Darrin Bond Sep 29 1993, 9:09 am show options Newsgroups: rec.games.chess From: dabst...@pitt.edu (Darrin Bond) - Find messages by this author Date: 29 Sep 1993 10:59:31 -0500 Local: Wed, Sep 29 1993 8:59 am Subject: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse I undrstand that the powers at be are still looking for a new host for server. I say again: The University of Pittsburgh Chess Clubs' offer to host the chess server is still valid and the machine is still waiting. Darrin L. Bond President University of Pittsburgh Chess Club Valentin Pepelea Sep 30 1993, 12:53 am show options Newsgroups: rec.games.chess From: valen...@netcom.com (Valentin Pepelea) - Find messages by this author Date: Thu, 30 Sep 1993 07:28:11 GMT Local: Thurs, Sep 30 1993 12:28 am Subject: Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse In article dabst...@pitt.edu (Darrin Bond) writes: >I undrstand that the powers at be are still looking for a new host for >server. I say again: The University of Pittsburgh Chess Clubs' offer to >host the chess server is still valid and the machine is still waiting. Perhaps theere's some animosity between you guys and them. You're only a stone trow away from each other, after all. (no pun intended) Valentin -- "An operating system without virtual memory Name: Valentin Pepelea is an operating system without virtue." Phone: (408) 249-6610 Internet: valen...@netcom.com - Ancient Inca Proverb Cory Tichauer Sep 30 1993, 6:28 pm show options Newsgroups: rec.games.chess From: volta...@cwis.unomaha.edu (Cory Tichauer) - Find messages by this author Date: Thu, 30 Sep 1993 22:39:46 GMT Local: Thurs, Sep 30 1993 3:39 pm Subject: Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse I think that this is a great idea!! Finally there woulkd be a dedicated server. Why doesn't anyone take this offer seriously? I think that the Darooha Master/ADMIN should let the chess community know why this isn't occuring! Come on ICS Admins lets have a response!! Voltaire P.S. The Club that offered the computer could always get a copy of the server and start there own!! I for one would love to have a server to connect to without limits on number of people!! -- _______________________________________________________________________ | Voltaire "Justice is only the perverse desire | | volta...@cwis.unomaha.edu to spread the pain around equally" | |______________________________________________________________________| Kirk Gunsallus Oct 1 1993, 6:56 am show options Newsgroups: rec.games.chess From: k...@alchemy.TN.Cornell.EDU (Kirk Gunsallus) - Find messages by this author Date: 1 Oct 1993 13:14:55 GMT Local: Fri, Oct 1 1993 6:14 am Subject: Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse - Hide quoted text - - Show quoted text - In article <1993Sep30.223946....@news.unomaha.edu> volta...@cwis.unomaha.edu (Cory Tichauer) writes: >I think that this is a great idea!! Finally there woulkd be a dedicated server. >Why doesn't anyone take this offer seriously? I think that the Darooha >Master/ADMIN should let the chess community know why this isn't occuring! >Come on ICS Admins lets have a response!! > Voltaire >P.S. The Club that offered the computer could always get a copy of the server >and start there own!! I for one would love to have a server to connect to >without limits on number of people!! >-- >_______________________________________________________________________ >| Voltaire "Justice is only the perverse desire | >| volta...@cwis.unomaha.edu to spread the pain around equally" | >|______________________________________________________________________| I am not an ICS admin so this information is secondhand but as you might guess, if the offer has been made for some time and hasn't been acted on, there's probably a reason. Daroo, as main developer (after all the good work done before him) of the ICS code wishes to maintain some control over it so that it does not become the mess it once was (from a programming point of view). Suffice it to say, the offer was contingent on the control being lost. This was not the only ocncern. I don't believe the code is public domain so I don't think one can just grab it and start their own server. We would all love to have a server without limits. Active searches are still taking place (evidenced by the recent attempt at rafael....) but until a site is found that allows _continued_ good operation, we must be content with what we have. Keep up the good work Daroo... K cagey@ICS Darrin Bond Oct 1 1993, 8:02 am show options Newsgroups: rec.games.chess From: dabst...@pitt.edu (Darrin Bond) - Find messages by this author Date: 1 Oct 1993 08:53:57 -0500 Local: Fri, Oct 1 1993 6:53 am Subject: Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse Dan Sleator was approached first to move the server, which he declined after some discussion. He was then asked for just the EXE files to which he also refused saying that they (?) had decided that it was best that only one server be operating in North America. We now stand at an impasse. The Pitt group consists of three programmers who are attempting to rework the old code written by M. Moore and R. Nash. Two are dedicated to the C programming and one is doing the network interface. We are also attempting to rework the code for the program that allows ratings to be compiled and stored from many servers as if there were only one (Dan killed this idea long ago). Unfortunnately for the ICS players it has been estimated that there is at least a man month (approx.750 hours) of work to do. Anyone interested in getting involved in this project is invited to email this account. Thanks, Darrin L. Bond President University of Pittsburgh Chess Club Axel Boldt Oct 3 1993, 7:27 pm show options Newsgroups: rec.games.chess From: a...@renoir.uni-paderborn.de (Axel Boldt) - Find messages by this author Date: 3 Oct 1993 09:25:33 +0100 Local: Sun, Oct 3 1993 1:25 am Subject: Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse dabst...