Page 6 of 10

Re: Interesting project about mouse/ gamepad latency

Posted: 09 Jul 2019, 21:54
by ad8e
You can count Linus's clock, it goes to 59 and then 00.

My vague recollection of my proofs from long ago is that under some conditions, pure left/right dodging is optimal. Various physics models here are AD instant constant velocity (Overwatch uses this) and AD instant constant acceleration with exponential damping (I think Quake uses this). So "standing still" would be simulated by rapidly alternating left and right, if the acceleration/position constants in the game make that a valid option under the optimal strategy.

This neglects further physics models such as if jumping evacuates the player's position in less than the aimer's reaction time, or Overwatch's crouching.

The dodger does have some memory, of his own position up to 200 ms ago (if the aimer has 200 ms reaction time). This is hidden state that he can use to make decisions with. So his dodging is not necessarily exponential until after he has moved for 200 ms in the same direction.

I understood your reasoning behind not using position waveforms, and it's true that a direct cross correlation would not work with position, but you can finagle it to make it work with some stupid cutoff tricks. With acceleration, it's much harder to make it work, and your method doesn't suffice. On the other hand, it'll be very hard to work with in either case; you will only be able to get a very loose upper bound on reaction time, since kukkii's aiming is not optimal and won't fit to a model. His mouse tracking is imperfect, like everyone else's in the world. And when the model doesn't fit properly, the results become more conservative.

For those who find the statistics hard, here's an easy visual way to see why:
When Marwan moves in a straight line for a while, kukkii wobbles his mouse like crazy but no correlation is being done.
When Marwan starts alternating his position, kukkii keeps wobbling his mouse, just like before, and now we get some really spurious correlations with Marwan's alternating position.
So when analyzed by the method in the article, kukkii's crazy wobbling gives him a big reaction time advantage when Marwan alternates, and doesn't hurt him when Marwan keeps moving straight. You can think of it like constantly guessing on a reaction time test, where the test doesn't penalize for wrong guesses.

Re: Interesting project about mouse/ gamepad latency

Posted: 09 Jul 2019, 23:11
by spacediver
ad8e wrote:You can count Linus's clock, it goes to 59 and then 00.
Oh, that's a good catch, bravo, I hadn't paid much attention to the real time counter in bottom right before.
ad8e wrote: I understood your reasoning behind not using position waveforms, and it's true that a direct cross correlation would not work with position, but you can finagle it to make it work with some stupid cutoff tricks. With acceleration, it's much harder to make it work, and your method doesn't suffice. On the other hand, it'll be very hard to work with in either case; you will only be able to get a very loose upper bound on reaction time, since kukkii's aiming is not optimal and won't fit to a model. His mouse tracking is imperfect, like everyone else's in the world. And when the model doesn't fit properly, the results become more conservative.

For those who find the statistics hard, here's an easy visual way to see why:
When Marwan moves in a straight line for a while, kukkii wobbles his mouse like crazy but no correlation is being done.
When Marwan starts alternating his position, kukkii keeps wobbling his mouse, just like before, and now we get some really spurious correlations with Marwan's alternating position.
So when analyzed by the method in the article, kukkii's crazy wobbling gives him a big reaction time advantage when Marwan alternates, and doesn't hurt him when Marwan keeps moving straight. You can think of it like constantly guessing on a reaction time test, where the test doesn't penalize for wrong guesses.
I understand your point, and it's a good one. I agree that figuring out the model underlying his performance could be extremely challenging, but in this case I think the cross correlation method on the accel waveforms still does the job (although I'll have to examine the data to confirm this).

I'll explain why I think this to be the case tomorrow - it has to do with the meaningful changes in kukkii's movements (above and beyond the noise of his wobbling) accounting for the most variance.

Re: Interesting project about mouse/ gamepad latency

Posted: 10 Jul 2019, 01:30
by ad8e
While acknowledging that you'll improve your understanding of your method tomorrow, I'll point out one thing about the science now and introduce some background behind why the critical analysis and interpretation is so important. Scientists pick stories that look plausible and interesting. Then they do experiments to check if their stories are true. For example, "audio reaction times are faster than visual reaction times because of a shortened neural pathway" is a story, and it sounds plausible and interesting, so people tested it and now the data shows that it's true.

