point against playing on it as my regular place for Go. The experience is ruined in the situation you outline, regardless of whether the time control is 15 minutes, 2 minutes, or 120 minutes in these cases. Sure, it may be a longer wait to have it ruined in longer time controls (although if this is your main argument, you lose the justification for the _bottom_ floor on time controls), but this is something the user has already taken on willingly - I'm not convinced anyone will appreciate this restriction being forced upon them.

Concerning your comparison to Real life, you have to take into account that the internet has anonymity. I have never seen anyone in real life escape, yet is one of those issues all servers have to deal with.
The internet is also less stable. People lose connections, get distracted, have technical issues, lose power.

A single user that resorts to this tactic will do so for every single game on long time settings.

It happens. It has happened to me on 15 minute games. On all servers I ever played in. Some servers have even taken measures to prevent this precise case.

And it will happen simply because the other user has nothing to lose by waiting instead of resigning, he can just leave the computer on. His inaction is prized with opportunity.

The motivation for bottom limits is that a game with a very low time limit produce very imprecise results. A 9d can easily lose to a 10k even game in a 2 minute absolute time game.
I dont think anyone advocates for those games to be rated.

Actually, if the main issue is that "This happens because there is no game adjourning.", why not added a feature that allow to adjourn a game?

There is a delicate balance here between escape control and playability.

KGS gives the power to a player to adjourn whenever he wants, which in turn allows escapism.
Requiring agreement would not solve the situation as the "troll-ee" would always deny it.

Thats why very long games are suited for deferred-go, one that doesnt require the players to be present at all times.


There is always an upper limit, that is unvoidable. The current one is arbitrary, so the question is How can we measure what would be the best maximum time allowed for rated matches?

[HermanHiddema]

Why does there need to be a cutoff? Why not weigh a games effect on the rating by its time limits?

[topazg]

I can assure you no man alive ever played a 9999 absolute time live rated game. He would have to be awake for 7 days straight without sleeping.

I can't work out if you're being deliberately obtuse here or not My point is that both fast and slow games are often deliberate choices by players who also want their games to be rated. There is plenty of evidence supporting the fact that people like to do this, and I have yet to see a single individual ask for time limit caps on rated games to prevent game-ruining abuse. Why are you taking this choice away from them?

Concerning your comparison to Real life, you have to take into account that the internet has anonymity. I have never seen anyone in real life escape, yet is one of those issues all servers have to deal with. The internet is also less stable. People lose connections, get distracted, have technical issues, lose power. A single user that resorts to this tactic will do so for every single game on long time settings. It happens. It has happened to me on 15 minute games. On all servers I ever played in. Some servers have even taken measures to prevent this precise case. And it will happen simply because the other user has nothing to lose by waiting instead of resigning, he can just leave the computer on. His inaction is prized with opportunity.

If I come across someone who does this, I'll add them to an ignore list or equivalent, and just not play them again. If a player just leaves the computer on, it will still time out and he'll still lose, it's hardly a great opportunity! Besides, why should one guy being a jerk stop me from playing long rated games with other people who want to? You aren't punishing the abusers with your policy, you're punishing the honest players who just want long games.

The motivation for bottom limits is that a game with a very low time limit produce very imprecise results. A 9d can easily lose to a 10k even game in a 2 minute absolute time game. I dont think anyone advocates for those games to be rated.

I don't think anyone advocates for them not to be either. Find me a single 9d who has played a 10k in a 2 minute absolute game. People are aware of what they are doing when they take these games on, and I'm sure the potential increase in error margins isn't very high in their mind. Again, if they are happy with the risks to their rating by doing this, why are you telling them to take it to another server?

[Mivo]

The current one is arbitrary, so the question is How can we measure what would be the best maximum time allowed for rated matches?

Why does there have to be a maximum if any given time setting requires mutual consent from both players? If they want to allot 5 hours to their game, why not? I'd not agree to those settings, but the system would also not force me to. I see no harm in choice.

How about a warning if a player is about to accept "unusually long" time settings? Like, "Do you really want to play a game with 75 hours base time and 555 periods at 60 seconds each? This could take a long time and games cannot be adjourned."

[Kaya.gs]

The current one is arbitrary, so the question is How can we measure what would be the best maximum time allowed for rated matches?

Why does there have to be a maximum if any given time setting requires mutual consent from both players? If they want to allot 5 hours to their game, why not? I'd not agree to those settings, but the system would also not force me to. I see no harm in choice.

How about a warning if a player is about to accept "unusually long" time settings? Like, "Do you really want to play a game with 75 hours base time and 555 periods at 60 seconds each? This could take a long time and games cannot be adjourned."