@pitt.edu (Darrin Bond) writes: >Dan Sleator was approached first to move the server, which he declined >after some discussion. He was then asked for just the EXE files to which >he also refused saying that they (?) had decided that it was best that >only one server be operating in North America. We now stand at an >impasse. >The Pitt group consists of three programmers who are attempting to >rework the old code written by M. Moore and R. Nash. Two are dedicated >to the C programming and one is doing the network interface. We are also >attempting to rework the code for the program that allows ratings to be >compiled and stored from many servers as if there were only one (Dan >killed this idea long ago). I hope that you guys will then put your code under the gnu license or something, so that the sources become available and this ridiculous situation ends. (I thought the times with hidden sources were long gone!?) Axel. Daniel Sleator Oct 4 1993, 5:29 am show options Newsgroups: rec.games.chess From: sleat...@cs.cmu.edu (Daniel Sleator) - Find messages by this author Date: Thu, 30 Sep 1993 23:17:14 GMT Local: Thurs, Sep 30 1993 4:17 pm Subject: Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse Darrin Bond's offer was considered and rejected. Among other problems: The machine he proposed can only accommodate 64 open files, or about 55 players. Bond wanted to eliminate all games shorter than 5 0 time control. He also wanted to eliminate wild games. (I may have talked him out of idea.) He wanted to have control over the selection of admins. Meanwhile, other promising options have appeared on the horizon. Daniel Sleator (a.k.a. Darooha) John Anthony Chanak Oct 4 1993, 1:00 pm show options Newsgroups: rec.games.chess From: John Anthony Chanak - Find messages by this author Date: Mon, 4 Oct 1993 14:35:28 -0400 Local: Mon, Oct 4 1993 11:35 am Subject: Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse Just a few thoughts and experiences... The idea of having a ratings system spread over a wide area is well and good, but not very easy. Big software houses producing real products have difficulty providing distributed database-type services. The problem is easy to state, *very* hard to solve. Danny Sleator and I talked about this for a long time about 6 months ago. We decided that yeah, it would be neat. We also decided that it would be next to impossible to do correctly. (Certainly beyond the time either of us could put into it) Keep in mind that Danny is a very knowledgeable programmer. Though his specialty is not distributed systems, we both have considerable knowledge in this area. Trust that we know it is not an easy task. (If you want a "simple" distributed system that breaks every so often, that is quite possible. However, how useful is this when you can have a robust, secure system on one machine?) A far better technique is to rewrite the network interface ICS uses. I approached this problem at one point, and just never got into it too much. (The basic idea is to distribute communication overhead over multiple hosts, keeping one machine as the central processor. I even had a rudimentary version of this running late this spring for when GM Rhode played Deep Blue. Unfortunately, it was not very stable.) -John Chanak (MataPato, at one time... sniff...) Karl Schwamb Oct 4 1993, 1:54 pm show options Newsgroups: rec.games.chess From: schw...@isi.edu (Karl Schwamb) - Find messages by this author Date: 4 Oct 1993 17:24:28 GMT Local: Mon, Oct 4 1993 10:24 am Subject: Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse In article <28hagf$...@fitz.TC.Cornell.EDU>, k...@alchemy.TN.Cornell.EDU (Kirk Gunsallus) writes: > I am not an ICS admin so this information is secondhand but as you might > guess, if the offer has been made for some time and hasn't been acted on, > there's probably a reason. Daroo, as main developer (after all the good > work done before him) of the ICS code wishes to maintain some control over > it so that it does not become the mess it once was (from a programming > point of view). Suffice it to say, the offer was contingent on the > control being lost. This was not the only ocncern. If this is true, I'm even more dismayed with ICS (er, than with quotas and lag!). Trying to monopolize the chess service is just crap. Chess servers should act as chess clubs not as electronic versions of the USCF. The ICS code (or some analogue) should be public domain so that a chess server can be started by anyone who has the resources. If consistent ratings are desired then someone should code up a ratings server that can interchange information with the chess servers. BTW, this would also solve the consistency problem with the US and European servers. If the software is being horded because some folks want to keep the code pretty, this is even sillier. Almost no one cares what the code looks like -- the users want a functional, fast server. If some dedicated person doesn't code well but still provides great service -- who cares! If there are several servers to choose from then the net users will flock to those that provide good service -- not good code design. > We would all love to have a server without limits. Active searches are still > taking place (evidenced by the recent attempt at rafael....) but until > a site is found that allows _continued_ good operation, we must be content > with what we have. What nonsense. There is no such thing as a truly dedicated server. The only way to provide chess service to the net users as a whole is to start up a number of servers so that a single node failure will not remove service for the Internet as a whole. -Karl James Aspnes Oct 4 1993, 5:54 pm show options Newsgroups: rec.games.chess From: aspnes-ja...