Manual tracking is similar to doing a regular reaction test, but replacing the button press with a mouse move, and wiggling the mouse before the test commences. All other differences are to the detriment of your manual tracking test. Since the mouse is moved, the initial acceleration can be detected, instead of a mouse button which only detects thresholds. And since the mouse is wiggled, the static friction of moving from a standstill is eliminated. Theoretically, this should be a really good input detection system, similar to Pain and Gibbs, which completes the story of how manual tracking reflex could, and should, exist.

The only reason this story was able to be written is because the papers were critically analyzed for the right explanations behind their amazing reaction time results. Without this critical analysis of the papers, there would be no reason to believe that manual tracking reflex is worth investigating, since little differentiates it from all the other possible reaction tests which are worthless. And since I don't believe that your correlation method has the potential to detect the right number, that would be the end of it even if the story is true. A story doesn't matter until it's supported by evidence, and the evidence is hard to get, and nobody will try for hard evidence unless they expect a payout.

But now, equipped with a good story, there's more reason to push forward when things become hard. So having the correct interpretation of results, not just the right results, is important because:
1. It helps us guess what the correct stories are. Good stories help us focus our research on the right area, instead of blindly fumbling. I don't know how you settled on LG out of all the possible things to test, since you didn't know the specific details behind the papers, so you couldn't have recognized the similarities. Maybe you got very lucky. But now that we have intuition, luck is unnecessary.
2. Correct analysis helps us design experiments.

kukkii jerking his mouse back and forth will be almost impossible to extract the correct information from. Even a statistician won't get good results, since his model will necessarily be so conservative as to kill any power. Exotic models need to validated to be used. (Yours is not.)

However, using our story, we can design a different experiment. The user pulls his mouse down vertically at a consistent nonzero speed. The reaction test asks him to jerk his mouse to the left when the screen flashes (or right, if left-handed). Now the filtering changes from nearly impossible to very easy, and this test still captures all the elements of our story. It is purely an effect of two physics phenomenon: elimination of static friction, and improved wavelet detection. Our new story has nothing to do with manual tracking or "distinct underlying neural mechanisms", which is good, because I can't possibly see how either of those things might exist.

But, since we have nobody fast to test on, I think that's the end of it. My reaction time is about 25 ms slower than do0om's and I don't have the equipment. Most likely your reaction time and Rejhon's are similar to mine, and everyone else abandoned this thread after I started talking about critical analysis.

Re: Interesting project about mouse/ gamepad latency

Posted: 10 Jul 2019, 02:38
by mello
ad8e wrote:...and everyone else abandoned this thread after I started talking about critical analysis.
I am pretty sure that a lot of people are reading the ongoing discussion in this thread.

Re: Interesting project about mouse/ gamepad latency

Posted: 10 Jul 2019, 02:40
by ad8e
mello wrote:
ad8e wrote:...and everyone else abandoned this thread after I started talking about critical analysis.
I am pretty sure that a lot of people are reading the ongoing discussion in this thread.
If so, that reflects really positively on them. Most people find something else to do when the thinking becomes difficult.

Re: Interesting project about mouse/ gamepad latency

Posted: 10 Jul 2019, 05:02
by 1000WATT
I can not understand your motivation and intentions.

Re: Interesting project about mouse/ gamepad latency

Posted: 10 Jul 2019, 10:01
by spacediver
ad8e wrote: The only reason this story was able to be written is because the papers were critically analyzed for the right explanations behind their amazing reaction time results. Without this critical analysis of the papers, there would be no reason to believe that manual tracking reflex is worth investigating, since little differentiates it from all the other possible reaction tests which are worthless. And since I don't believe that your correlation method has the potential to detect the right number, that would be the end of it even if the story is true. A story doesn't matter until it's supported by evidence, and the evidence is hard to get, and nobody will try for hard evidence unless they expect a payout.

