Expert HD Mode (Tetris Friends)

[Paul676]

<:O
: o>

This thread has been redirected to the pre-defined tone: happy

On another point- even if the tf pointer is totally fair, there are other problems why one could fault it- for example for uneven targeting times, even if they are random, are an annoying pest for pro players

[Noogy]

You're one to talk. I've been seeing a lot of inflammatory, self-righteous posts coming from you. You might want to tone it down a bit.

'tis what i do

[Anonymous]

1. conduct that 1 sample t test over 100 games

er, not trying to butt in, but t-tests are supposed to be used for small samples c;

On another point- even if the tf pointer is totally fair, there are other problems why one could fault it- for example for uneven targeting times, even if they are random, are an annoying pest for pro players

Also, I thought one of the creators or someone said that they wanted to put in targetting to add more "strategy" or something, so I don't think they will get rid of targetting.

[Noogy]

you're right, my mistake

[Paul676]

if they make the targetting fair and equal times but still garbage to 1 then it might be more fair still and less easily criticised/debated about if it's unfair

[chopin]

I don't think targetting or anything else that is seen as "random" is unfair. It provides a changing terrain that weeds out better players. That's why you always see better players winning regardless - because they can adapt. There's no point of having a "multiplayer" if you just play with yourself. This interactive randomness is essential to a good multiplayer. A pro will get the job done.

[XaeL]

er, not trying to butt in, but t-tests are supposed to be used for small samples c;

7 is far too small.
maybe 50?

[Anonymous]

7 is far too small.
maybe 50?

http://www.statisticssolutions.com/methods...-sample-t-test/

"In one sample t-test, sample size should be less than 30."

Although it really doesn't matter, because his sample doesn't meet any of the other assumptions/conditions. ;p

[Paul676]

I always thought a more accurate answer would result from more repeats of experimentation. So the more the merrier!

[Anonymous]

I always thought a more accurate answer would result from more repeats of experimentation. So the more the merrier!

Normally, you do want to have a big sample size; but I was just saying that you use a t-test for small sample sizes.

[XaeL]

doesnt it have to meet some certain requirements
e.g.
1) random
2) representative
3) etc.

[Kitaru]

if they make the targetting fair and equal times but still garbage to 1 then it might be more fair still and less easily criticised/debated about if it's unfair

It's a bad system regardless. Again, as it stands, it doesn't give the player enough control. Give the player too much control and it'll lead to degenerate strategies like sniping individual players out of a room. It's a convoluted solution to a simple problem, as always.

Until they come up with something that is fleshed out more, they should probably go with something simple like even division of garbage. Even division of garbage is nice because it approximates 1v1. Until they have a way to make FFA an interesting experience in its own right, this is probably a good approach.

I don't think targetting or anything else that is seen as "random" is unfair. It provides a changing terrain that weeds out better players. That's why you always see better players winning regardless - because they can adapt. There's no point of having a "multiplayer" if you just play with yourself. This interactive randomness is essential to a good multiplayer. A pro will get the job done.

It may generally balance itself in the long term, yes, but it leads to discrepancies in the short term. In FFA, a much lesser skilled player such as myself can get lucky and beat Blink without any real mistake on his part -- just opportunism based on where the garbage falls. There really doesn't need to be any potential for this situation to ever arise.

It seems like a lot of people are trying to defend an "improved" targeting system mostly as a way of upholding the status quo. Can someone instead please explain to me what merit the system has that would warrant its continued use over a more elegant solution?

[Impact009]

I wasn't going to use my tiny, non-randomized sample of course.  Independent, random samples would easily be possible, but the fact that the auto-handicap (assuming it's even there) having its own variables affecting it would cause it to be a confounding variable itself.  I had incorrectly assumed that I could easily find the population mean, but that's another hurdle itself.

Perhaps control it so that one player always wins, and go for a Z-Test with a larger sample size.  The randomization would be screwed then by trying to control the auto-handicap, and there's still the problem of needing to know the population mean and having normal data.

Player A always wins, and the dependent variable can be how many items he or she obtains per game, or how many Tetriminos need to be locked before he or she obtains an item.  Player B will have the same dependent variable, with the control to always be losing.  Maybe a two-sample test would be better.  Now to find the population mean and figure out randomization...

Well, now I'm actually interested in what a good procedure would be.  Does anybody have any ideas?  I'm looking at item frequency being a dependent variable, but perhaps I'm looking at it the wrong way.

...Or I could probably just go and ask a dev LOL.  Not that anybody on this forum would probably care, but now I'm hoping to find somebody else who would find a project like this to be entertaining.

As for targetting, I also agree that equal distribution of garbage would be the best idea in FFA.  Give players any amount of control, and it creates the possibility for a group of people to snipe somebody.  I know this is going to be Expert HD and and the top players probably wouldn't snipe anyway, but I'm in favor of seeing fair gameplay for as many players as possible.  Yes, a much better player would be able to overcome things like that, but then it creates a gap for those players who are better, but not that much better.

I don't see sniping being much of a problem in legitimate teams matches though.

[Anonymous]

doesnt it have to meet some certain requirements
e.g.
1) random
2) representative
3) etc.

yep, it didn't meet all the conditions and assumptions ;(

...Or I could probably just go and ask a dev LOL.  Not that anybody on this forum would probably care, but now I'm hoping to find somebody else who would find a project like this to be entertaining.

if you wanted to do any statistics, it would probably be more worthwhile to do something about the targetting system (and maybe introduce blocking between different ranks and stuff). C:

[chopin]

Kitaru, can you? lol. When I watch Blink play TFA he wins all the games unless he messes up on his own. And about dividing garbage, it may create an entirely different game. Would  matches be slower?