@cs.yale.edu (James Aspnes) - Find messages by this author Date: 4 Oct 1993 19:36:58 -0400 Local: Mon, Oct 4 1993 4:36 pm Subject: Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse In article John Anthony Chanak writes: The idea of having a ratings system spread over a wide area is well and good, but not very easy ... Actually, it's not all that hard, as long as you don't care about getting instant updates on rating and are willing to give up a little security. Each local ICS site can simply mail out a message (using the existing mail infrastructure) to the ratings server every time a game finishes giving the players, the score, a timestamp, and (to prevent trivial cheating) a cryptographic checksum of the message appended to a secret key. The ratings server itself can be a batch job that just processes all the mail once a day (say) and mails the adjusted ratings back out to the local sites. You could do most of this in about forty lines of perl. ( If you want a "simple" distributed system that breaks every so often, that is quite possible. However, how useful is this when you can have a robust, secure system on one machine ?) The above system, if implemented correctly, would be as robust as the underlying mail system, and more robust than ICS currently is (I've both gained and lost rating points to machine crashes). Its only defect is that it doesn't give immediate feedback on rating changes. But individual sites could still show "performance ratings" that would be updated immediately, and which would follow the official ratings pretty closely except for players who used multiple sites. A far better technique is to rewrite the network interface ICS uses. I approached this problem at one point, and just never got into it too much. (The basic idea is to distribute communication overhead over multiple hosts, keeping one machine as the central processor. I even had a rudimentary version of this running late this spring for when GM Rhode played Deep Blue. Unfortunately, it was not very stable .) This sort of thing is certainly possible (it's been done for MUDs), but it's not clear for example that MIT would be happier if the ICS traffic swamping their network were going to multiple machines instead of just one. I see no reason why there could not be regional servers competing with each other for traffic on the basis of response time, robustness, and friendliness of the local culture, except for the understandable but ultimately wrong-headed desire of those who have worked very hard on the ICS code to maintain control over the product of their efforts. I hope that in the long run this desire will wear off. --Jim Aspnes Karl Schwamb Oct 4 1993, 5:55 pm show options Newsgroups: rec.games.chess From: schw...@isi.edu (Karl Schwamb) - Find messages by this author Date: 4 Oct 1993 19:48:20 GMT Local: Mon, Oct 4 1993 12:48 pm Subject: Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse In article , John Anthony Chanak writes : > (If you want a "simple" distributed system that breaks every so often, > that is quite possible. However, how useful is this when you can have a > robust, secure system on one machine?) Well the simple approach seems the way to start. If you try to solve the general problem first, then nothing gets improved. I would be happy with a system that sent updates to a single ratings server. If that ratings server went down, then game results could be logged locally and ratings extrapolated by the local chess server. When the ratings server was back on-line it could notify the chess servers and dbs could be sync'ed again. If the same ratings algorithm were used by the chess server and the ratings server, then users who used the same chess server exclusively wouldn't notice any difference. If users hopped back and forth across servers then ratings could change for those users if the ratings server wasn't kept up to date (or was down). But that's no big deal, that's just life on the Internet. Of course, service could be improved by having two or more rating servers which kept each other up-to-date to protect against single-point failures. As far as the all-on-one-server idea goes, if you run a ratings server and a chess server on the same machine, you could always provide an efficient back door to communicate ratings changes between the two processes. This might be marginally slower than the existing design while allowing multi-site chess service/ratings support. If your site had a lot of resources then a chess server and ratings server could be provided. If you had fewer resources, then you might want to provide only chess service and depend on remote hosts for rating service. If we wanted to improve rating accuracy, then rating servers (w/o chess servers) could be run at additional sites. Each time a user would log into a chess server, query all active ratings servers for the latest rating info. After each game, update the user's rating and broadcast the result to other active servers. Of course, there are scenarios where this simple scheme would fail but it would 1) be an improvement on the existing scheme and 2) provide infrastructure that would be needed in a general solution. -Karl Paul Colley Oct 5 1993, 8:38 am show options Newsgroups: rec.games.chess From: col...@qucis.queensu.ca (Paul Colley) - Find messages by this author Date: Tue, 5 Oct 1993 15:20:20 GMT Local: Tues, Oct 5 1993 8:20 am Subject: Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse In article <28pm8c$...@venera.isi.edu> schw...