But now, equipped with a good story, there's more reason to push forward when things become hard. So having the correct interpretation of results, not just the right results, is important because:
1. It helps us guess what the correct stories are. Good stories help us focus our research on the right area, instead of blindly fumbling. I don't know how you settled on LG out of all the possible things to test, since you didn't know the specific details behind the papers, so you couldn't have recognized the similarities. Maybe you got very lucky. But now that we have intuition, luck is unnecessary.
2. Correct analysis helps us design experiments.
You've got it completely backwards (again, you're making assumptions). At the time of the article, I'd played with kukkii for years, and as a long time quaker, I'd long been fascinated with high level lg performance, having experienced it for myself. That was the impetus for the article, and the manual tracking investigation was completed before I even began writing or researching the article.

In other words, the idea about manual tracking had been brewing in my mind for years.
ad8e wrote: However, using our story, we can design a different experiment. The user pulls his mouse down vertically at a consistent nonzero speed. The reaction test asks him to jerk his mouse to the left when the screen flashes (or right, if left-handed). Now the filtering changes from nearly impossible to very easy, and this test still captures all the elements of our story. It is purely an effect of two physics phenomenon: elimination of static friction, and improved wavelet detection.
That would be a good control to test against the manual tracking condition. However, since I'm hypothesizing the existence of a manual tracking reflex, I'd still need to measure performance during a manual tracking task!
ad8e wrote:Our new story has nothing to do with manual tracking or "distinct underlying neural mechanisms", which is good, because I can't possibly see how either of those things might exist.
Can you not concoct a plausible story about how the central nervous system is able to function more efficiently under certain conditions?

When it comes to eye movements, we have good evidence that smooth pursuit reflects a different neural pathway than saccades, so at the very least there is an existence proof of this idea of distinct circuitry underlying different modes of action.

Then there's the finding that manual tracking movements with the hand actually enhance smooth pursuit eye movements - something I just learned, see https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4669211/ where they investigate this with a randomly moving target (they actually use a remarkably similar set up to the one in my article).

Another important aspect is that we often find a relationship between reaction times, and the degree to which the perception of the stimulus, and the resulting motor response, are conscious (the excerpt from the Jordan Peterson video I linked to in the article is an excellent example of this). I wouldn't be surprised, for example, if the vestibular oculomotor reflex has a lower latency (relative to the onset of head rotation) compared to saccades, and indeed the vestibular oculomotor reflex is quintessentially unconscious (at least it is for me!)

One of the ideas in neuroscience is that conscious processing reflects recurrent and reentrant processing in the brain, where processing is dynamically updated between "lower" and "higher" brain regions, ultimately converging on a (perceptual) decision. This mode of processing, while powerful, also has a significant latency cost.

A distinct sensation that occurs when one is in the zone with the lg is that the movement feels unconscious (I discuss this in the article). Phenomenologically, then, this is a good starting point for investigation. It may be the case that this is simply a reflection of being in the zone, and perhaps manual tracking facilitates this more than other modes of aiming. One of my best profs used to talk about how the flow state involved being "dynamically coupled" with one's environment, and manual tracking of a target moving in real time is certainly a condition that lends well to this dynamic coupling.

So this investigation didn't arise out of a vacuum, and to preemptively dismiss the hypothesis, as a result of not being to see how it could possibly exist, might simply reflect a lack of imagination or open mindedness on the question at hand.

Note that this is very different from claiming that it does exist. It's a hypothesis, and may well be completely wrong. But I believe it's certainly worthy of investigation before we begin to rule it out.

I'll post the stuff about cross correlation later and the manual tracking data later, have too many things to take care of right now.

Re: Interesting project about mouse/ gamepad latency

Posted: 10 Jul 2019, 10:59
by ad8e
spacediver wrote:You've got it completely backwards (again, you're making assumptions). At the time of the article, I'd played with kukkii for years, and as a long time quaker, I'd long been fascinated with high level lg performance, having experienced it for myself. That was the impetus for the article, and the manual tracking investigation was completed before I even began writing or researching the article.

In other words, the idea about manual tracking had been brewing in my mind for years.
I didn't get it backwards. The ordering of events is only in reference to the new explanation, which is something that happened just now, and not in reference to your original motivations to study it. Your original reasons were not the justifications I describe in my last post, because you weren't aware of those justifications until I brought them up. You noticing that LG reactions might be sharper than click reactions is a nice spot.
spacediver wrote:That would be a good control to test against the manual tracking condition. However, since I'm hypothesizing the existence of a manual tracking reflex, I'd still need to measure performance during a manual tracking task!

