tasuki, you have every right to be biased, since you have created an excellent site that makes people want to try themselves out in games they otherwise would not play. but.
but to claim is that go, go13 and go19 (or the hexes and reversis) are completely different games is ridiculous.
just look at the monster rankings. in the top 20 there are only a handful of players who would not double or triple their points only because they are good at go (hex or reversi). such all-round players are rare: ray garrison, miuek, t-devil, for example.
but there any many more players who are mediocre besides their top-heavy go performance: xaver gubas and bela nagy are prime examples. beyond the top 20 DZ, gambatte, 12, sorin gherman have a high monster ranking also only due to their go performance. so it is not just nao
Bloke, you have played only a little go here (have played more anywhere?). But I think you have played enought, to at least agree with tasuki with 9 and 19go, aren't they completely different games? If you don't, play more games. About the hexes and reversis, I can't tell, I've only played 13x13 hex, and no reversi.
Ruben, what is "different games" is simply a question of definition. In this context, I prefer to consider games with identical rules, but with different board sizes, to be different games. However, I know very well that there are huge differences between playing Hex13 and Hex19. However, The difference between Hex13 and StreetSoccer is far, far bigger. Again, it's a matter of definition.
Heh Marius you fooled me with your first message. Sure, the difference between let's say, SS and Go9 are bigger than Go9 and Go19, but the skills you use when playing Go19 are not the same you use when you play Go9 (local tactics vs global strategy, things like that).
bloke: Xaver has over 1100 points (that's about a third of his total points) from those games he's "mediocre" at.
I wonder, why is no one complaining that street soccer is included too? It is a completely random game, so the ratings are obviously distorting the monster rating
And once again - as soon as I have the time, I'll make it possible for people to make their own custom monster ratings, based on the criteria they choose. :)
Looking forward to the custom site, tasuki. You're doing a great job! Anyway, I must protest that street soccer is "a completely random game". Sure, chance plays a part - a significant part, even. But there is still skill involved. Stochastic games can be even more complex than other games, as you have to compute probability distributions rather than "just" a game tree. However, due to the random element, rating differences will likely be smaller in random games (SS and Einstein) than entirely deterministic games (at least as long as the deterministic games are sufficiently complex - in Tic Tac Toe, every game would end in a draw, of course).
Ruben: Sure, there are differences between Go9 and Go19 - noone disputes that. But like I said, what constitutes a "different game", is simply a matter of definition. There are many reasons (good ones, in my opinion) why games with identical rules, but different board sizes, should NOT be considered different games, but opinions differ on this matter.
I kind of overlooked the fact that littlegolem removed < and > from my post... It should have been:
I wonder, why is no one complaining that street soccer is included too? It is a completely random game, so the ratings are obviously distorting the monster rating
I'm sorry for the inconvenience and I hope it's clearer now :)
It seems so obvious to me: Every strong Go player is strong at all the three sizes of Go, even if the games are quite different. Every Go novice is weak at all three. So of course in Tasukis rating Go has weight 3 and the honorable game of street soccer has weight 1.
tasuki: I bet that playing streetsoccer if difference between dices are plus minus 10, I'll defeat you 8 from 10 times. Playing with tigis you may win 1 of 10.
in the 2nd MOLG they are NOT considered different for a very obvious reason, i suppose. the reason why i have not signed up for each of the MOLG tournaments is my non-existent go game. but i like the new rules so i will sign up for MOLG3.
I have to thank tasuki for the excellent page: I started playing all sorts of new games, and I was very flattered to be on the 0st page (it is really the first but called zero). I was surprised how easy it has been to move up from position 45 to 28. I don't think I'll come much higher than 20 though being a bit over rated in Hex and because of the quality of the opponents up there.
So, I sacrificed yesterday's evening and added the possibility to ( insert fanfare) create your own variations!
I already added two variations (the default rating is a variation for technical reasons, and the new one is "weighted" with go counting only 1/3 and hex 1/2).