@isi.edu (Karl Schwamb) writes: > Trying to monopolize the chess service is just crap. I see no evidence that the authors of the ICS are trying to monopolize chess service. There's nothing to hinder you from writting your own chess server. The authors of ICS wrote the ICS code. It is THEIRS. It would be nice if they gave it away for free; but like any donation to a cause, such donations are the choice of the giver, not the recipient. The correct reaction is thankfulness for what has been given, not whining that we'd like more. Keeping one's code is not trying to monopolize anything! Go write your own server. Having two choices would be a good thing! There are things about the ICS interface that I think could be improved. > The ICS >code (or some analogue) should be public domain so that a chess server can >be started by anyone who has the resources. Fine. If you strongly believe in this, write a chess server. I'm not being facetious. If it's trivial, do it instead of complaining. If it's a huge amount of work, well then, you might understand why the authors feel like protecting THEIR work. The GNU project is a miracle, something to wonder at, that so much quality software has been distributed with so little restriction. But not everybody wants to do hard work to have somebody else stomp all over it. Even the FSF asserts copyright, because there are things they don't want to see happen to their code. >If the software is being horded because some folks want to keep the code >pretty, this is even sillier. Almost no one cares what the code looks like Hate to tell you this, but authors of novels get upset when a director trashes the movie version of their work; and software authors get upset when somebody takes their finely crafted code and produces a bug-ridden monstrosity. To call them silly is to insult the pride they have in their workmanship. There is only one person whose opinion matters: The author's. The author created the code/novel/painting, and is morally and legally entitled to control over what happens to their creation. The correct response is simple: Write your own. >-- the users want a functional, fast server. The users only get what they want if somebody puts the time and effort into it. Since the "users" don't pay, and don't write code, they are going to have to be thankful for what they get. In private Email, I told somebody the amount I was willing to pay to use ICS. The response was somewhat idiotic, that "everything on Internet must be free." I guess it was somebody who hasn't seen the costs of providing Internet access to their site... > If some dedicated person >doesn't code well but still provides great service -- who cares! If there >are several servers to choose from then the net users will flock to those >that provide good service -- not good code design. Good point. So, what we want is a dedicated person who is willing to write a server and give it away freely. We don't have this mythical person yet. We have people who have written a server and are willing to let us use it for free in the meantime. Count your blessings; without what has been done, you would have no server at all. In the future you may have the public-domain code you crave, but not until somebody writes it and gives it away. - Paul Colley (Who has no intention of writting a chess server) University: col...@qucis.queensu.ca Home: pacol...@ember.uucp watmath!ember!pacolley +1 613 545 3807 "Quantum Mechanics: The dreams stuff is made of" - Ken Burnside Paul Colley Oct 5 1993, 8:47 am show options Newsgroups: rec.games.chess From: col...@qucis.queensu.ca (Paul Colley) - Find messages by this author Date: Tue, 5 Oct 1993 15:34:24 GMT Local: Tues, Oct 5 1993 8:34 am Subject: Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse In article <28qc2qINN...@PINE.THEORY.CS.YALE.EDU> aspnes-ja...@cs.yale.edu (James Aspnes) writes: > except for the understandable but >ultimately wrong-headed desire of those who have worked very hard on >the ICS code to maintain control over the product of their efforts. I >hope that in the long run this desire will wear off. I don't know if "wrong-headed" is fair. What motivates someone to write a server that total strangers can use? Some of the ideas that come to mind are: - Being the first to do it - Being the only one to do it - Joy in writting good code that others covet - Having control over the server All of these would disappear (except the first) if the code were public-domain. Ultimately, somebody is going to write a competing server, and ICS' heyday will begin to wind down. But to call the ICS authors "wrong-headed" requies knowledge of their goals, and I don't think that is public knowledge. If the authors only goal was to promote on-line chess, then they I think they are wrong-headed. But I suspect that some of the other goals I listed are important too. ------- I too hope that the authors change their mind and release the code. But I can't see cause to criticize their decision. - Paul Colley - Hide quoted text - - Show quoted text - University: col...@qucis.queensu.ca Home: pacol...@ember.uucp watmath!ember!pacolley +1 613 545 3807 "Quantum Mechanics: The dreams stuff is made of" - Ken Burnside Jeffrey C. Ely Oct 5 1993, 10:17 am show options Newsgroups: rec.games.chess From: j...@econ.Berkeley.EDU (Jeffrey C. Ely) - Find messages by this author Date: 5 Oct 1993 17:09:40 GMT Local: Tues, Oct 5 1993 10:09 am Subject: Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse In article , - Hide quoted text - - Show quoted text - Paul Colley wrote: >I don't know if "wrong-headed" is fair. What motivates someone to >write a server that total strangers can use? Some of the ideas that >come to mind are: > - Being the first to do it > - Being the only one to do it > - Joy in writting good code that others covet > - Having control over the server >Ultimately, somebody is going to write a competing server, and ICS' >heyday will begin to wind down. But to call the ICS authors >"wrong-headed" requies knowledge of their goals, and I don't think that >is public knowledge. Anyone who knows Darooha can immediately pick out the elements on your list that apply. Jeff &&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&& Jeff Ely* "He said the Sun's not yellow... Department of Economics, UC Berkeley It's Chicken!" j...@econ.berkeley.edu &&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&& *No opinions, unemployed Erik E. Wachtenheim Oct 5 1993, 11:09 am show options Newsgroups: rec.games.chess From: Erik.E.Wachtenh...