Page 2 of 21

Re: A new server is being developed: Kaya.gs

Posted: Sun Aug 14, 2011 6:46 am
by Kaya.gs
Excellent feedback guys! some people have been saying things that are already in the development list :). Let me answer each one.

Haha, Araban kind of beat me to the punch on this one.

For me, one of the greatest things about Gmail was its support of pop3--and especially imap, which very few services at the time offered (actually, I think it may still be one of the few free webmail providers that also offers imap access...although I've not really bothered looking around since 2004).

As for desktop applications being a "forsaken technology"...I understand where you're coming from on this one, but I'm sure I'm not the only one with something of a disdain for webapps. I'll take the forsaken technology of local, always available software any day, rather than something that may or may not work based on a multitude of factors, including browser version, network connectivity and performance, and server compliance. Sure, many of these factors are implicit with any online game, but with the KGS et al, I can at least fire up the client and replay any of my archived sgf's (cgoban happens to be my favorite editor/viewer, at the moment...even though it admittedly leaves something to be desired).

With that all said, there is nothing wrong with more competition, and especially innovation. Some of the things you're talking about do sound pretty neat.


Everyone is entitled to their opinion of course, and you can disagree. What i am trying to convey to you is that pages like facebook , or online email clients like Gmail/hotmail etc could have never been what they are if they were done through a java client. You can see thunderbird from mozilla as an example of a java client that lost badly to online email clients. Still i dont intend to follow this analogy, only wanted to answer this. To each his own :).


let's ask some random questions here to see how you are going to run this server.

what kind of rating system are you going to use on your server and to which other server does it most compare?

how will you cope with nuisances like escapers? also the policy for bans, will they be long/short or possibly permanent?

I suppose there will be a chatroom, will this be like KGS, or more like an IRC chat?

will there be something like KGS+? are there ideas on how much such a thing will cost?

When is there gonna be beta testing and how could one sign up?^^


Many things are still being defined but thought of.

Rating system: i want to implement an ELO system, very much like Wbaduk and Tygem and most Asian servers. I will make it so that the first set of games define the rank like KGS does which i think is the only advantage of the system.
So a few games to get a solid rank quickly, and then a transparent point based system.

One thing im talking with my partner about the rating system which we will try is the following: "14 victories gives you a rank up". These is something that happens with some club-rating systems like in the Nihon-kiin.
This means that we would always adjust the strength difference of any game, so if a weak 5d plays a strong 5d, there will a difference in komi. It is a more granular way to handle the rating system, and also a very predictable one. Each game you play is basically worth the same.

KGS rating system has some disadvantages, the main one being that accounts get "heavy", hence people always have multiple accounts. Wbaduk, for example, poeple never make more than 1 account. But it still a drag that when you are close to rank up, you have to play way more many games.

Escapers: similar to Asian servers. The whole idea from kgs to keep the games is interesting but it just doesnt work. IF a player is disconnected or leaves, he can join back in a certain amount of time (15 mins ~) or else he loses. I dont want Kaya.gs to have escaper policies in the main room to explain it to people. Simple is good :).

Channels/Rooms: This is also a key point. Good question :). In Kaya.gs we differenciate between a room and a channel. It is very important that people can make a flexible group to chat with however they want. On the other hand its not cool that you have 500 rooms with 1 people, making it harder to find the active/valuable ones.
So there is a channel list, very open and people can open them up as many as they want. Its simply a chat-tab that people can talk simultaneously in.
And there is a room list with much more features and support. Only active communities, or qualified services get on the rooms list which will be very limited and on sight. A room will have a description, links and other website related technology. Depending on the service, joining a room could be opening up a new "client". Think of joining the room goproblems.com, where you can solve problems and share demonstrations on specific problems you or other people want to solve.

We are still ironing details on how we are going to do the latter, but the differenciation of channels and rooms is here to stay. Also we want to put IRC-like commands on the chat. So you will be able to write to someone when you mention their name (i.e. Dexmorgan what do you mean? would be shown in bold to me) and other similar commands.

I will make a public beta call when its time :).

1. Fischer time controls

2. Something for tsumego lover


Nice!. Fishcer time is actually the first time system im pushing my partner to do. i never tried it but i dont see how it could possible not be better than byoyomi. It is a great time setting and i think blitz people will love it.

There will be something special about tsumego..but that is classified information :)


9x9 and 13x13 wont be ranked which is something i feel strongly about. However, we will make an option when you create your account to mark it as 'beginner'. Beginner accounts will have a progress window to guide them into getting out of beginner-ship.

We will start with something very simple on that like solving a certain set of problems, and playing a certain number of Atari-Gos, 9x9 and 13x13. We really want to make the server a great place to learn Go from scratch, although most beginner stuff will come later on when the server is already mature.

Re: A new server is being developed: Kaya.gs

Posted: Sun Aug 14, 2011 7:18 am
by Kirby
Good luck. Many people have tried to create a new go server, but have given up.

I wonder if you will be different. I look forward to any results that you may have. In the event that nothing ever results, thanks for trying.

