Thread in 'Discussion' started by Kitaru, 2 Jan 2012.
My point is : skills of slower but consistent player >> skills of faster but irregular player.
Billtsar: If I get Death M maybe one in every five games, what is happening the other four games? Are you seriously trying to argue that there is no variation in the performance people do when they play?
I've gotten Death Gm precisely once, ever. I'm nowhere near the level of someone who can get Death GM in four out of seven games.
We all have good games and bad games. Some are down to lapses in concentration, or maybe a poor piece sequence with no I pieces for ages, or maybe we stacked badly but the randomiser was kind and we got away with it. There are loads of reasons why we don't get anywhere near the exact same score every game. Sometimes those reasons will all happen to go the wrong way and I screw up and die at level 70. Sometimes they go the right way and I get Gm. The fact that they don't go all the right way every game is exactly why I don't get Gm every game for my current level of skill, even though I'm capable of it.
Your skill affects where your average games lie, and whether a "good game" is 300 or 500 or Gm, but when you are getting different scores every game due to statistical variations, then your PB is always going to be the game where all those factors combined together in the nicest way for your given amount of skill.
I think he's trying to argue that it doesn't make sense to draw an arbitrary trade-off between consistency and speed or play quality, because they generally increase in tandem (so a person who is extremely consistent is also likely to be faster, etc). He's obviously not saying there's no variation because he said outright that getting Gm 10 times in a row is basically not human. So, his point isn't (I think) that your average game is or should be your record, but that a player's PB and average game are in a fairly linear relation, such that if you ordered all players by PB and then again by average game, the two lists would come out fairly similar (similar enough that a higher PB implies a better average game, and therefore skill level).
Thanks zircean, that's what I'm trying to say. As for the 10 GM in a row paradigm I said it's feasible but only for a sub11 player not for someone who completes it in 13min.
And of course I don't disagree with your last post rosti.
"If I get Death M maybe one in every five games, what is happening the other four games?" 1/5 means you are super beast and the other 4 games have very high average.
"I've gotten Death Gm precisely once, ever. I'm nowhere near the level of someone who can get Death GM in four out of seven games." In this case, if you watch the video of each player it's easy to recognize who is the best judging from their gameplay. And of course you can't be faster than that guy so time factor shows the difference here as I mentioned in a previous post.
But what's to stop me just getting a slightly more favourable game without actually increasing my skill at all? My Death GM is a good example, because I lose a ton of time to level-stops and would have been a lot faster without that (I worked out from the video that I lost around 12 seconds to level stops just from 0-500). It would have been perfectly possible for me to play the exact same game but just happen to clear lines at X99 instead of X97 and I'd save all the time I lost. If I'd gotten a bit luckier with the stops I could have been under six minutes, and that'd put me ahead of Kitaru, c_t and Amnesia on the leaderboards.
This is the point I'm making - that because they are a record of a single game, personal bests are affected by game-to-game fluctuations (they are a direct result of them) as well as just skill. This becomes especially relevant at higher levels where the margins for error are so damn tiny.
And like I said, I'm not arguing that personal bests are a stupid metric - I just think that some measure of average skill is better because it overcomes all the game-to-game fluctuations. Maybe it does give pretty much the exact same rankings for player order, but that doesn't stop it being fundamentally a better way to measure.
If that "favourable" game occured and you had cut 15sec, in your one and only GM don't you think that the same would happen to a better player in one of his 50 GMs? Come on rosti. The only case where what you say applies is in a small sample. 1 GM 6:10 for you, 4 GM 6:15-6:30 for the other player.
I am technically a TGM1 Gm. A moderately fast one, too, with sub-12. Could I play in a "The Masters" tournament? No. I have not yet gotten a second Gm, and I most certainly can't shake one out of my sleeve at a The Masters event. Nuff said.
I honestly don't have a clue what you're even trying to say with this post. And we're talking about small samples here - how many players can actually get Death GM on a semi regular basis?
My point was that my single Gm could have potentially been a run lucky enough to put me above players who are almost certainly better than I am, and who shouldn't be behind me in the rankings. Because we look at personal bests. Because we look at single games. Because we look at small samples.
A very skilled player could play conservatively and in a way that, because of their skill, lets them reliably get GM. On the other hand someone could play in a way where they purely go for speed, meaning that in that one-in-a-thousand game where they finally get GM, they have an epic time because they're going balls-out. And they'd be higher up on the rankings even though their average (and their skill) is substantially lower than GM, just because they're playing such a high-variance way.
I feel like you don't truly appreciate just how rare and valuable these sorts of grades are when you're just on the level of getting them. Things like Death Gm, or Master Gm are games where the margin for error is so small because at full tilt one misdrop can completely throw you off and kill you. For master it's even more skewed because a misdrop at level 50 is completely different to a misdrop during the invisible credit roll.
When I get to 500 in Death, all I want for the next 500 levels is to not make a mistake, because making a mistake will usually kill me, or put me in a spot where I end up topping out eventually anyway. How likely I am to make a mistake is down to my skill, but that doesn't change the fact that if I play enough games I'll eventually get a lucky run where I don't make a mistake.
I really couldn't be arsed doing this earlier, but I'll do some proper maths to prove my point.
Say the time from 500-999 is three minutes. If I make a mistake on average once ever three minutes and use a Poisson distribution, the chance I'll live through those three minutes is 36.8%, one in every 2.7 games.