@dartmouth.edu (Erik E. Wachtenheim) - Find messages by this author Date: Tue, 5 Oct 1993 17:11:15 GMT Local: Tues, Oct 5 1993 10:11 am Subject: Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse I have been reading the discussion concerning the ICS server and would like to put in my 2 cents worth. I have been a user of the server since the Moore days when there were no increments, no special output to help interface programmers, etc. I have seen the program evolve into what I consider to be a very good program and an excellent gift to all of us chess players with internet access. Daniel Sleator is responsible for many of the recent improvements including formatted ouput of the moves, and I think he is doing a very good job at satisfying the needs of the users as they arise. As the co-author of one of the graphical interfaces to the server I very much appreciate Daniel's desire for control over the code. I am all for re-locating to a computer which could handle a large number of simultaneous users and at the same time be reliable. I must admit, however, that I am somewhat hesitant about the idea of some "other" group taking over the code and re-working it. It has already been mentioned here that if someone is that eager to make major modifications to the code and is not content with the current server and its faults, they should take on the task of writing their own server. I, for one am very grateful to Michael Moore, Aaron Putnam, and Daniel Sleator for making realtime networked chess a possibility. -Erik Wachtenheim eew@ics John Anthony Chanak Oct 5 1993, 12:14 pm show options Newsgroups: rec.games.chess From: John Anthony Chanak - Find messages by this author Date: Tue, 5 Oct 1993 14:18:52 -0400 Local: Tues, Oct 5 1993 11:18 am Subject: Distributed ICS, was Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse You say: **** Each time a user would log into a chess server, query all active ratings servers for the latest rating info. After each game, update the user's rating and broadcast the result to other active servers. Of course, there are scenarios where this simple scheme would fail but it would 1) be an improvement on the existing scheme and 2) provide infrastructure that would be needed in a general solution . **** This is a good idea, but: Duplicate ratings information requires that you be careful with your queries/responses. In particular, you need to satisfy some quorum to ensure that you get the most up to date information. This means that you have to know and agree on three things. (globally agree) 1. How many ratings servers there are with ratings information. (A) 2. How many ratings servers must be writtten to on write. (B) 3. How many ratings servers must be read from on read. (C) To ensure validity of up-to-date information, A + 1 >= B + C *must* be true. This is the *only* way to ensure that you get up to date information. Not only that, but A, B, and C must not ever change, and must be recognized by all servers. (Actually, A, B, and C can change only such that this change does not invalidate the above stipulation. i.e. if B = 5, and C = 5, A can be anywhere from 5 to 9.) Now, from a programming perspective: Assume there are 7 servers out there. We decide to read from 4, write to 4. (A = 7, B = 4, C = 4) When someone logs in, we must query 4 servers. (These are 4 network connections.) You have added programming complexity in ensuring that these queries are non-blocking. (Not too hard.) Next, when you need to write, you also need to write to 4 servers, again non-blocking. (And for those that prefer to fork a process to do this, no way- forking is way too expensive) Next, you need to have some amount of confidence that your write succeeded. In all, this adds up to a lot more complexity. It can be done, but it is not at all trivial, and the overhead incurred will not be minimal. Some other things to consider: 1. Remote servers going down. What happens if you cannot write to 4 servers when you need to? 2. What happens when servers are added/removed? 3. How do you ensure some degree of security? -John Patrick Surry Oct 6 1993, 7:38 am show options Newsgroups: rec.games.chess From: p...@epcc.ed.ac.uk (Patrick Surry) - Find messages by this author Date: Wed, 6 Oct 1993 09:01:06 GMT Local: Wed, Oct 6 1993 2:01 am Subject: Re: Distributed ICS, was Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse I think a much simpler way to do this to start off with is to do "global" ratings updates at a *much* lower frequency than once per game played. An adequate solution from my point of view would be to have one central repository of "true" ratings and players (which could itself in principle be distributed, but this leads to all of the recognised problems of distributed memory coherence, on which there is a large literature - the problem of reliability with respect to node failures makes the problem even more difficult). A number of chess servers could exist around the world, which would take some snapshot of the players/ratings list from the rating server at the beginning of each day say (or each hour, or whatever). Then it would manage games and modify ratings locally to give an approximate picture of what your current rating is, but it would log the game scores of all rated games played (eg. players, win/lose/draw, match type). People playing multiple games against each other would allow big compression of this data. Then, every day/hour/minute whatever, the chess server would report its list of game scores to the central server, and once receipt was acknowledged it would grab another snapshot (or set of changes since many players ratings would presumably not be changed) of the player list & rating info. The comms could be accomplished by mail or direct connection or whatever, and it would be fault tolerant in that local servers would just keep logging game scores until they got acknowledgement from the rating server that they had been included in the global