...

Note that this is very different from claiming that it does exist. It's a hypothesis, and may well be completely wrong. But I believe it's certainly worthy of investigation before we begin to rule it out.
If you want to further test manual tracking specifically, that's your decision and it's fine with me. (I'm not the one doing the work!) Although, I won't help you in this direction, since I feel it's a waste of time.
spacediver wrote:Can you not concoct a plausible story about how the central nervous system is able to function more efficiently under certain conditions?

...

So this investigation didn't arise out of a vacuum, and to preemptively dismiss the hypothesis, as a result of not being to see how it could possibly exist, might simply reflect a lack of imagination or open mindedness on the question at hand.
I can't concoct such a story at the moment. The idea that manual tracking could be better than the other top visual reaction systems by any meaningful amount (even just 10 ms), in improving a reaction test like yours, doesn't agree with my understanding of the human brain. The reason I am ok closing my mind off to such ideas is because I have confidence in my worldview on the human mind. This is not something I can easily transfer to a reader; it's built off of years of research to get an accurate picture. As for the connections that you list from the literature, omitted from my quote, my expert worldview doesn't see how they might contribute to the existence of improved reflexes under manual tracking. But listing off an "expert worldview" is just arguing from authority, and since there is no logical reason for others to believe that I am a top expert on the mind or that my judgment is accurate enough, this makes it hard to convey a convincing message. My area of research in the mind is general cognition and not the typical neuronal studies, so feel free to keep bringing up research articles; I haven't seen them beforehand despite self-describing myself as an "expert".

As an analogy, if an unknown number comes up as an answer to a math problem and it's a number between 1 and 2, many people will guess it's the golden ratio. But as a mathematician, I understand how the golden ratio works and I wouldn't guess it except in special situations. This is because the depth of my knowledge gives me a better picture of what to expect and not expect, whereas the average guesser treats the golden ratio as a magical number which shows up unexpectedly. My guesses can still be wrong, but my background in mathematics sharpens my guesses.

In the context of manual tracking, your understanding of the mind is less than mine (this is only my own assertion). Since the brain is more mysterious to you, many things seem plausible, just like the golden ratio showing up in random places, because you don't have the information to rule them out.

But the evaluation of plausibility of stories is a matter of expert subjective judgment, not one of data. Even the experts often disagree on stories, which is why they pursue different research.

Re: Interesting project about mouse/ gamepad latency

Posted: 10 Jul 2019, 12:01
by Chief Blur Buster
On a slightly different topic,

A study on reaction-time stopwatch-starting and stopwatch-endings in the esports sphere would be lovely, to at least have something peer-reviewed released out there.

There are some situations where stopwatch-start can be a bit nebulous, or it's simply continuous (e.g. following a moving target that's randomly accelerating/decelerating), and contains very different reaction pathways that are not as well-studied.

Some highly trained professional esports players use multiple simultaneous cues such as pre-existing movement, peripheral vision, pattern recognition, audio cues, and more. Sometimes 4 simultaneous cues are occuring at the same time (e.g. 3 visual + 1 aural, etc), and some papers are only measuring 1 or 2.

I have seen (in person) an esports player do ultra-precise mouse movements similar to this:
https://twitter.com/slasher/status/1148 ... 09761?s=21
(Just one random impressive example)
So there are many situations where their precision isn't even cheating at all, so even admist the cheating, there's genuine uncanny precision like that Twitter video, confirmed by my eyes, in person. So I have no reason to disbelieve that such skill exist.

It's amazing how much mastery some professionals achieve. Not all of them though. Scientists and researchers need to cherrypick these mouse Mozarts and keyboard Beethovens to determine their tight-follow reaction time response (rather than one-event stimuli).

I almost bet my mortgage that milliseconds will be shaved off previously conceived limitations once various additional cues/factors/etc formerly unrecognized are factored into future studies.

Re: Interesting project about mouse/ gamepad latency

Posted: 10 Jul 2019, 12:02
by do0om
If you care about reaction time tests, we have an even better candidate in this thread, do0om, whose reaction times/equipment seem even better than flood's. I can supply him with a top-tier reaction test, if he's willing to do the trials.
If you want me to test it, I can