I have explained in the previous posts why agreement is not sufficient, in practice. Confirmation of agreement does not solve the expectation of your opponent to actually play go with his alloted time.

I also add that there is a lighter limit on the byoyomi time itself.

In speedchase's case, he can use 30 minutes main time, and 60-70-80 seconds byo-yomi period.
He can get his slow game also, but preventing an absurd abuse of time as i stated before.

[palapiku]

I'm also very perplexed by this decision.

From a business perspective, why are you giving players a legitimate need to play on other servers?

From a potential for abuse perspective:

* Players know what they are signing up for;
* Admins should deal with blatant abuse;
* If someone wants to waste their opponent's time, a time-honored way to do it is to finish the game then never accept the result. This can happen in games with any time limit. On KGS people occasionally do this (happened to me once) but it's not a big deal because admins are there to deal with it.
* No other server has this limit and yet this does not seem to be an issue in practice.

From an esthetics perspective, it's ugly to have an arbitrary time limit where none is needed.

From a practical perspective, I personally occasionally want to play or watch games slower than 30 minutes, so I'm against this

[MarkSteere]

Concerning your comparison to Real life, you have to take into account that the internet has anonymity.

Yes and no. Most of the time you'll play people you've already played many times before and whose playing habits you're familiar with. When you play an anonymous person, or more to the point, someone unknown to you, their stats should be accessible. What percentage of their wins were escapes? With this knowledge in hand, you can constrain the time limits accordingly.

Of course there will be some arbitrary time limit. But you don't seem to be acknowledging that currently it's WAY TOO LOW.

[topazg]

Ok Gabriel, quick question:

You're seeing lots of people opposing this here, and quite a few of them opposing it strongly. Do you have a similar number of people feeling it's a superior decision to the alternatives? From a business perspective, it's clear you want the best for your server, yet opinion here is that the decision is to the detriment of the server. Much like wms' decision on Fischer timing, it's your prerogative to do whatever you like ... but I'm surprised you seem to be unswayed by the proportionally large lack of support it seems to be getting.

Do you have another reason for the policy that isn't related to the players' desires that over-rides the popular opinion in this thread, or is it a case of you feeling you know what's best for the players more than the players do?

[MarkSteere]

The biggest issue so far on beta.kaya is the lack of game offers rather than abuses.

Is it a case of you feeling you know what's best for the players more than the players do?

Gabriel, you got $13K in donations. How about investing in a box of Q-Tips? Family size.

[Admin: Please keep post on topic. In what way is this relevant to the Time settings on kaya.gs?

3. Meaningless Posts
Keep posts relevant to the topic and make sure to use the edit button when appropriate. Don't post for the sake of post counts. Spam is not allowed. Also, personal conversations between members should be done via Private Messaging, not in a thread. ]

[Kaya.gs]

We are all on the same side here, we are all go players, trying to look for a better solution. Lets keep that in mind.

I'm also very perplexed by this decision.

From a business perspective, why are you giving players a legitimate need to play on other servers?

From a potential for abuse perspective:

* Players know what they are signing up for;
* Admins should deal with blatant abuse;
* If someone wants to waste their opponent's time, a time-honored way to do it is to finish the game then never accept the result. This can happen in games with any time limit. On KGS people occasionally do this (happened to me once) but it's not a big deal because admins are there to deal with it.
* No other server has this limit and yet this does not seem to be an issue in practice.

From an esthetics perspective, it's ugly to have an arbitrary time limit where none is needed.

From a practical perspective, I personally occasionally want to play or watch games slower than 30 minutes, so I'm against this

Leaving the game as it is for 40-45 minutes hoping for a disconnection is not against the rules of the system itself. An admin cannot intercede in those cases.
If one user decides to lose on time, there is nothing to do but to wait. Can you imagine the helplessness of dealing with this over and over?

Their stats should be accessible. What percentage of their wins were escapes? With this knowledge in hand, you can constrain the time limits accordingly.

This is not useful at all. First of all, there is a lot of work to do on statistics just to try to predict your opponents behavior, which is a very complex solution for a user to learn to use.

Second, those users are more likely to WIN games because their opponent escaped, if there is any forfeit at all. And lastly, they are playing within the rules of the game their opponent accepted.

Bear in mind that the whole point here is how to respect the rating system, and how good a game with long settings is to the rating system itself.
Figure that if 20% of the 60m+ games ended up in forfeit, they would be pretty terrible.

Of course there will be some arbitrary time limit. But you don't seem to be acknowledging that currently it's WAY TOO LOW.

There is more than one conversation ongoing here. Some believe there should be no upper limit. You imply that you do believe there is a need for one.