database. People's ratings would be very close to "correct" all the time, though it would of course be possible to see slightly different ratings on different chess servers if you jumped around quickly with respect to the update time. But I think it would be more than acceptable to me to have a "real rating" that lagged a bit and an instantly updated "provisional" one. The whole thing would seem to be trivial to implement (in principle anyway, there would be a good deal of code to write :-), and would remove the bottlenecks on player limits, etc that seem to irk everyone so. I think the key idea is not to try and build a perfectly fault tolerant, truly parallel/distributed chess server, but to go for something slightly simpler. Just my two cents worth - what do others think? Cheers, Pat (ICS: patrick, friend of Norm :-) --------------------------------------------------------------------------- Engineers think that theory approximates reality | Mathematicians never Physicists think that reality approximates theory | make the connection --------------------------------------------------------------------------- Patrick Surry, JCMB 2254, Edinburgh, Scotland e||) Edinburgh Parallel EMail: p...@epcc.ed.ac.uk Tel: (031) 650-5960 c||c Computing Centre --------------------------------------------------------------------------- Yibing Fan Oct 6 1993, 9:48 am show options Newsgroups: rec.games.chess From: y...@alchemy.chem.utoronto.ca (Yibing Fan) - Find messages by this author Date: Wed, 6 Oct 1993 14:15:26 GMT Local: Wed, Oct 6 1993 7:15 am Subject: Re: Distributed ICS, was Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse - Hide quoted text - - Show quoted text - In article p...@epcc.ed.ac.uk (Patrick Surry) writes: >I think a much simpler way to do this to start off with is to do "global" >ratings updates at a *much* lower frequency than once per game played. >An adequate solution from my point of view would be to have one central >repository of "true" ratings and players (which could itself in principle >be distributed, but this leads to all of the recognised problems of >distributed memory coherence, on which there is a large literature - the >problem of reliability with respect to node failures makes the problem >even more difficult). >A number of chess servers could exist around the world, which would take >some snapshot of the players/ratings list from the rating server at the >beginning of each day say (or each hour, or whatever). Then it would >manage games and modify ratings locally to give an approximate picture of >what your current rating is, but it would log the game scores of all >rated games played (eg. players, win/lose/draw, match type). People >playing multiple games against each other would allow big compression of >this data. Then, every day/hour/minute whatever, the chess server would >report its list of game scores to the central server, and once receipt >was acknowledged it would grab another snapshot (or set of changes since >many players ratings would presumably not be changed) of the player list & >rating info. The comms could be accomplished by mail or direct connection >or whatever, and it would be fault tolerant in that local servers would >just keep logging game scores until they got acknowledgement from the >rating server that they had been included in the global database. >People's ratings would be very close to "correct" all the time, though >it would of course be possible to see slightly different ratings on >different chess servers if you jumped around quickly with respect to the >update time. But I think it would be more than acceptable to me to >have a "real rating" that lagged a bit and an instantly updated >"provisional" one. >The whole thing would seem to be trivial to implement (in principle anyway, >there would be a good deal of code to write :-), and would remove the >bottlenecks on player limits, etc that seem to irk everyone so. I think >the key idea is not to try and build a perfectly fault tolerant, truly >parallel/distributed chess server, but to go for something slightly >simpler. >Just my two cents worth - what do others think? >Cheers, >Pat (ICS: patrick, friend of Norm :-) >--------------------------------------------------------------------------- > Engineers think that theory approximates reality | Mathematicians never > Physicists think that reality approximates theory | make the connection >--------------------------------------------------------------------------- > Patrick Surry, JCMB 2254, Edinburgh, Scotland e||) Edinburgh Parallel > EMail: p...@epcc.ed.ac.uk Tel: (031) 650-5960 c||c Computing Centre >--------------------------------------------------------------------------- Hi, guys, I read enough debates, but still couldn't find out WHERE IS THE new ICS location. I tried logon rafael.metiu.ucsb.edu 5000 several times and nobody is there except me. What happened to ICS? yibing (yfan on ICS) Michael Moore Oct 6 1993, 5:15 pm show options Newsgroups: rec.games.chess From: m...@cis.ohio-state.edu (Michael Moore) - Find messages by this author Date: 6 Oct 1993 19:07:26 -0400 Local: Wed, Oct 6 1993 4:07 pm Subject: Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse Paul Colley wrote in article : >In article <28pm8c$...@venera.isi.edu> schw...