GameKnot Forum
[ mod this thread ] << - < - > - >>
| From | Message |
| duchess 7/13/2003 10:08:44 [ report this post ] |
 Subject: GK rating systemMessage: The GK rating system has been discussed here in more than one thread. I never posted there because always found lots of people with ideas so close to mine that I felt me well represented. Most of opinions were against the actual GK rating system but... they got no result, because everything remains the same. That's why I'm opening once again the discussion. I think the main problem is this: We should not colect points playing against an oponent, for instance, 800 points lower than our rating. Those games should be automaticaly unrated. If this remains cyrano, for instance, soon will have 3500, or 4000 points. That is ridiculous. My CC rating should be about 2200-2300 and my GK rating is reaching 2700(!). That's why I stoped playing with low rated players. What I tell them is that, as soon as I can play with them without collecting points I will do it. One of the problems this is bringing is that each time it is more dificult for me to enter or start a minitournament because every day there are less players inside the 400 points needed to enter. My proposal is this: Points colecting with a win over a player rated up to 300 points under our rating - 8 points 300 to 400 points under our rating - 3 points 400 to 500 points under our rating - 2 points 500 to 600 points under our rating - 1 point more than 600 points under our rating - unrated Also, before the begining of the new system, all the ratings over 1200 should be compressed, beeing multiplied by a coefficient like, for instance r1=r(0.8+(3000-r)/10000) where r1 is the corrected rating and r the old rating. With this a player with actual 3000 points will have 2400, with 2000 will have 1800 and with 1500 will have 1425. That would diminish the distances and then we could start a new rating system closer to the ICCF Elo system. Sorry for my poor and confusing english but I think you can understand me. |
| parrvert 7/13/2003 10:35:32 [ report this post ] | I agree Message: that the rating system should be changed but I don't think that it should be changed so your rating is 2200. GK rating system should show relative strengths within the gameknot community not try to match other rating systems. In a way I think it would be better to have a system where the highest ratings were say 1000, to make it clear that the rating system does NOT match FIDE or ICCF Elo. |
| mlazar 7/13/2003 11:02:10 [ report this post ] | unrated games Message: sound great,but would the elite players here other then cyrano play us lowly patzers? Thanks again cyrano for your class and style.. Mike |
| duchess 7/13/2003 15:05:58 [ report this post ] | answer to mlazar Message: Yes, I am ready to play with any low rated players if the game is unrated. And I think the games, that way, may be much more interesting because then we can freelly coment the moves and give advices, because there will be no rating problems. |
| mlazar 7/13/2003 16:09:11 [ report this post ] | that would be Message: great...Hope it can somehow happen... |
| silverwolfwsc 7/13/2003 17:09:19 [ report this post ] | unrated Message: mike did make a comment that said unrated games were coming... |
| kumpan 7/13/2003 22:20:57 [ report this post ] | The most elegant way Message: to make an end to all these discussions is to implement the ordinary ELO-System what is available everywhere. This would show us at most the real strongness of a player. By the way Pedro, what happend in your system when a low rated player plays draw or wins against a much higher rated? The opportunity to play unrated games would be a very good one. Gives the possibilty that a master can play prepared games with his students. Games where the master makes moves what give to the student the possibility to find e.g. winning positions etc. |
| pawnstar 7/14/2003 07:17:55 [ report this post ] | - Message: With cyrano over 3000 the time has finally come for a change. Especially with Mike contemplating blitz chess on the site where the rating inflation will be ridiculous. |
| cuthbert 7/14/2003 07:22:34 [ report this post ] | I agree Message: with Duchess - it's time for a change!!! Most of the GK users want a new rating system - the effort is rather low... compared to Java based Blitz games ;-) Cuthbert |
| lexherman 7/14/2003 08:12:20 [ report this post ] | i agree Message: I had promised myself not to react anymore about this subject. So many times this has been asked for...so many ideas have been given. Nothing is done about it.I think a lot of players don't react anymore about this subject because like me they have given up hope. So...this is really the last time : GK MUST REPAIR IT"S RATING SYSTEM! |
