Testing limits of control latency perception (16)

1 Name: #!usr/bin/anon 2005-10-21 15:58 ID:s3ZinKUR

I coded a test application to ABX test moving a white square around a black background using the arrow keys at 60Hz synced to vertical refresh rate, with 0 or 1 frames of latency added to the controls.

I tested myself with 15 trials. I found it extremely tiring and difficult, and these 15 trials took over 2 hours. Final result was 12 successes and 3 failures.

Using the binomial theorem we can calculate the probability of achieving this result by luck:
p = (15 nCr 12) * 0.5^12 * 0.5^3
= 0.014

Therefore is it overwhelmingly liking that I was indeed able to tell the difference between 0 and 1 frames of added latency, and everyone who said it shouldn't make a difference is wrong.  Therefore triple buffering is not suitable for fast action games.

I think the oldness/softness of my keyboard made things much harder, I'll rewrite it to use mouse buttons instead and test again. Also release the source so others can test themselves. But not now, because testing is very unpleasant and I don't feel like going through it again just now.

2 Name: #!usr/bin/anon 2005-10-21 21:17 ID:pRr4kx6G

Huh? What's a success? What's a failure?
How about trying it with more than one item on the screen?
You're missing a lot of stuff to try and conclusively say something here.

3 Name: #!usr/bin/anon 2005-10-21 21:32 ID:s3ZinKUR

Success being I matched the X control set to the correct A or B control set. Failure being I matched it to the wrong control set. A and B were randomly swapped each trial (one lagged, one unlagged), X was randomly equal to A or B. No feedback on score was given until after all the trials were complete. Switching between control sets was not done while the square was moving. I can conclusively say that I can identify differences in timing of 17ms using visual feedback.

4 Name: !WAHa.06x36 2005-10-22 02:10 ID:Heaven

Ugh, man, you already had ONE thread, did you really have to make another? Don't chop up your discussion into tiny bits. Please use the other one instead, and let this one die.

5 Name: dmpk2k!hinhT6kz2E 2005-10-22 03:48 ID:Heaven

Since the discussion is now here, might as well stick with it.

It took two hours just for fifteen runs on a simple scene, and has a sample of one? In order to be convincing, there would need to be a number of gamers, playing an actual game. Of course, something like that is actually worthy of a research paper.

However, it's still an interesting result. Obviously, some people can tell a difference, so I think it's worth further investigation.

6 Name: #!usr/bin/anon 2005-10-23 15:55 ID:Heaven

> Since the discussion is now here, might as well stick with it.

Don't encourage lousy behaviour.

7 Name: sage 2005-10-23 18:09 ID:s3ZinKUR

>>6
This thread is about human perception, the other one was about prevalence of various specced monitors.

8 Name: #!usr/bin/anon 2005-10-24 03:51 ID:Heaven

>>7
sage goes in the link field, idiot.

9 Name: #!usr/bin/anon 2005-10-24 18:29 ID:Heaven

>>8
According to the FAQ, it goes in the non-existent email field. There was a 50/50 chance.

10 Name: #!usr/bin/anon 2005-10-27 02:25 ID:Heaven

>>9
Nobody uses the FAQ. Use the Wiki.

11 Name: #!usr/bin/anon 2005-10-27 14:27 ID:s3ZinKUR

>>10 FAQ is linked at the top on the page, Wiki link isn't even visible unless you scroll past all the crap on the navigation frame. It's reasonable to assume that the FAQ has precedence.

Amidst all the hostility, nobody noticed that I fucked up the stats! (see "Prosecutor's Fallacy") But your chance to gloat is over, as I now present the correct two-tailed significance: 0.035. Still <0.05 though.

12 Name: #!usr/bin/anon 2005-10-29 05:48 ID:Heaven

>>11

Thanks to your whining and carrying on, the FAQ has been removed and now points to the Wiki. Lets hope this solves any other problems for future idiots like yourself.

13 Name: #!usr/bin/anon 2005-10-29 19:08 ID:Heaven

Such hostility in this thread!

14 Name: #!usr/bin/anon 2005-10-29 21:15 ID:Heaven

Such is the fate of the Shitty Thread.

15 Name: #!usr/bin/anon 2005-11-15 20:57 ID:pN6WE/Vf

>>14
fuck you. why do you have to ruin threads before they bloom?

16 Name: #!usr/bin/anon 2005-11-16 00:52 ID:Heaven

>>15
Fuck you too. Why do you have to bump shitty weeks-old threads just to be a moron?

This thread has been closed. You cannot post in this thread any longer.