Re: A new server is being developed: Kaya.gs

Posted: Sun Aug 14, 2011 8:31 am
by daniel_the_smith
Kaya.gs wrote:9x9 and 13x13 wont be ranked which is something i feel strongly about. However, we will make an option when you create your account to mark it as 'beginner'.


Have you considered publishing separate blitz and normal ratings? If so, you could make separate ratings for 9x9 and 13x13 (I agree they probably shouldn't be mixed with 19x19 rating).

Have you heard of Whole History Rating? This paper is worth reading if you haven't already: http://remi.coulom.free.fr/WHR/

Re: A new server is being developed: Kaya.gs

Posted: Sun Aug 14, 2011 9:04 am
by iazzi
Kaya.gs wrote:You can see thunderbird from mozilla as an example of a java client that lost badly to online email clients.


Since Mozilla Thunderbird is not a java client (nor are most IGS clients, nor WBaduk or Tygem as far as I can tell) I wonder what you are calling "java"...

Re: A new server is being developed: Kaya.gs

Posted: Sun Aug 14, 2011 9:55 am
by stalkor
i've heard that the server is gonna be built in html5, so that is compatible with all browsers if you like it or not:)

Re: A new server is being developed: Kaya.gs

Posted: Sun Aug 14, 2011 10:32 am
by Bantari
What has been written here about the new server outlines a very ambitious project.
And the details and possibilities look awesome... I will be the first to sign up if you guys are successful.
I think it is correct that the web-based server has many advantages over local client-based server, and its good to see people thinking about that.

On the other hand, it seems to me a huge undertaking to run a Go server. And I mean 'RUN', not just build. Building it and programming it and implementing it - this is the FUN part for any programmer and Go player. But the real work starts after that... the day-to-day duties a server master has to fulfill. Setting policies, making decisions, maintaining, and taking crap from all sides. And doing it day after day after day... It scares the heck out of me even to think about that.

Not many people are able to do that, even if there are a lot of us skilled enough to actually build such server. Will the owners of Kaya.gs be able to? After all, a viable go server is only viable if it is allowed to grow over a long period of time, especially if it has to compete against existing successful services. So it will have to, most likely, run with minimal membership for quite a while - see the other web-based servers. Years, maybe. Until something happens and the 'critical mass' of users will be reached.

And another thing - building a server is a good thing, but there has to be a business model behind. So, how is the new server to be financed? Membership dues, ads, sponsors? Or will it rely on volunteers all the way?

I dunno... this is what I think when I read this thread.

However, what I absolutely DO NOT agree with is the people saying 'well, we have KGS why even try to do anything different or better?' Very narrow horizons, I think.

Re: A new server is being developed: Kaya.gs

Posted: Sun Aug 14, 2011 11:17 am
by Kaya.gs
iazzi wrote:
Kaya.gs wrote:You can see thunderbird from mozilla as an example of a java client that lost badly to online email clients.


Since Mozilla Thunderbird is not a java client (nor are most IGS clients, nor WBaduk or Tygem as far as I can tell) I wonder what you are calling "java"...


Sorry i meant desktop client for thunderbird. Thunder is made on C++ which is certainly something we wont use for Kaya.gs.

Im surprised about wbaduk if its not made with Java. The homepage does ask for Java web-start to run the client.


Regards.

Re: A new server is being developed: Kaya.gs

Posted: Sun Aug 14, 2011 12:04 pm
by Li Kao
Doesn't sound very attractive to me:
* Bad rating system (WHR 4tw)
* Bad escaper policy
* No way to notify the user of chat messages or that it is his move to play apart from audio
* No context menus
* Can't autosave games and chatlogs to my harddisk

I'd strongly prefer an open source client to a pure website. One could work around that with a custom program that embeds a browser control and that offers a better GUI.

Re: A new server is being developed: Kaya.gs

Posted: Sun Aug 14, 2011 12:55 pm
by hailthorn011
One thing I'd like to see is user created tournaments. I'm not sure if you can do that on any existing Go servers, but it would certainly be a cool feature. And sure, you can set up informal tournaments and whatnot, but imagine if there was a button for creating tournaments?

Example:

-Click Create Tournament
-Decide rule set/time
-Decide tournament format-
-Minimum number of players/maximum number of players
-Then post it

So it'd be kind of like creating a game on KGS, except it would create a tournament branch. And I suppose there should also be a way to check winners and losers.

But that's for tournament nuts like me who can't go out and play in them in real life.

Re: A new server is being developed: Kaya.gs

Posted: Sun Aug 14, 2011 2:59 pm
by karaklis
Li Kao wrote:Doesn't sound very attractive to me:
* Bad rating system (WHR 4tw)

Indeed, the rating system of Wbaduk (and Tygem which is very similar) is broken in the kyu area. Prone to sandbaggers. KGS' ranking system is actually good except for that the ranking changes as the ranking of previous opponents change.


Li Kao wrote:* Bad escaper policy

Why do you think this escaper policy is bad? Which one would you think is better? Everybody here is complaining about the escaper policy of KGS, but I haven't seen anybody complaining about the others' escaper systems.