| duchess 7/14/2003 13:06:23 [ report this post ] | answer to Message: kumpan, who asks me �...what happend in your system when a low rated player plays draw or wins against a much higher rated?� In my system a game with a player more than 600 points under our rating should be definitively unrated, so... nothing happens, we just have a good time playing chess. Points, only playing with mates more or less close to our rating. |
| optimist 7/15/2003 13:40:01 [ report this post ] | Leave it as it is Message: Mike has more important things to do. Well, one slight change. Opposed to winning 8 points when drawing against a player with more than 400 points above your rating, you should also lose 8 points if you draw against a player with more than 400 points below your rating. I think this simple solution will also put an end to this rating inflation. No further adjustments, as I said before, Mike has more important things to do. For instance the analysis board suggestion from nosredla in another topic here (gameknot.com/fmsg/chess3/1965.shtml) |
| bogg 7/15/2003 15:36:32 [ report this post ] | optimist Message: IMO there is no higher priority than putting a reasonable rating system in place. A chess site that does not have a reasonable rating system must be considered a bottom feeder as compared to those that do no matter what nice, but unnecessary features it has in place. That is why I am not going to continue playing at GK when I have finished my games in progress. |
| mlazar 7/15/2003 16:39:55 [ report this post ] | Bogg Message: I hope you can find that site. You've been here 3 months and have played 3 games. You are the most important person you will ever meet.Why not ask yourself for an autograph? |
| nosredla 7/15/2003 17:19:48 [ report this post ] | I don't care ... Message: I play chess casually and for fun. I don't care what my rating is and, to be honest, would rather not have one and not know my opponent's. The existence of the ratings introuces a number of distortions to my simple pleasure of choosing (or being chosen by) and playing against an unknown opponent. That said, I understand perfectly why the ratings are important to many here and the concern that the present rating is, as someone else recently remarked, more like a "score". However, from my perspective, Mike's time could be better spent on other things. |
| weenink 7/16/2003 01:02:35 [ report this post ] | It is important Message: To change the rating-sysyem to a good one (ELO) imo. Ratings mean something on this site, they are important in teamplay and in mini-tournaments, and therefore the ratings should reflect the playing-strength of a person, which is of course not the case with the current system. Therefore, I agree with lexherman and disagree with optimist. As to the idea of duchess; why to replace a very bad system with a bad system? If you change it, you can as well change it to somthing good (ELO). So I most agree with kumpan :) |
| omus 7/16/2003 01:36:19 [ report this post ] | The ratings Message: are all relative to each player on GK. If someone is 100 points better than me I will generally find it hard to win or draw. |
| bogg 7/16/2003 08:13:24 [ report this post ] | omus Message: The closer an established GK player's 'rating' is to 1200 the more closely it represents the relative playing strength of the player. GK ratings for established players become worse and worse as the player's 'rating' gets higher and higher becoming nearly random at around 2200. The main argument against changing the rating system voiced in the forums is that Mike can better spend his time elsewhere. This is inacurate on several points: 1) the Elo straigt line aproximation system or a similar system is a simple equation. In most programming languages it would take 2 lines of code, that is less than the system in place now! Written in any language we are talking about a couple of minutes work. 2) others, myself included have offered to write the code for any of several rating systems for Mike, all he would have to do is install it. When I started playing here I was under the impression that the GK 'rating' system was going to be changed. When I offered to write the code for Mike I was notified that this was a closed issue, that he liked the system, that the system worked as he desired, and that it was not going to change. That is when I decide that I was not going to continue playing on this site after finishing my games in progress. |