@isi.edu (Karl Schwamb) >writes: >> If some dedicated person >>doesn't code well but still provides great service -- who cares! If there >>are several servers to choose from then the net users will flock to those >>that provide good service -- not good code design. >Good point. So, what we want is a dedicated person who is willing to >write a server and give it away freely. Hmmm... I thought I did that. Perhaps you meant to say "willing to write a server" and give it away for free. I would have done that if I had had the time. I'd do it now, if I had the time. But to me the most important thing at the time was to prove that ICS was a good idea, and to give an example of its implementation. I figured if I could get people interested things would take off from there, and others would emerge to take on the responsibility ICS entails. I realize it didn't help that the server I wrote was bug-ridden, to the extent that Daniel Sleator finally determined it would be easier to rewrite than debug. But I was right. People did get interested, and others found it worth while to invest a good deal of time to making ICS solid and professional. I respect the time they spent doing this. I feel they are perfectly within their rights keeping control of the code. And even if I didn't, it wouldn't matter, because they are completely within their legal rights as well. Personally, I would prefer they let others use the code. It falls more in line with the reasons for which I wrote the orginal ICS. The internet has benefited greatly because people have selflessly given of their time and efforts so that the whole might benefit. Also, writing a chess server is not that hard. About the most complicated algorithms you have to worry about are checking the validity of chess moves. Handling all of your clients and parsing their commands is straightforward. Producing compact output for interface programs like GIICS is simply a matter of compacting data you already have. I wrote the original ICS in 5 days. About 30 or 40 hours worth of coding. To do the job right would probably take me about two or three times that. If someone wants to pay me I'll happily prove it :) >Count your blessings; without what has been done, you would have no >server at all. In the future you may have the public-domain code >you crave, but not until somebody writes it and gives it away. My understanding of the current situation is that there is a server at MIT but there is a player limit, and so people are unable to play chess there when they would like. If that's true, then why doesn't someone just compile the old server and let people use it while they can't get into the MIT one? The old server works well enough most of the time. After all, hundreds of people used it for several months before Daniel even began modifying the code. Isn't ICS with annoying bugs better than no ICS at all? -Michael Moore Richard "red" Nash Oct 8 1993, 2:39 am show options Newsgroups: rec.games.chess From: n...@visus.com (Richard "red" Nash) - Find messages by this author Date: Mon, 4 Oct 1993 14:14:58 GMT Local: Mon, Oct 4 1993 7:14 am Subject: Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse In article <28m29t$...@renoir.uni-paderborn.de> a...@renoir.uni-paderborn.de - Hide quoted text - - Show quoted text - (Axel Boldt) writes: > dabst...@pitt.edu (Darrin Bond) writes: > >Dan Sleator was approached first to move the server, which he declined > >after some discussion. He was then asked for just the EXE files to which > >he also refused saying that they (?) had decided that it was best that > >only one server be operating in North America. We now stand at an > >impasse. > >The Pitt group consists of three programmers who are attempting to > >rework the old code written by M. Moore and R. Nash. Two are dedicated > >to the C programming and one is doing the network interface. We are also > >attempting to rework the code for the program that allows ratings to be > >compiled and stored from many servers as if there were only one (Dan > >killed this idea long ago). > I hope that you guys will then put your code under the gnu license or > something, so that the sources become available and this ridiculous > situation ends. > (I thought the times with hidden sources were long gone!?) > Axel. Here! Here! I've been asking for the code for weeks! -- +---------------------------------------------------------------+ | Richard V. Nash | Visual Understanding Systems, Inc. | | n...@visus.com | Tel. (412)-488-3600 Fax. (412)-488-3611 | +---------------------------------------------------------------+ Axel Boldt Oct 8 1993, 6:25 pm show options Newsgroups: rec.games.chess From: a...@renoir.uni-paderborn.de (Axel Boldt) - Find messages by this author Date: 8 Oct 1993 06:01:38 +0100 Local: Thurs, Oct 7 1993 10:01 pm Subject: Internet should be free (was: ICS Relocation) Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse col...@qucis.queensu.ca (Paul Colley) writes: >In private Email, I told somebody the amount I was willing to pay to >use ICS. The response was somewhat idiotic, that "everything on >Internet must be free." That idiot was me, thank you. >I guess it was somebody who hasn't seen the costs of providing >Internet access to their site... These costs were paid with taxpayers money, i.e. with my money. Therefore I'm entitled to use these services for free, like I use libraries, highways or the police. And since I'm not the only taxpayer, everyone should have free access to the net. Have a nice day, Axel. Richard "red" Nash Oct 9 1993, 5:36 pm show options Newsgroups: rec.games.chess From: n...@visus.com (Richard "red" Nash) - Find messages by this author Date: Thu, 7 Oct 1993 13:27:09 GMT Local: Thurs, Oct 7 1993 6:27 am Subject: Re: ICS Relocation Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse In article John Anthony Chanak writes: > Just a few thoughts and experiences... > The idea of having a ratings system spread over a wide area is well and > good, but not very easy. Big software houses producing real products > have difficulty providing distributed database-type services. The > problem is easy to state, *very* hard to solve. Danny Sleator and I > talked about this for a long time about 6 months ago. We decided that > yeah, it would be neat. We also decided that it would be next to > impossible to do correctly. (Certainly beyond the time either of us > could put into it) Actually it isn't that hard to solve. Simply update the central database in batches, say once a day or so. There is no reason to have real time ratings updates. The ICS nodes can keep approximate ratings the same way the do now and get updated from the central database every so often. This is the kind of thing that could get done (I am willing to do it) if the ICS monopoly could be broken and the code was released. Having one person dictate to the rest of the network chess community the way an ICS should work was never intended when the ICS was started. I hope Mr. Sleater sees the light and puts the code into public domain. I would suggest an ICS boycott but I know we chess addicts couldn't stay away. -- +---------------------------------------------------------------+ | Richard V. Nash | Visual Understanding Systems, Inc. | | n...