Re: A new server is being developed: Kaya.gs

Posted: Sun Aug 14, 2011 7:54 pm
by Kaya.gs
hailthorn011 wrote:One thing I'd like to see is user created tournaments. I'm not sure if you can do that on any existing Go servers, but it would certainly be a cool feature. And sure, you can set up informal tournaments and whatnot, but imagine if there was a button for creating tournaments?

Example:

-Click Create Tournament
-Decide rule set/time
-Decide tournament format-
-Minimum number of players/maximum number of players
-Then post it

So it'd be kind of like creating a game on KGS, except it would create a tournament branch. And I suppose there should also be a way to check winners and losers.

But that's for tournament nuts like me who can't go out and play in them in real life.


This is going to be available. Im actually talking to a developer that wants to make a website to organize tournaments: kaya.gs would simply utilize that website and achieve the desired functionality.

This is what kaya.gs is all about, that people want to do new things and that they can be integrated into the server.



Doesn't sound very attractive to me:
* Bad rating system (WHR 4tw)
* Bad escaper policy
* No way to notify the user of chat messages or that it is his move to play apart from audio
* No context menus
* Can't autosave games and chatlogs to my harddisk

I'd strongly prefer an open source client to a pure website. One could work around that with a custom program that embeds a browser control and that offers a better GUI.


There is no point in discussing if you like Kaya.gs because it doesnt exist yet. ELO is by far the most tested rating system out there, hardly to say a bad rating system. EGF ratings use point based system.

I dont see why you say that there is no way to to notify the user, in the dev server of Kaya.gs we have played games with sound and visual notification. Also we have context menus. And also we can autosave games on the server. And chat logs arent saved by KGS as far as i know: but wait, we can save them. We might not want to, as opposed to KGS that has right to every single private message you send.

What has been written here about the new server outlines a very ambitious project.
And the details and possibilities look awesome... I will be the first to sign up if you guys are successful.
I think it is correct that the web-based server has many advantages over local client-based server, and its good to see people thinking about that.

On the other hand, it seems to me a huge undertaking to run a Go server. And I mean 'RUN', not just build. Building it and programming it and implementing it - this is the FUN part for any programmer and Go player. But the real work starts after that... the day-to-day duties a server master has to fulfill. Setting policies, making decisions, maintaining, and taking crap from all sides. And doing it day after day after day... It scares the heck out of me even to think about that.

Not many people are able to do that, even if there are a lot of us skilled enough to actually build such server. Will the owners of Kaya.gs be able to? After all, a viable go server is only viable if it is allowed to grow over a long period of time, especially if it has to compete against existing successful services. So it will have to, most likely, run with minimal membership for quite a while - see the other web-based servers. Years, maybe. Until something happens and the 'critical mass' of users will be reached.

And another thing - building a server is a good thing, but there has to be a business model behind. So, how is the new server to be financed? Membership dues, ads, sponsors? Or will it rely on volunteers all the way?

I dunno... this is what I think when I read this thread.

However, what I absolutely DO NOT agree with is the people saying 'well, we have KGS why even try to do anything different or better?' Very narrow horizons, I think.



It is no less than ambitious. We are truly building the next generation Go server and it deserves to go as high as we can take it. To your concerns about running the server, i can only answer you that we have the blind-sighted entrepreneur confidence that we can take anything that shows up :).

We are building a serious and professional server. There is a business model but its not sensible to go public with it now as its not going to be implemented in quite a while.
Most of your concerns are very real, but i cant really discuss them so openly at this stage of the project , as they can change drastically.

Re: A new server is being developed: Kaya.gs

Posted: Sun Aug 14, 2011 8:08 pm
by hailthorn011
This is going to be available. Im actually talking to a developer that wants to make a website to organize tournaments: kaya.gs would simply utilize that website and achieve the desired functionality.

This is what kaya.gs is all about, that people want to do new things and that they can be integrated into the server.


Sign me up, then!

Re: A new server is being developed: Kaya.gs

Posted: Mon Aug 15, 2011 1:16 am
by karaklis
Kaya.gs wrote:ELO is by far the most tested rating system out there, hardly to say a bad rating system. EGF ratings use point based system.

Not sure whether the EGF system is the best system. I've heard rumors that the Glicko rating system is used for chess ratings, which is an improvement over ELO. Or better: use Glicko-2.

Re: A new server is being developed: Kaya.gs

Posted: Mon Aug 15, 2011 1:47 am
by hilltopgo
This is the first I've heard of Glicko. Its "rating volatility" parameter sounds interesting, and essentially similar to the AGA's "sigma."

Re: A new server is being developed: Kaya.gs

Posted: Mon Aug 15, 2011 1:59 am
by danielm
hilltopgo wrote:This is the first I've heard of Glicko. Its "rating volatility" parameter sounds interesting, and essentially similar to the AGA's "sigma."


It's just ELO with the RD parameter. It's used on FICS for example, but they don't call it Glicko. It makes sense, but you also get "heavy" accounts as on KGS of course. This should be balanced so that it never becomes sensible to create a new account, just so the rating system can catch up to your latest improvement...