| zoobrenok 7/16/2003 08:37:05 [ report this post ] | let me ask you a question Message: Assume that Kasparov only plays people with 1500-1700 rating and wins 100% of all those games. What his rating would be? - I am not an expert on ratings, but the problem is not in the system, problem is that people are not playing opponents of their level. And this is an area where trying to fix something 'from top' (by Mike) won't help - there should be change in attitude here on GK for ratings in 'high places' to become more realistic. But then again - GK is not FIDO, so as omus said: ratings only make sense withing GK community. Here is an example: I find it really hard or almost impossible to win against someone rated above 1800 and I play opponents or more or less same rating. Therefore my rating is ~1600. For example, a player has no problem winning against opponents at 1800, so his rating is higher - let's say 2100 And bogg, has no problem winning against people above, so his rating should be even higher. And then we have cyrano, who simply do not loose to anyone, while playing against everyone - so he should be top rated player. And of course if Kasparov comes here - he should be 4-5 thousand rated player after 5 completed games :) This is GK rating. Leave it alone! And if you don't beleive me - check results of Mikes poll. Majority of people think same way |
| duchess 7/16/2003 10:23:42 [ report this post ] | Please Message: zoobrenok can you tell the link to the poll about this matter? I can't find it. Thanx. |
| riduan77 7/16/2003 10:28:19 [ report this post ] | Don't need Change of ratings system. Message: I just don't understand the idea of changing the rating system at all.. Mike might have different idea for his wonderful website/rating.. he might want to have different rating system which is far much different from FIDE or what ever Elo ratings that existed in this EARTH.. HE simply don't want it to be same like other ratings.. he wants a rating that shows GK uniqueness.. a rating which shows GK own brand of rating.. So what if the ratings balloons to 5000++, it will be much better.. as it will shows that GK rating is definately not the same with FIDE rating!! Anyway i like the idea of promoting GK own brand of rating system.. it's the only way that people around the world knows that GK has its own brand of rating system which will be definately different from FIDE ratings.. So mike, stick to this rating system, it will benefit this chess website even more as it grows bigger & bigger with time.. regards, riduan77 |
| bogg 7/16/2003 10:32:36 [ report this post ] | zoobrenok Message: I can not find a poll referencing the GK rating system either. Additionally, I don't see much point in a poll when discussing a problem that doesn't really impact the majority of participants. Just because the 'rating' system more or less works for the majority of participants is not a valid reason for not implementing one that works for all participants. Instaling a real rating system that works for everyone not just the players close to the norm is a trivial task and should be done. |
| zoobrenok 7/16/2003 23:31:13 [ report this post ] | poll Message: I am talking about this poll: gameknot.com/vote.pl?l Just analyze it bogg, if it works for majority of folks, why to change? ;-) |
| lexherman 7/17/2003 01:19:19 [ report this post ] | I analyzed it Message: And maybe i overlook something but i can not see any opinion about the current rating system. Please explain zoobrenok |
| bogg 7/17/2003 01:34:03 [ report this post ] | zoobrenok Message: I don't see any mention of the 'rating' system in that poll. You asked, 'if it works for majority of folks, why to change? ;-)' I answer with another question: As putting a rating system in place that works for everyone would be a trivial task why shouldn't GK have a rating system that works for everyone? The fact that the 'rating' system here at GK meets the standards of 99% of the people doesn't seem relevant to me. If it fails completely for 1% of the people, and it does, why not replace it with something better? Using myself as an example: I expect my first provisional rating to be about 2700 which should indicate that I am slightly better than florinserban when I would guess that he is at least 100 points stronger than I am. |
| duchess 7/17/2003 02:25:09 [ report this post ] | The arguments Message: against GK rating change that we got in this thread are very weak. They have 2 main apearence: 1 - This rating system is a characteristic of our beatiful club; it �shows GK uniqueness�. In other words, maybe is a bad system, but makes us diferent so is beautiful, so is good. 2 - This works well for me so, why to change it? As I told you, and bogg tells it better than I because he wrights a very good english, the GK rating system is worse as the rating increases. I would like to see here, defending this system, an high rating player. I also found nothing about this matter at the poll recommended by zoobrenok You cannot say �this is just a rating� because this rating regulates lots of events here at GK and it regulates them �very badly�. In my way of feeling this club, the worst consequence is that mates who should be 200 or 300 points far from me are now 800 points away, and I feell me alone :o( I am diminishing my games number and probably, when I reach the zero, I'll stop playing. |