If I make a mistake every minute and a half, it becomes 13.5%, one in every 7.4 games.
If I make a mistake every minute, it's 5%, one in every 20 games
If the crappy player who makes mistakes every minute just plays more games, he can still have as many Gm grades as the good player. If we assume that all other skill factors besides how many mistakes they make are comparable, then they'll have roughly the same record. This assumption is probably wrong, but you can argue that a player who makes fewer mistakes is likely to be more skilled and therefore faster, and you can also argue that a player who makes more mistakes is likely to do so because he plays faster, so I feel the assumption is OK for what I'm trying to say.
The Ti grading system doesn't care how many times you get the rank to give you the qualification - it looks only at the rate, and in this case the player who can get Gm one in every 2.7 games has a far better chance of qualified than the person who does it every 20 (though with the figures, neither of them would be likely to get it).
This means that simply playing more games isn't enough to get the best grades, and you can't just rely on your 1-in-100 lucky games to push you up the rankings when you play more than other people. It's a measure of skill rather than skill combined with games played.
And with this stupidly long post, if you don't understand my point of view now then I honestly don't know how else I can better explain it.
Can you give a definition of "mistake"?
You still use the word "lucky" so there is no need to discuss it further.
Maybe I'm crazy, who knows? I give up.
A mistake could be loads of things. Maybe I double-lock, maybe I press the wrong button, maybe I make a really bad stacking decision, maybe I lose the DAS charge, maybe my mind just goes blank for a second and throws me off. It could be pretty much anything which is likely to kill me and is directly my fault.
I use "luck" because I'm talking about statistical fluctuations and events which are going to happen at irregular intervals. You might technically have control over the events, but you don't really, because it's not like you ever make a mistake on purpose. You can only really control how often they happen - not when they happen or how long the gap between any two will be. Sometimes that gap will be long, sometimes it won't be, and whether it is, is far more a factor of statistical probability and luck than anything else.
TGM is luck based simply because it features a randomizer.
Thank you for the very entertaining discussion.
"TGM is luck based simply because it features a randomizer."
Assuming you agree that TGM features a human player as well ... I understand you both agree with Rosti LFC, who is basically saying, that in the long run skill will make the difference and in the short run it is luck combined with skill...when looking at just the next game it might just be luck...
So everybody is basically right, depending whether you think a single game or a lot of games should be considered most important.
This brings us back at the beginning of the discussion.
So let's start the second round by me saying:
I think a player with a better average is better, but I can't be bothered keeping a statistic of my results, so I prefer judging skill by PB's.
I'd agree with this, and keeping those stats of results is basically what the Ti qualification system does, which is why I like it.
So we are well in the dictatorship of the regularity/consistency.
A player is good if he has a good average and THAT'S IT.
But if someone try to push it for converting some prudence into more top speed, he is less good.
So DEATH mode, jago is definitely better than me because he has 15 Gm with best at 6:25, than me who has 4 GM for a best at 6:05.
And what if jago today is not able to get 6:05 even in trying?
It is impossible to check anyway.
I think too much people subjectively give too much supremacy into the consistence.
Hell, the very notion of "skill" is subjective.
So what I think is we should stop take all of this so seriously and just play because we want to..
For me, fun comes first, then competition.
m.kevin thums up.
So back to the title of the thread:
My tetris resolution for 2012 is to become a TGM grand master.
I would say he's still better than you
But I guess my opinion is screwed by playing too much with players who can sit down in front of the game and do TGM sub 9:30 or Death M or Shirase 1000 each time. You realize then your single PB gotten in Mame doesn't mean ANYTHING, really...
That's why my 2012 resolution is to become consistent first. Time matter will come later naturally.
I think when you actually have to put money into a cab it makes you concentrate a whole lot more, or at least that has been my experience with playing shumps at the arcade in London.
On MAME I can sit back and play for an hour without ever really concentrating, whilst when I've spent money and have limited plays I try my best to make each one last as long as possible.
Thanks so much man, and wow, so sorry about your issues. Really hope it gets solved so you dont have to walk around in that kind of "haze" or whatever.
Cheering for you!
Also, not that I have much to say about what neither Rosti or billtsar have said already, I agree with basically all of it, and yeah the most fair thing would surely be some kind of scientific average, so this is just for some kind of personal feeling of "skill level":
When you are in kind of a 'zone' you can have a whole evening, even a couple of days of veeery high averages in my experience. Why is this so? Hard to analyze well for me, but some kind of relaxed, extremely smooth and focused mind. To some extent this is of course influenced by playing very intensely maybe a week leading up to this small period, but not completely. I have had good 'zone' plays both after playing extremely bad, pretty good, after playing very much, very little or nothing at all for long periods.
There will still be fluctuations from game to game, but way less. I have some memory of having several games in the 930-999 (Death that is) in a row a number of times in an evening or over a couple of days back when I was still playing well.
Maybe this is just a different way of expressing something billtsar wrote earlier about there being no disturbances like a dog biting your leg, an itch and so on, but for me all the distractions, disturbances and such are done by my own mind. I prefer to look at my skill this way, feels more motivating. Sure doesn't mean I could walk into some arcade and expect to find myself in this so called zone though of course.
Maybe I have misunderstood some of what has previously been written, if so, sorry about that. Tired.
Cheers to everyone, my wishes for happiness!
Separate names with a comma.