Nay?

And in case you don't want to feed the resident +1 whore:

Yay?

yay, but w/e

Nay?

Fuck nay. I got into the top 100 in GR playing less than 10 games. It was easy.

That aside, isn't the ability to raise your rating independent of the level of competition? As a raw value that is, not as a rank. If your competition is twice as stiff as it used to be, they'll have twice the ghost rating, and you'll beat them half as often for double the prize.

Net zero gain. Right?

Seems likely to me that it's a bit better. I haven't been around forever, but the progressive nature of the site (new openings, new strategies, new people) always make it a possibility. There have been so many developments in the people here that it only makes sense it would be so.

LOL, ghug. Not everything I post is meant to be a +1, and 2WL beats me by a mile here.

Sometimes I have a thought and I would merely want to express it.
And sometimes I am genuinely curious and until polling is an easy thing to do on the site, I thought keeping a yay/nay count would be easiest this way.

Players now have better workout routines, better nutrition programs, and top notch trainers. Old timers like MadMarx must be on the juice in order to keep up.

Heh, I have to give you shit for it anyway. I'll also give you a +1 because you put me in a list with a bunch of people who are better than me (and some loser gunboaters, but whatever).

I feel like there's definitely a greater wealth of good players than there were even when I joined, but I think that that's probably just as much a function of a growing user base than anything else. The number of good players also doesn't make it any harder to game GR by playing bad players.

I agree with ghug, he isn't very good at all ;)

And no, you are incorrect, YJ on two different counts.

First, yes you got into top 100, but you've played only one top GR game AND you got beat squarely if I recall correctly.
Second, your statement is only true if there is a system that can accurately measure someone's skills quickly, efficiently and without fail. The GR system that we have is good but it still measures only the approximated "true" rating, not the actual "true" rating. So, no, if your level of competition is twice as stiff, then their true ranking would be twice as high (relatively speaking) if and only if you can measure the true ranking. As it is, GR measures relative strength or how you perform in relation to the people in your game, and so if the competition is higher, then it would take more games or more wins to get to the same ranking than when you played with easier competition.

This is easy to test out for you: why don't you try playing a few more games with the current top echelon and see if you can keep your 40% win rate? :)

Fight me YJ. Do you even lift?

#fuckyj #stophiding #pussy

"...make it any harder to game GR by playing bad players."

I've often wondered how true/possible this is. You could easily make the argument that that's how I got so high so fast. I think though, that it is self-controlling to a large degree, because even MadMarx will lose to 6 noobs one time in 40, and the GR hit from that loss will be enormous. Eventually, playing talent at any caliber is a zero sum game with GR. That's the beauty of the system - you will hover around your natural level.

Still, one has to wonder if your "natural level" as determined by playing only rookies would be higher than as determined by playing persons of equal skill. I'm tempted to say not, because your ghost rating as a raw value only has meaning when compared to the baseline of 100, - i.e. ghug at 300 is 3x more likely to win a game than somebody who has never played before, MadMarx at 900 is 9x more likely, and MadMarx is 3x more likely to win than ghug. All of these statements are accurate, and the relative skill of players is captured regardless of where they stand on the scale. I don't think the system can be gamed like you suggest.

ok, ignore what I just said while I ponder what Zultar said...

It's hard to tell. Some of the best players I've ever known (Ivo, Babak, Treefarn, PolarBear) haven't been active in some time. With that being said, I'm sure there are plenty of new players that are very good as well.

hmm, beat squarely, but not soundly. All I can say is that I belonged in that game, and I think that the projected 8% chance of "winning" (ignoring draws) I had based on GR was fairly accurate.

But I thought about your second paragraph and I was still missing something. To my understanding, any flaw in the GR system is because of "transient" effects as it normalizes to your true skill over time. But after stepping through the below exercise I think you're right.


Let me try again, and if I still don't get it, tell me where I'm confused (again, ignore draws as they complicate the issue):

Scenario 1:

I am a brand new player and playing 1,000,000 games with 6 other brand new players of equal skill. All of us never get any better, so each will win 1/7 of the games. All of us will have, after 1,000,000 games played, a GR of 100, just as we started with

Scenario 2:

I am a brand new player and playing 1,000,000 games with 6 other brand new players who are in truth double my skill. None of us gets any better, so the other six players each win 2/13 games, and I win 1/13. After 1,000,000 games played, their ghost ratings will settle at some point double mine, and mine will settle at some point lower than 100. Therefore, my raw score depends on the caliber of opposition.

OK, I'm sold, you're right. But am I right in the manner with which you are right?

I think this is probably one of those things where you will always think that the first "elite" players you are exposed to will be the ones you consider the best players since they were the ones you learned the most from. For me, it is players like Babak, davebishop, MadMarx, rhino86, djbent, and yeah, I'll say it, AgentK

OK then moving on, even if that is true, doesn't maintaining one's high GR become self-regulatory as time passes? Since the true skill level of the opposition doesn't influence how much you lose in GR when you are defeated, if the level of competition becomes stiffer, won't the best players be harder pressed to maintain their current raw values?

I know you're a math guy, and this is probably a lot more intuitive for you than me, so thanks for going slow with dumb old YJ.

So, I think that would be an appropriate question: won't the best players be harder pressed to maintain their current raw values/rating (not ranking)?