| omus 7/17/2003 03:31:40 [ report this post ] | if Message: if the rating system was to change then imagine the problems associated with changing every players rating a a given time. For example: some players who have fought hard and feel a sense of achievement to be in the top 100 may find themselves dropping in the rankings overnight. |
| omus 7/17/2003 03:34:10 [ report this post ] | also Message: What is more important to players: their position in the GK hierarchy or their rating? |
| duchess 7/17/2003 05:00:45 [ report this post ] | No Message: omus. The top 100 would stay just the same. Only the ratings would be compressed to be more real. Maybe their positions would change in the future, because the rating would be more fair. But now, only the exagerated ratings would be diminished, keeping the relative position. |
| lexherman 7/17/2003 05:12:22 [ report this post ] | Omus Message: To your first message: I think everyone can keep it's rating for start, if the new system is implemented the rating system will gradualy correct ratings.(correct me if i am wrong) Second question: I don't understand , explain. We just want to repair the system which causes in the top inaccuracy in indication of relative strenght. That is all.For most players things won't change very much. But a lot of top players get annoyed about this ,that nothing is being done about it or a decent rational reply is given. Mike can't leave this issue alone....a lot of players feel they are not taken seriously....and some of top players might decide to leave. |
| omus 7/17/2003 06:19:51 [ report this post ] | lexherman/duchess Message: "if the new system is implemented the rating system will gradualy correct ratings.(correct me if i am wrong)!" If everyone kept their rating it would take an age for this gradual "correction". i.e. would the top few ever then be overtaken ?? (presumably you guys want a tougher system....no points for beating players x points below the winners rating for example). to explain my second query... I am just wondering whether players find it more pleasing to be in the top 10 say, or are more concerned about their rating (2000, 1200 or whatever). duchess... o.k. seems reasonably fair enough |
| lexherman 7/17/2003 06:50:34 [ report this post ] | Omus Message: Thank you for explaining. Now in reply to your question.Ofcourse i try to get as high as possible in the ladder but that has for me nothing to do with my wish for a change in the system. What i find most annoying in the system is the inaccuracy of the relative difference in chess strenght expressed in points. To make it more specific: my difference with the top player here is about 650 points. Ofcourse i am not as strong as this player but the difference is not 650 points. Within this system his theoretical chance to win from me is 16:1. In a few month the differences because of this system will be so big that theoreticly the top can beat players like Duchess, Florinserban 16:1. So i want to ask you Omus and others this question: DOES THE DIFFERENCE IN POINTS AT THE TOP OF GK REFLECT REMOTELY THE RELATIVE difference in STRENGTH BETWEEN PLAYERS??? IF NOT: Is this not an important issue for a respectfull chess site/community ?? |
| omus 7/17/2003 08:04:21 [ report this post ] | lexherman Message: O.K. its not perfect - and not just at the 2000+ end of the range - but it is certainly not worth players stating they will no longer play here because of it. The webmaster is one of the most reasonable people I have ever 'met' and no doubt he is looking at this forum feedback. Of course any site is continually evolving and well reasoned discussion is the answer to improve this great chess arena. |
| lexherman 7/17/2003 08:32:05 [ report this post ] | Omus Message: In reply to your answer. I am glad you admit it is not perfect, but i think that is an understatement . In a few cases my estimation of the incorrectness is about 500 points! I agree with you we should be able to discuss this in a rational way as to improve this site, which is a very good one.I have seen many suggestions and ideas pass by and there is no response at all. The discussion is very one sided up to now. Mike is no doubt a very reasonable person, but on this point i would expect him to make a statement. I think the way a lot of players who put this issue on the table are neglected and this is not respectfull. I think this lack of respect is more important even then the issue itself and thereby may be a reason to leave. I hope this will not happen. |
| bogg 7/17/2003 11:30:43 [ report this post ] | omus Message: The correction of the existing 'ratings' would not take that long with a good rating system in place. Depending upon the K variable used in the equation it should only take about 50 games. Using Cyrano as an example, in a previous post I showed that he is about 400 points overrated by GK standards. That means that with a real rating system when he plays florinserban, where he is about a 3:5 underdog in an eigth game match instead of losing 16 points he would lose about 160 points, again depending upon the K variable. |