New variations, opinions and bugreports are welcome. :-)
Whenever a new variation is created, a new indexed column is added to the player database. So I check for people trying to create a variation with a name of already existing column, and of course I forgot to check case-insentitively.
So it happened, that when someone tried creating a variation called "GWG" (there's a possibility to sort by game, btw, no need to create a new variation for that, you ******), it wrote the data over the existing gwg column. So I'm now updating the whole database. Now, when creating a new variation, it checks for every column, so in case some idiot decides to call his variation "last_updated", it will not (hopefully) fuck up my database.
Geez, this made me so angry... (and yes, I know it's my mistake, but I really don't have the time and energy to deal with this so please just stop breaking my stuff, okay? :-)
Great work Tasuki! People should not believe in one correct order by strength of the players on Littlegolem. I like "weighted" most, but thats just my personal taste.
I really appreciate your effort, tasuki! This is becoming a great addition to Little Golem.
I have one very small suggestion for an improvement, if you should find the time: In the "weighted" variation, new ratings for the weighted games are calculated and displayed. To me, at least, it would be preferrable to show the actual ratings, but weigh them in the total score.
That is, I'd prefer to do Score=sum(weight(i)*game(i)) rather than Score=sum(wgame(i)), where wgame(i)=weight(i)*game(i).
Marius, I am not sure I understand what you are saying.
Here is the row (ok, I admit I removed all the undecipherable arrays and replaced them with human-readable variables) that calculates how to display rating for each game (the ratings are saved in the database as I got them, with go being converted the way explained in the about page):
Marius: This is funny, your suggestion is pretty obvious (the rating variations don't affect players' ratings in individual games, whatever way they are counted), yet it never occured to me. I did this to be able to verify that the rating calculations are correct (knowing that I wrote it myself, I was actually pretty surprised to find out they are correct ;-)
And I think I'd like the displayed values to somehow relate to the rating, although I'm not sure whether my way (translating the values so that the cutoff value is 1500 again) is good.
For now, if you want to see the "real" ratings, use the default variation :)
I created a games_with_dice_1000 rating, where only EinStein and StreetSoccer ratings count, with 1000 the cut limit.
The results are ok at the Ratings column but not at the streetsoccer and einstein columns. Ratings appear to be 500 higher, although the calcualtions are right.
ypercube: Actually, they are recalculated so that 1500 remains the cutoff value (I think it makes sense to display it this way, but I might have to rethink it, since Marius already had the same complaint as you (just three posts above :-))
I'm with Marius. It's just plain odd to see your familiar steady-state rating jumping all over the shop depending on what variant you're looking at. However, all in all, it's a great site - thanks!
It seems like the updating does not work. The rating for each game is updated, but the sum is not. I've only checked the weighted variant, though, so I don't know if the others are OK.
It's sad that the excellent feature with variations no longer work properly. I completely understand that tasuki has other and probably more important things to do than fix this, but it's sad nevertheless.
The bug is rather strange, as the ratings for each individual game is updated, yet the total rating is not. And for the "original" variation, everything works just fine. I keep copying the ratings to Excel and calculate the total ratings there, but it's definitely not a desirable solution.
First - I'm really sorry to have left you waiting for so long.
The bug was actually fairly trivial - I didn't check the length of the variation name, and in the table containing variations I only saved first 20 characters. Of course, in the player table, there was a column with the original full length name, hence the query failed completely.
It's all working again (I hope, at least), and I promise to be more responsive and responsible next time :)
Anyway, it's not a hard set limit, it's just the number of rows I decided to initially put into the database (now extended to 20000).
On a completely unrelated note: I decided not to do regular updates of the ratings, as I figured it might be a heavy load on littlegolem. Now I see that googlebot and yahoo slurp have taken care of regular updating while distributing it over longer time periods :-)
Side note: I knew I shouldn't have bothered with the variations - now I had to go through all of them and guess how the creator would have dealt with havannah...
Side note #2: I didn't update the ratings (just few ppl to see whether it works), I'll leave it up to users and googlebot.