I would argue yes. I'm going to pick out a few examples: MM, MarekP, and PE. I talk to MM and PE regularly, so I hope that won't get too upset with me here and they both know how I respect them. And I apologize to MarekP before hand since I don't know him at all.

It is harder for MM and PE and MarekP to maintain their raw values/ratings. When they have played more than a few games recently, they have dropped their raw rating. Just compare their current raw vs all time raw. My argument and evidence would be that it is indeed harder to maintain their top raw scores when they play with the top echelon now. My instinct/intuition tells me that there would be a general but noticeable, maybe even significant, decline in the raw rating for the old echelon if they competed with the new crop.

I think the number of good players goes up naturally as site membership expands. The drawback to this is the corresponding increase in bad players who quit, CD, and generally decrease the overall quality of play, making it more difficult for those good players to get a quality game unless they set it up beforehand. I see this phenomenon with races quite often. As the field expands, the winning time may become faster, but the median time increases with more new runners.

OK Zultar, we'll just have to see if that happens, then. I do follow your argument, but I really have no idea whether there overall caliber of players is now higher than it once was.

GR primarily measures time, not skill. Therefore, it's nearly equally easy to raise your GR now and earlier, I would think.

Madmarx, babak, ivo ivano, djbent, rdrivera2005, fortknox, rlumley, Crazyter, Spell of Wheels, TheWizard, Plastic Hussar, Tru ninja, centurian, themastergamer, chrispminus, President Eden, Crazyter.

Those are the players I played against in my first year here who were some of the best on the site. A lot of them don't play as much, or at all anymore. In all of my games I have only been completely destroyed and owned by two players. Babak in one game, and djbent in another. I've played against most of the current top players and never got thrashed like I did against djbent. Now on the flip side, I've gotten much better since then, and grabbed a 3 way draw in a game both MM and Babak were eliminated in, so I'm not sure my perspective is clear.

I do agree with Zultar in one respect though, many of these players ghost ranking, (ex. PE's all time is 530 or so), is inflated higher then it would be against the competition today. The site is much more mature in stopping solo's. Another large factor is a few years ago not many people understood how to manipulate the GhostRating system as today, so the people that did were able to increase their rating quicker and higher then they would be able too today.

The more recent shift from points fascination over to ghostrating system made more people aware of how it works, and hence people are more cautious making it hard to increase your ranking quickly. And as Zultar said, there are a lot of newer players who are really skilled, everyone he listed, goldfinger0303, Tom bombadil, Zultar, 2ndwhiteline and some others are all excellent players.

I'm not sure the competition is "better" skillwise, but overall, yes the competition makes quick ranking rises harder.

And not a single +1 was given to zultar that day.

I would say nay. Good players leave just as often as good new players show up. There are tons of quality players who used to be active, but who aren't anymore. Dangermouse, babak, Ivo (okay, I managed to draw him back on), TheMasterGamer, djbent, etc. But, as you've said plenty of good new players have shown up to fill their spots, so I'd say net neutral overall.

Also, YJ you would've done well if you didn't fuck me over that game lol

CS that's really not true at all. Go through the arguments that zultar and I already layed out.

The one thing that GR is perfect at is measuring the skill of a repeatable system given infinite time. If you pull any 7 players, and "lock in" their true current skill level (i.e. assume they don't get any better), and have them play 1,000,000 games together, their ghost rating will be reflect their relative skills approaching perfection. Even if their true skill does fluctuate over time, their relative ratings will capture it.

Basically, the only thing you can say about GR being inaccurate in regards to time is that you need X number of games before it starts to paint a truly reliable picture. But that's just sampling theory - and true of any stat system.

I fucked up big time in lots of ways that game, goldfinger. I actually don't remember why it was I did what I did. I do know it was obviously not the right choice. Will have to go reread that press.

@YJ: I've worked through the formula. GR does not attempt to measure relative skill. It attempts to measure a combination of relative skill and longevity. The formula is designed to converge to the right value strictly from below and only after a much larger number of games than played by the typical player.

It would be easy to design a rating system that attempts to measure skill rather than a combination of skill and longevity, but that's not the one that's been opted for.

I've been on the site for nearly two years, and I only recognize a few of the names y'all are listing. I guess that makes me a noob. I have been in a draw twice with MichiganMan.

Thanks, YJ. And as ghug pointed out, MAN UP and play some, you puss. :)

To continue the argument based on 2WL's observation, I think the upper echelon's competition is higher but the bottom echelon's competition is lower partly due to the increase in the number of players (good and bad) and the level of knowledge of the top players (better job at stopping the solo due to stalemate lines and coalitions and the knowledge/care of GR as jmo pointed out). I don't know about the average or median player though and I don't think I've made any claim regarding them.

In regards to CSteinhard's comment, GR measures approximated true skills with respect to the level of competition you've played as YJ pointed out. I think what you are thinking about is partly true if we do not take into account "all time rating," but that's part of the debate: if the old top timers who climbed so high in raw score would be able to do it now or as easily.

In regards to goldie's comment, while there has been good players who left and good players who joined, I contend that the top level of competition is better now than it was before. One thing we can do is to go through all time gr and current gr to find multiple examples like the ones I have given.

Would the top 50 or top 100 all time GR have earned as much rating now as they did then?