| zoobrenok 7/17/2003 18:52:50 [ report this post ] | Message: i think someone mentioned before that problem is not in the system, but in number of games people play and in win %. I pretty much sure that if Kasparov would play hundred games a month, his rating would grow higher and higher. Anyway, my point is simple - if you folks care about rating, play stronger people :). Don't accept rated challenges from people whom you can eat instead of breakfast and yet be hungry. GK has a very simply rating system that can be explained and understood by everyone. KISS - keep it simple stupid ------------- About poll: yes, there is no question about rating system there. But to me results of that poll mean that even if we remove ratings all together, people would still play same number of games here. In other words - majority of people do not pay attention to rating and because they are happy right now, they might not appreciate system change. And again - if 1% is unhappy, we do live in so called 'democratic' society, don't we? I personally would not appreciate rating system change - I have BA in Applied Math, but I hate to use something more complex for my chess then +/-16 or 8 or 1 or 0 :). 1. How many of you play at Yahoo? 2. What is your rating there? 3. How many games have you completed recently there? After answering 3 questions above, tell me why you do not like Yahoo rating system :) --------------- maybe arguments to keep this system are not strong, but I also do not see strong arguments to change it: if you think it is unfare to get points by winning against weaker players, simply don't play them. |
| bogg 7/17/2003 21:50:20 [ report this post ] | zoobrenok Message: You seem to be completely missing the point! Fairness has nothing to do with it. The point is that you can't determine the relative strength of players based upon their GK 'rating'. PS Democracy doesn't mean the majority rules it means what is fair for all. In other words if 99% of the people aren't affected by something and 1% are, only the 1% have a say. |
| lordnguyenvo 7/17/2003 22:08:03 [ report this post ] | zoobrenok Message: i found that your sentence doesnt agree." majority of people do not pay attention to rating and because they are happy right now, they might not appreciate system change".talking about logic here,if "majority of people" dont pay attention to the rating (i mean REALLY)then they dont care about the rating change.If they do care about the change,it means they do pay attention to the rating system. "GK has a very simply rating system that can be explained and understood by everyone." i agree it is simple and thus easy to understand but thats probably the reason why it contains problem(everybody knows what the problem is). "I have BA in Applied Math, but I hate to use something more complex for my chess then +/-16 or 8 or 1 or 0 :). " Its not like u have to calculate it by yourself anyway(they have devices u know). "maybe arguments to keep this system are not strong, but I also do not see strong arguments to change it" I think a lot of people already state that the reason for changing is to make a clearer rating system in which it reflects the relative strangth between players. " if you think it is unfare to get points by winning against weaker players, simply don't play them. " its not that simple like u said.even when u dont do it,somebody might,and then u might beat that person thus all the ratings are inaccurate.and its hard to know whos weak to avoid them when the rating system isnt showing the real relationship between a player's ability and his rating. |
| zoobrenok 7/18/2003 09:42:22 [ report this post ] | Message: but if what 1% says negatively affects 99%, then people should look into other options. And I do think democracy is about majority rulling, otherwise I don't get it - who needs it if one person does not like something he can change it even when rest of world likes it?.. And don't we get President who was voted by majority? - majority rules :) Anyway. Back to GK related topics :). Rating system is good and ain't going to change. Let me ask you this: would you rather have rating system changed or some other feature added to the site? We all know that Mike is not what some think webmaster should be and does not implement hundreds of changes every other month, instead he gets just few every now and then. And once again - history shown that about 700 users come to Forums more or less regularly. This brings my back to the point that people like the way things are going and therefore don't care right now, but they will care if you make a change. In software they say: 'if it ain't broken, don't fix it' In my opinion this rating system does