Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Insomniac]

Looking at council room he does have some real games

[Ozle]

Ohh, he is my new hero. I may offer to have his babies!

[Razzishi]

He could easily have other accounts that he has dump games to him; all it takes is a new email address to get someone with a starting skill.

Note that his variance is extremely high, as is fitting for someone who has only played against a few players and presumably hasn't lost.  The fatal flaw in the TrueSkill algorithm is the use of only two parameters to quantify the known information about the skill of a player so that nothing is known about the skewness (or the kurtosis, or whatever else) of the distribution of possible skill values given the data.  The best part of TrueSkill is that it's able to find good matches for people so that nearly everyone has a 50% win ratio; when people are able to pick weak opponents (instead of being forced to play those of similar rank) its utility declines.

It seems to me that if this sort of manipulation is possible, it's quite likely that if you win significantly more than 50% of your games, TrueSkill will overrate you.  Not knowing exactly what the formula looks like I couldn't say for sure, but recall that people had "problems" before with someone else who got very high on the ratings board playing only weak competition and declining games from high ranked players.

[olneyce]

Speaking of which, kyrmanakos.  This person is 38-2 against 'kyrman' and otherwise has basically a .500 record.  You'll notice that 'kyrman' manages to still be a level 22 despite that astonishing 2-38 record against kyrmanakos by playing reasonably well against everyone else.

Seems like a new approach - managing your sock puppet up to a decent level so your slew of victories against them count for a bunch more.

The result is the highest skill level on the board, but 13.8 variance, producing a Level 42.  So it's not even really that successful of a troll.

[shark_bait]

Speaking of which, kyrmanakos.  This person is 38-2 against 'kyrman' and otherwise has basically a .500 record.  You'll notice that 'kyrman' manages to still be a level 22 despite that astonishing 2-38 record against kyrmanakos by playing reasonably well against everyone else.

Seems like a new approach - managing your sock puppet up to a decent level so your slew of victories against them count for a bunch more.

The result is the highest skill level on the board, but 13.8 variance, producing a Level 42.  So it's not even really that successful of a troll.

I didn't even realize that.  I've been looking the board a lot lately because my alter ego Luckyfin has been creeping up to the top in terms of skill and I've have been rather disappointed that I was second these past few days.  I guess with kyrmanakos being a semi-troll, that makes me the highest legit skill on the board currently. 

[olneyce]

Speaking of which, kyrmanakos.  This person is 38-2 against 'kyrman' and otherwise has basically a .500 record.  You'll notice that 'kyrman' manages to still be a level 22 despite that astonishing 2-38 record against kyrmanakos by playing reasonably well against everyone else.

Seems like a new approach - managing your sock puppet up to a decent level so your slew of victories against them count for a bunch more.

The result is the highest skill level on the board, but 13.8 variance, producing a Level 42.  So it's not even really that successful of a troll.

I didn't even realize that.  I've been looking the board a lot lately because my alter ego Luckyfin has been creeping up to the top in terms of skill and I've have been rather disappointed that I was second these past few days.  I guess with kyrmanakos being a semi-troll, that makes me the highest legit skill on the board currently. 

Hilariously, this is not true because MY alter ego (The 9th Doctor) is actually slightly ahead of Luckyfin.  In fact, that's the reason I noticed this at all.

[shark_bait]

Speaking of which, kyrmanakos.  This person is 38-2 against 'kyrman' and otherwise has basically a .500 record.  You'll notice that 'kyrman' manages to still be a level 22 despite that astonishing 2-38 record against kyrmanakos by playing reasonably well against everyone else.

Seems like a new approach - managing your sock puppet up to a decent level so your slew of victories against them count for a bunch more.

The result is the highest skill level on the board, but 13.8 variance, producing a Level 42.  So it's not even really that successful of a troll.

I didn't even realize that.  I've been looking the board a lot lately because my alter ego Luckyfin has been creeping up to the top in terms of skill and I've have been rather disappointed that I was second these past few days.  I guess with kyrmanakos being a semi-troll, that makes me the highest legit skill on the board currently. 

Hilariously, this is not true because MY alter ego (The 9th Doctor) is actually slightly ahead of Luckyfin.  In fact, that's the reason I noticed this at all.

Hmm... I hadn't noticed you creeping up there too, but I did accomplish it a couple days ago.  And not having lost with Luckyfin since then I guess I just assumed no one had passed me.  Look's like I'll just have to find more time to play dominion 

[olneyce]

Speaking of which, kyrmanakos.  This person is 38-2 against 'kyrman' and otherwise has basically a .500 record.  You'll notice that 'kyrman' manages to still be a level 22 despite that astonishing 2-38 record against kyrmanakos by playing reasonably well against everyone else.

Seems like a new approach - managing your sock puppet up to a decent level so your slew of victories against them count for a bunch more.

The result is the highest skill level on the board, but 13.8 variance, producing a Level 42.  So it's not even really that successful of a troll.

I didn't even realize that.  I've been looking the board a lot lately because my alter ego Luckyfin has been creeping up to the top in terms of skill and I've have been rather disappointed that I was second these past few days.  I guess with kyrmanakos being a semi-troll, that makes me the highest legit skill on the board currently. 