They could be different arguments. I said before im not opposed at all to change it. But if i change it to say, 45, then why not 46? and so on. Thats why i asked a very important question in bold.

Ok Gabriel, quick question:

You're seeing lots of people opposing this here, and quite a few of them opposing it strongly. Do you have a similar number of people feeling it's a superior decision to the alternatives? From a business perspective, it's clear you want the best for your server, yet opinion here is that the decision is to the detriment of the server. Much like wms' decision on Fischer timing, it's your prerogative to do whatever you like ... but I'm surprised you seem to be unswayed by the proportionally large lack of support it seems to be getting.

Do you have another reason for the policy that isn't related to the players' desires that over-rides the popular opinion in this thread, or is it a case of you feeling you know what's best for the players more than the players do?

First of all, let me summarize the request:

I want to be able to play a long time settings with:

1- with any kind of time configuration.

Remember that right now, as it is speedchase can get a slow or even slower game than 40mins 40/5.

If you assume byoyomi starts around move 125, and there are in average 120 moves more, 40mins 40/5
is a maximum of about 168 minute game. (80 min + ((120moves + 10 periods)*40 secs/60secs))

With a 30 minute main time, 50 second/5 byoyomi, he gets a 168 minute game.
Even better, in the second case, if he gets to byoyomi early, the game lasts even longer!

But, the difference stalling in one game to the other is 43 minutes to 34, Might not seem like a lot, but each minute added to the main time enlarges this scenario even more.

Even more, you have up to 120 secs byo-yomi. With this calculation, the longest live game right now for Byo-yomi is over 5.5 hours!

2- and force my opponent to be online/live all the time during the game

When deferred go is in, any player could step out, disconnect, etc and deal with any situation that could arise in a game.

EDIT: As per a user's suggestion on Kaya, im expanding on this:

Deferred game on a live server like ours basically means there is no escaper clock. The clock would still be running, but the user has the option to leave for a certain amount of time, or deal with any technical issue he has with some time to spare.
So if a user is actually planning to spend more than 5hours on a game, he can easisly set up a deferred game, and the case where a user stalls is not terrible: several deferred games could be played at once, and you dont have to wait online for your opponent to lose on time.

3- and has to be rated

Free games have laxer limits and they could easily get expanded. In any kind of problem scenario, resiging is a viable option for any player.

Its a VERY specific request. If any of the above options is ceded, you can play it with the current system.

On the other hand, after hearing some people saying they want exactly this i said:

1- There is the problem of users stalling games, preventing you to play another, or finish the current one, and having to be online for a very long period of time doing nothing but waiting.

2- The current number is arbitrary, but any other number also would be arbitrary. How to make a less arbitrary decision? or How can we measure what would be the best maximum time allowed for rated matches?

3- From personal experience, and by example on FoD, this problem is real
It happens all the time. It happens on 10 minute games. I know for sure that this will happen even on 30 minutes, that we will have to deal with people in that situation that will be much more angry during the occurrence, and there is nothing to do for that case as there is no rule violation.

4- Many games will not be played under high time settings, mainly because only a few handful of users could take up all long games without ever actually playing, just fooling around. Personally i havent seen many non-tournament-rated-equally-handicapped 5.6hour games online.

Even after exposing all this, i still said i would change it to another number we can find suitable, however the only proposal has been "limitless", which is not a viable option without adjourning.

All voices are heard and votes are counted.
This debate could have been the other way around , "why not allow 1 minute absolute games be rated?" And many ultra-blitzers could have the opinion that it should.

This is in the interest of slow players, which are very likely to receive abuse and also affect the rating system with matches as valid as throwing games away.

Are you sure that 5.5 hour games are not enough? And if not, what is enough?

As i said, changing the limit is a single number, i can do it in a second, and im not too attached for a number i picked 4 months ago. IM looking for the next number being less arbitrary than this one. Also, what should we do if we do increase the number, and stalling becomes a regular issue?

[topazg]

Even after exposing all this, i still said i would change it to another number we can find suitable, however the only proposal has been "limitless", which is not a viable option without adjourning.

AFAIK, your server is the only one that restricts the rated status of games based on length, so limitless clearly is viable. If I want to play a 3 hour rated game on KGS I can. I can't remember the IGS settings in enough detail, but if I can play the length at all, I can play it rated. Tygem and wBaduk I don't play on, so I've no idea. Flyordie, Playok, Yahoo etc don't really count as proper Go servers (no disrespect intended to those who play on them). You aren't addressing a problem that others have battled with and found clever solutions, you are addressing a problem that no-one else thinks is a problem.