provide accurate player placing. And changing it won't help you guys to achieve what you want because of the number of games people play -- there still will be high ratings for not so strong folks simply because they will keep on winning against much weaker opponents. Look - 18 months ago everyone complained that dj222 is way to strong. Then they complained that brunetti is way to strong because he has high rating and plays 300 games. Then people complained that cyrano is not playing fare. My point is that it is not high rated player who is strongest, therefore whatever rating you pick - it won't make a difference. Who is Chess World Champion? Is he a guy with highest rating outthere? - No. So, rating system does not mean that much. But - just for you guys ;) -- we should have GK-wide poll here with just one question and two possible answers (one to pick): Do you like current rating system? Yes No And if majority answers Yes, then let's talk about something else :). BUT --- this poll has to be sponsored by Mike and be system wide, not just for Forum visitors in order to get accurate results |
| hell-on-wheels 7/18/2003 09:49:02 [ report this post ] | Zoobrenok Message: The best rating system around is the Elo system, which should have as large a data base as possible, as it is a stastical system. Most groups using it require 50 games to establish a permanent rating. How long does that take? Many groups will rate an entire tournament at once (such as a mini-tournament.) But many players want to play "coffee-house" chess, one game at a time. I don't think that a sosfisticated system will work. The simpler the better. |
| bogg 7/18/2003 10:30:14 [ report this post ] | zoobrenok Message: You asked 'And don't we get President who was voted by majority?' The answer is no we do not get a president because the majority vote or him. Kennedy and Bush recieved less votes than Nixon and Gore. Back to the main topic. There are two points to consider when discussing a change to the system: 1) the difficulty of implementing it. 2) the desirability of changing it. To point 1. It has been explained that the task is trivial, mere minutes. It does not even vaguely compare to adding a new feature. We are talking about 2 or 3 lines of code. My membership alone would make Mike's pay for the effort several hundred dollars per hour. To point 2. I haven't read a single message stating that someone liked the current system, only that it was good enough and that the person commenting thought Mike could better spend his time on adding new features. Hence my statement of 99% vs. 1%. It isn't that 99% like the current system and 1% dislike it. It is that 99% don't care and 1% dislike it. I actaully would be surprised if the numbers are not much closer. I personally find it difficult to believe that anyone could prefer a 'rating' system that doesn't rate the participants to one that does. PS It is not necessary for someone to win even one point for beating someone significantly worse than themselves. As a matter of fact, it is ridiculous to recieve anything for beating someone 800 or more points below you as you are a 100% certainty to win the game, if the rating system in place is gives reasonable relative ratings. A player 800 points stronger than his opponent should still be a big favorite giving piece odds. |
| zoobrenok 7/18/2003 13:56:49 [ report this post ] | hell-on-wheels Message: Maybe it is just me ;) One of the reasons I joined GK is because rating system was simple to understand (after first 20 games of course) - I played at Yahoo before and you never know what your rating will be after game is over. Here I can do math calculations in advance. I think you pointed it correctly - GK is like 'coffe-table chess' :) |
| aqeel 7/19/2003 09:36:23 [ report this post ] | I think duchess is correct! Message: A rating of 4000 is ridiculous!!! un realistic.. due to the same reason the low rated players dont get a place some times because the veterans are always climbing up and up! |
| hell-on-wheels 7/19/2003 09:38:48 [ report this post ] | Zoobrenok Message: If a player is rated 800 or more points than his opponent, he offers his opponent a handicap, he would be expected to swindle his way to a win. In such a case, the game should not be rated at all. Thus there will be no feeding off an inferior player for that one point. But an inferior player should be able to play a superior player to improve his play -- if he will learn from his mistakes.But the difference should be limited to 100 or 200 points for rating. Then the basic system for ratings can remain simple. The other problem is underrated players feeding on players with higher ratings. |
| ckytep 7/21/2003 08:00:15 [ report this post ] | Well... Message: Is there any other place, similiar to gameknot, that uses the standard Elo-rating system (or even Glicko)? If not, please someone do it quickly.. I think this GK rating system is stupid. Great place to play chess though.. |