Hilariously, this is not true because MY alter ego (The 9th Doctor) is actually slightly ahead of Luckyfin.  In fact, that's the reason I noticed this at all.

Hmm... I hadn't noticed you creeping up there too, but I did accomplish it a couple days ago.  And not having lost with Luckyfin since then I guess I just assumed no one had passed me.  Look's like I'll just have to find more time to play dominion 

Success!  The 9th Doctor is #1 at the skill element with 56.429.

All this really proves is that good players can bump up their skill by creating new accounts and erasing all their fumbling around from when they were new at the game.

[^_^_^_^]

I bet when everyone moves over to the new application being made by RRG the ranking will be much different but people that currently struggle to get above, say, level 36(you know who you are ), will probably have a much higher chance than before at rising to the top 5-10 people.

[ftl]

Why do you think so? I assume the ranking system used will be different so a direct comparison of numerical levels will make no sense, but I would expect relative rank to stay mostly unchanged - the same group of people at the top will stay at the top, and so on. Or do you think that the trueskill ranking works badly enough that a reset makes a big difference? Despite olneyce's comment about The 9th Doctor having a really high skill, his original account (olneyce) is level 43 and the new one is 45, not actually that big of a difference...

[WanderingWinder]

Why do you think so? I assume the ranking system used will be different so a direct comparison of numerical levels will make no sense, but I would expect relative rank to stay mostly unchanged - the same group of people at the top will stay at the top, and so on. Or do you think that the trueskill ranking works badly enough that a reset makes a big difference? Despite olneyce's comment about The 9th Doctor having a really high skill, his original account (olneyce) is level 43 and the new one is 45, not actually that big of a difference...

Yeah, but skill is your real rating. Level is... a novelty, basically.

[ftl]

Heh. I haven't been thinking of it like that - the level seemed to me to be a more reasonable estimate of how good you are, precisely because a rating system shouldn't be rewarding a player for its own uncertainty. Like, a new player who hasn't played any games at all starts out at a skill of 25, but that's not really a valid estimate, their level (0) is a much better indicator. I'd thought the level system works pretty well - at least, on this site I haven't heard grumblings about how people are under-ranked and should be much higher level, or how people are over-ranked and should have a much lower level (except for players gaming the system, which is a separate issue). Have I just not been paying attention?

[Young Nick]

The only reason that people in mid 30's or so would receive a slight bump is because there would now just be more players. Theoretically, with more players, there will be less experienced online players, so for at least a little bit, those of us with Isotropic experience will perform relatively better against new comp than we would otherwise.

[jonts26]

Heh. I haven't been thinking of it like that - the level seemed to me to be a more reasonable estimate of how good you are, precisely because a rating system shouldn't be rewarding a player for its own uncertainty. Like, a new player who hasn't played any games at all starts out at a skill of 25, but that's not really a valid estimate, their level (0) is a much better indicator. I'd thought the level system works pretty well - at least, on this site I haven't heard grumblings about how people are under-ranked and should be much higher level, or how people are over-ranked and should have a much lower level (except for players gaming the system, which is a separate issue). Have I just not been paying attention?

I would think that level is a better predictor for new players but once the player has a certain number of games played, mean skill would be better.

[Deadlock39]

I have no complaints, but since I have been playing less lately I have noticed that my level has sunk a few points from low 30s to high 20s.  I suspect this is because my variance has increased, but I highly doubt my actual skill has decreased in any significant way due to playing less games per week.

[lespeutere]

Like, a new player who hasn't played any games at all starts out at a skill of 25, but that's not really a valid estimate, their level (0) is a much better indicator.

Is it? Not everybody starts playing dominion on isotropic. I think level 0 suits our intuition but that's it. I'm neither speaking pro nor contra looking at levels or skills, I'm just questioning your statement. ;-)

[Davio]

I have no complaints, but since I have been playing less lately I have noticed that my level has sunk a few points from low 30s to high 20s.  I suspect this is because my variance has increased, but I highly doubt my actual skill has decreased in any significant way due to playing less games per week.

But other players may have gotten better?

Because this still is an evolving game I like the sliding window as it shows who masters the current theories and accepted strategies best. If we would have a non-moving leaderboard, the #1 could have only played 1000 games when Prosperity was the newest edition and he would sit there forever. With new expansions out, can we still consider him the best Dominion player ever? Of course not.

I like having a moving leaderboard to make players prove over and over that they still have got what it takes. There's still a lot of research to be done. Maybe the moving board could come to a halt one year after the last expansion once the dust has settled and the main strategies are well known.

[Deadlock39]

Of course, my level definitely could have gotten lower for many legitimate reasons, and I also don't care about it, which is why I said I have no complaints.  I was just stating that my level decreased, and I don't feel that my level of play has gone down.  I am still winning at about the rate I expect, and haven't changed behavior in what opponents that I play.  What I have done is started having periods of time where I don't play any games, and every time I come back from one of those, my level has dropped 1 or 2 ranks.  I have frequently gained one back if I play a lot of games in the day.  I haven't been explicitly looking at the leaderboard and my variance, but I am pretty confident my level would level back out at 30-32 where it was before if I started playing about every other day like I was before.

I think it is a good thing for a leaderboard, but I don't think the rate of level decay in the current system is an accurate gauge of actual skill loss from lack of play.