Escapers can be a problem, and there are a number of solutions implemented to deal with them on different servers - but I'd much rather see you address how to fairly handle escapers (and you'd get a really lively debate on here to help with that) than to restrict and hamper the honest users of Kaya, which is what you're currently doing.

All voices are heard and votes are counted.
This debate could have been the other way around , "why not allow 1 minute absolute games be rated?" And many ultra-blitzers could have the opinion that it should.

I am of the opinion that if two players set up a 1 minute ultra blitz game and want it to be rated, they should be able to. It's their choice, why restrict them? There are loads of 1x0:10 blitz rated games on KGS, even between high dans, and quite a few of those are over in 5 minutes (total time) - more power to them I say.

...Also, what should we do if we do increase the number, and stalling becomes a regular issue?

Develop a system that punishes regular stallers, rather than a system that punishes honest users.

FWIW, I think you're approaching this from entirely the wrong direction (and as someone that has developed a still running and popular Go server, I think I speak from at least some successful experience here) - you're trying to prevent a theoretical phantom problem that concerns you, rather than trying to maximise the potential and power of your server. Your job to create Kaya as a large and successful environment for playing Go is to empower your player base, particularly those who want to spend most of their time there. Features are great, but basic functionality is crucial, and long rated games I see as a core functionality of a Go server. Maybe I'm the only one who does, but I doubt it.

It's like video game DRM. Software pirates don't suffer from it, honest users do, and that's why that's such a fuss about it - it's done with good intentions, but the people that suffer are the loyal ones who are supposed to be being rewarded for their loyalty. You're falling into the same trap here.

Bottom line: Empower your faithful user base, don't restrict and hamper them.

Having said all that, if you want a suggestion on timing systems, I'd say assume a game has 250 moves (125 for each side), and byo-yomi is typically reached after 150. Model your time controls to take into account the maximum time those 250 moves would take under this criteria, and set a per player limit of maybe 2 or 3 hours each. I would assume taking 80% of byo-yomi time for each "life" is permissable, and allow all the lives to expire but the last one.

So for fischer, 60 mins + 15 secs per move would be 60 + (0.25 * 125) = 91.25 minutes, well within the cap. For canadian, 60 + 10stones/5mins, you'd have 60 + (50 / 10 * 5) = 85 minutes, again well within the cap. For Japanese, 60 mins + 5x0:30, you'd have 60 + (2 + (50 * 0:24)) = 80 minutes, again well within the cap.

Looking at those, and using those formulae, I'd opt for 2 hours for each side as a cap (as a compromise, I'd still prefer 3 hours), although I'd probably be happy going as low as 90 minutes if it meant there was at least a change in the system.

I'd still be in favour of not restricting your users at all, but you asked for a proposal, so there's a new cap proposal and a formula that makes it easy to calculate for each of the timing systems I'm aware that you use (obviously absolute time isn't hard to work out )

[speedchase]

Just FYI in the op I meant 5periods of 40 seconds each.
Either way, the argument against no limit seems to revolve around the concept that he users are stupid and don't understand what they are getting into, as well as made up data. It's allows any amount to time for rated games, and I have never once hear someone complain about people starting long games and quiting,

[shapenaji]

I just have to ask the question:

which is worse?

1) The blue moon case where someone takes advantage of someone who did not have time to play the game they signed up for (This is a terrible way to gain rank btw, 2 hours for a single win? Their rank isn't going anywhere anyway... why make it unrated?)

2) A common case where players get suboptimal functionality from the server.

This, as with so many other problems is a risk calculation, and I think you're coming down on the wrong side of it.


Further, you may not be able to adjourn, but the timer keeps on going right? if you get disconnected, you can reconnect no? If anything, disconnects are less of an issue in long games.

[MarkSteere]

One possible rational response: "Thanks for all the feedback, guys. It's exactly what I need to develop a superior Go server. In respect of your wishes, I effectively removed the time limit. Hope this helps. Thanks again."

[lemmata]

A lot of interesting points are being made in this thread.

One comment: Kaya is still in beta and it is under active development. Most parts of the server are subject to future changes. We should keep on making suggestions, but let's avoid offending the developers if possible (and if you are interested in getting them to change something).

More on topic: Personally, I would prefer it if all rated games were forced to be under fixed time settings. We cannot even pretend that 10m/3x10s is even remotely close to being the same game as 20m/3x30s or 45m/5x30s.

A similar solution is to have ratings for different time settings. We could have a "blitz" rating, a "middle" rating, and a "long" rating.

Lastly: If there is a feature you strongly want to see on Kaya, they do have that "Your Feature" donation option...