| cuthbert 7/21/2003 08:46:55 [ report this post ] | Change it Message: Like many other said before in this thread: It is obvious that the "GK rating system" is NO rating system. !!! A rating system should be able to predict chess results !!! So if two people with the same rating are playing a certain number of games against each other, the outcoming result should be the same number of wins, draws, lost for each. Or to make it more concrete - when some like Cyrano (3054) plays others like Florinserban (2666), mestrinho(2743), duchess(2686), cairo(2670), mateintwo (2755) (just to mention a few of the top players) everybody expects Cyrano to win at least 75% of his games against them - but he doesn't. (Remember 400 pts. difference are expected to win to 100%!!!) This does not mean that Cyrano is a bad player - it just shows us that the so called "GK rating system" is NOT A RATING SYSTEM!!! I do not care wether the rating system is called ELO or SONAS or whatever - but please let's get rid of the current one. Cuthbert |
| ckytep 7/21/2003 09:01:59 [ report this post ] | Agree with Cuthbert 100% Message: Yes! |
| chess_champion 7/21/2003 09:37:30 [ report this post ] | me too... Message: i also agree with cuthbert. |
| sjoroos 7/21/2003 14:21:31 [ report this post ] | Well, well, well... Message: I know i PROMISED not to care anymore about this topic, but as others are starting to realize what i've been trying to say for a year and a half i will give all "doubters" an example that simply tells why this "points" system sucks !! When i started here the top players were mostly at 2000-2300. I started out as many newcomers playing games against other 1200 rated players before i realized that my initial rating would be far to low compared to equal players if i continued that way. Therefor i challenged higher rated players and ended up with a rating of 18-1900 after 20 games (don't remember exactly). Now i am playing in some tournaments and also taking challenges from other good players which surely has made my game improve since i started here. As i do NOT play thousands of games against poor opponents to boost my rating i have established around 2250-2300 which probably is about 300 points more than i would have o.t.b in ELO rating. So far everything is good, but consider the following and you will clearly see the problem with this system: Players will continue to boost their rating and in time someone will reach say 3500. Let's say someone registers as a new member and plays 20 games against the 3500 rated player or someone in the same class. After LOSING ALL 20 GAMES, that new player would end up with a rating of 3100 !!!!!! That is about 800 points more than me, meaning that normally i would have no chance against him!! Can you seriously stand for your point saying that NOTHING is wrong with that system ???? A RATING system (not points) MUST be built in a progressive way, meaning that the higher you get the harder it must become to earn more points. A champion of 2003 should be able to be compared against a champion of 2005, or do you really think that the highest rated player at GK 2005 will be DOUBLE as good as todays highest rated ??? So, you that try to tell us that this system is OK, stop and think once more !! As i see there are only 2 good options for GK: 1. Get a REAL rating system 2. Get rid of the whole thing |
| hell-on-wheels 7/21/2003 18:49:02 [ report this post ] | Rating inflation Message: is a problem with systems which simply compare two players by adding points. If the higher rated player "loses" 5 points on a loss, and the lower rated players get 6 or 7 points on a win, obviously a rating "inflation" occurs because 1 or two points are added to the total ratings of both points. And over the years, you cannot compare players from 1995 to players from 2003. Periodically, an adjustment must be made. A good system automatically adjusts as you go. |
| zoobrenok 7/21/2003 21:28:09 [ report this post ] | hell-on-wheels Message: So, could it be that the answer to some of our problems is in limiting how high player rating will be after 20 games (or once he/she stops being a prvisional player?) For example, if during first 20 games I can only play players with rating of no more then 400 points comparing to mine, it would limit how high I can get early on. In addition, if I do play a person who is rated much higher then me, then during first 20 games when calculating my rating all opponents with gap of 400+ points will be assumed as myrating+400 points? -- This way it would be harder to get initial strong rating and as far as I can see, our main problem is that formula for provisional rating does not work for strong players who are relatively new to GK? (like bogg for example) |
| Maximum number of reply messages has been reached | |