@visus.com | Tel. (412)-488-3600 Fax. (412)-488-3611 | +---------------------------------------------------------------+ John Richards Oct 11 1993, 5:56 am show options Newsgroups: rec.games.chess From: j...@r-cube.demon.co.uk (John Richards) - Find messages by this author Date: Mon, 11 Oct 1993 08:34:36 +0000 Local: Mon, Oct 11 1993 1:34 am Subject: Re: Internet should be free (was: ICS Relocation) Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse In article <292s7i$...@renoir.uni-paderborn.de> 6500a...@ucsbuxa.ucsb.edu writes: >col...@qucis.queensu.ca (Paul Colley) writes: >>In private Email, I told somebody the amount I was willing to pay to >>use ICS. The response was somewhat idiotic, that "everything on >>Internet must be free." >>I guess it was somebody who hasn't seen the costs of providing >>Internet access to their site... >These costs were paid with taxpayers money, i.e. with my money. >Therefore I'm entitled to use these services for free, like I use >libraries, highways or the police. >And since I'm not the only taxpayer, everyone should have free access >to the net. Gosh. That's jolly decent of you. Now all I have to do is tell British Telecom they shouldn't charge my company for dialling up to get Internet access, and Demon Systems, who provide us with the link, that they shouldn't be charging us for the connection (reasonable as the cost is). Tell you what, how about if Paderborn University provide us with the link and we make a reverse charge call? That should help get some of my taxes back. John Richards Paul Colley Oct 12 1993, 2:50 pm show options Newsgroups: rec.games.chess From: col...@qucis.queensu.ca (Paul Colley) - Find messages by this author Date: Tue, 12 Oct 1993 21:06:29 GMT Local: Tues, Oct 12 1993 2:06 pm Subject: Re: Internet should be free (was: ICS Relocation) Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse In article j...@r-cube.demon.co.uk (John Richards) writes: >>And since I'm not the only taxpayer, everyone should have free access >>to the net. >Gosh. That's jolly decent of you. Hey! I talked to him first! AFTER he pays the costs of Onet in Ontario and gives free access to all of us here, he can use any left over money for the UK. Suckers are strictly first-come, first served. Take your place in the line. (:-) for the humour-impaired). - - Hide quoted text - - Show quoted text - Paul Colley University: col...@qucis.queensu.ca Home: pacol...@ember.uucp watmath!ember!pacolley +1 613 545 3807 "Quantum Mechanics: The dreams stuff is made of" - Ken Burnside Randell Jesup Oct 12 1993, 8:21 pm show options Newsgroups: rec.games.chess Followup-To: alt.internet.services From: j...@cbmvax.cbm.commodore.com (Randell Jesup) - Find messages by this author Date: 13 Oct 93 02:17:31 GMT Local: Tues, Oct 12 1993 7:17 pm Subject: Re: Internet should be free (was: ICS Relocation) Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse 6500a...@ucsbuxa.ucsb.edu writes: >>In private Email, I told somebody the amount I was willing to pay to >>use ICS. The response was somewhat idiotic, that "everything on >>Internet must be free." >>I guess it was somebody who hasn't seen the costs of providing >>Internet access to their site... >These costs were paid with taxpayers money, i.e. with my money. >Therefore I'm entitled to use these services for free, like I use >libraries, highways or the police. >And since I'm not the only taxpayer, everyone should have free access >to the net. I noticed you posted from a German site, BTW. Whose taxes are we talking about? A lot (most) of the internet is NOT publicly funded anymore. Sure, ARPAnet was and NSFnet is publicly funded, but an awful lot of the net is either commercial (like how I'm on the net), or only indirectly or partially publicly funded. Also, I don't know about Germany, but just because something is (partially) publicly funded doesn't mean it's free, unless you live in a very socialist state. National parks are publicly funded; but you need to pay to get in. The space shuttle is publicly funded; that doesn't mean that anyone who wants to put something in space gets to do so for free. The real answer is that the net is no longer primarily purely publicly funded. It's too big. There are an immense number of companies on the net, many universities (most of which probably do not get direct funding for being on the net), major commercial hubs like psinet and uunet, intercontinental links, even local libraries and schools. (Chester County Net is starting here, which will link the libraries, schools, university, and county government to the internet, and eventually local companies and even homes. It'll be free for a while, and then they'll figure out how much they _need_ to charge for it - it costs money for an internet connection.) Even if net "access" were free, the hardware to run the chess server isn't necessarily free. If someone provides something for free, great. Don't assume the net/government "owes" it to you. We're way off the topic of chess, let's drop it. -- GNU Emacs is a LISP operating system disguised as a word processor. - Doug Mohney, in comp.arch Randell Jesup, Jack-of-quite-a-few-trades, Commodore Engineering. j...@cbmvax.commodore.com or rutgers!cbmvax!jesup BIX: rjesup Disclaimer: Nothing I say is anything other than my personal opinion.