Interesting project about mouse/ gamepad latency

Everything about latency. Tips, testing methods, mouse lag, display lag, game engine lag, network lag, whole input lag chain, VSYNC OFF vs VSYNC ON, and more! Input Lag Articles on Blur Busters.
ad8e
Posts: 68
Joined: 18 Sep 2018, 00:29

Re: Interesting project about mouse/ gamepad latency

Post by ad8e » 12 Jul 2019, 12:28

flood wrote:comment:
i think the humanbenchmark test works by turning green at a random time, but this random time is selected between 2s and 5s
which means that anyone who's done this test enough times, will (consciously or subconsciously or both) anticipate more and more if the color stays red for longer.
For me, it's consciously, but it doesn't make a big difference until around 50 trials, and then I start to get annoyed with how crappy and slow their reaction test is. These uniform distributions appear to be very popular with researchers, maybe because it's easy to describe the methodology. Looking at my own test, I see 800 ms + 800 ms * -log(random_float(0, 1)). The equivalent-ish uniform distribution requires an annoying amount of time to reduce anticipation to acceptable levels.

mello
Posts: 251
Joined: 31 Jan 2014, 04:24

Re: Interesting project about mouse/ gamepad latency

Post by mello » 12 Jul 2019, 16:08

Chief Blur Buster wrote: Experimenting with oneself's different stimuli, and mental states, is very educational!

I speculate that on a per-player basis may tune themselves subconsciously to their respective most-efficient cues of visual stimuli. People who are good readers (less dyslexic) might respond faster to text changes, while people who aren't partially colorblind may respond faster to color changes, and so on; as a speculation. Naturally, the drive for winning means the multi-stimuli opportunity of a esports session has many opportunities for an individual player to fine-tune to their most optimal stimuli automatically in a subconscious manner, as a rote of training.

I have to observe that esports is full of unexpected interacting stimuli factors
- FOV of stimuli (fullscreen-game-flashes, versus a single faraway target suddenly appearing, and everything in between).
- Simultaneous visual stimuli (shape-change, color-change, brightness-change, text-change, size-change, peripheral, flicker-flash, etc)
- Supplemental stimuli (audio)
- Gently focussing oneself on a specific stimuli of a multi-stimuli (like focussing on brightness/color instead of text or icon).
- Calm zen factor (professional solders/military/self-defense forces/etc fear for lives but trained to aim calm+stable)
- Pumping/priming oneself but not pumping onself to distracted-panic-attack levels.
- Nonstressed and I am slow (too relaxed), too stressed I am slow (too shaky), so prime myself to the correct level.
- Fear factors (fear of loss of rankings etc)
- Etc.
There are even more factors to consider:

- time and quality of sleep, and i would like to refer to Matthew Walker's book "Why We Sleep". You can also watch a lot of interviews and presentations from him on youtube. He describes how important sleep is, both when it comes to learning, cognition, performance, productivity, creativity and overall mental health and mental performance. You will learn why undersleeping and chronic sleep deprivation are one of the worst things you can do to your body.

- stimulants / "performance enhancers", meaning coffee, nootropics and drugs like adderall. To be honest, personally i would not focus on these things unless the effectiveness and a proper dosage is proven via RCTs. Also, just because something makes you feel different or 'good' it does not automatically mean that it improves your performance. You can often take way too much of something and make your performance worse, by being overstimulated, irritated or being way to relaxed.

- nutrition. Yes, contrary to the popular belief nutrition plays a major role in overall well being, and mental health performance. When you deprive yourself of proper nutrition, cause deficiences and malnutrition (even a mild ones), create nutritional and chemical imbalances in your body (again, even a mild ones), it all has a negative effect on your mental health and performance. These things make induce mental health problems, cause a person being irrational, easily irritated, aggressive, confused, and this in the end will negatively affect mental and physical performance. What is best brain food ? And again, to the contrary to the popular belief and to what nutritional science and nutritionists would like to tell you, the best foods for the brain and human body are basically meat, fish, eggs, liver etc. Basically animal based products, which is a species specific food for humans. You should focus on these foods if you care about your physical and mental performance and longevity.

- age. This also plays a role. Young people seem to be much more primed to perform very well in things like gaming and esports. Younger bodies are also much more robust and resilient to self abuse via things like drinking, partying, drugs, undersleeping, being overweight without too much consequences etc. Not saying that you can't reach an elite level in something while being in your 40s or 50s, but it is apparently much harder to keep up with the younger generation when it comes to gaming and esports.
spacediver wrote: Agreed, I've not delved deep, but I believe peak performance is a legit field of research, and it seems plausible that being in a flow state, or highly physically primed state, could have an effect on reaction times. If so, reaction time tests would themselves be a great way to test whether someone's in a state of peak performance.
Flow state requires positive feedback and positive results. In fps gaming, it means that you need to have to activate an internal reward system via overcoming hard and 'stressful' situations. It means a combination of things like: perfect flick shots, crazy actions and highlight worthy plays etc. Basically everything or majority of things needs to end up in your favor. Everything that can create praise and make you feel good (dopamine hit etc.). This way, as the game progresses, you are more and more in tune and closer to the flow state. When that happens, things like learning processes, motivation and cognitive performance are being enchanced.
Last edited by mello on 12 Jul 2019, 19:18, edited 1 time in total.

User avatar
Chief Blur Buster
Site Admin
Posts: 11647
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: Interesting project about mouse/ gamepad latency

Post by Chief Blur Buster » 12 Jul 2019, 17:20

Yes, good useful extra factors. I admit I had half a cup of coffee before my tests, so that may have somewhat helped accidentally.

So that’s called a “flow state” in your terminology! I am a more common garden variety 200ms mortal but I was able to fine tune myself to 150-thru-170 consecutive spurts after metaphorically clicking into that groove (or “flow state” essentially).

Given the prevalence of caffeinated-drinks sponsors (hello Monster, GoFast, and all the clones) and their continuous supply directly to the players, it certainly suffices to say that caffeine isn’t generally a banned substance in esports. I recall coffee was essentially defacto banned for athletes in Olympics from 1984-2004. Not anymore though (if in moderation).

In esports, the anticipation factor tends to blur into it all. A known-size level with a known-player-count, combined with known behaviours of your enemy....

....means an enemy may zoom by an opening (e.g. doorway in a specific location of a specific map) at a less-than-fully-random pattern or interval, and the slight excess of ammo in some levels, may produce anticipation-amplified events where some players will almost accidentally fire into a coincidentally appearing player, like a false start. This might be statistically meaningful.

Filtering anticipation might be sometimes hard in some esports contexts, given a potentially contentious boundary between a true false-start and a true triggered reaction....
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

Image
Forum Rules wrote:  1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
  2. Please report rule violations If you see a post that violates forum rules, then report the post.
  3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!

mello
Posts: 251
Joined: 31 Jan 2014, 04:24

Re: Interesting project about mouse/ gamepad latency

Post by mello » 12 Jul 2019, 19:17

Chief Blur Buster wrote: So that’s called a “flow state” in your terminology!
Yes, this is my interpretation of a "flow state" when it comes to gaming. This is also why warm-ups are important, both when it comes to optimal physical and mental performance. Same thing when you wake up, you are not able to produce your best effort in anything, your body needs some time to "wake up" on chemical level (hormones etc.). Both your brain and your muscles need time to get ready to whatever you will be doing during the day. And because of that, testing your own reaction time should also be planned and performed when you are at your peak (or close to it) during the day when you basically feel at your best. Also, considering an outside factors, such as good/bad sleep, being tired, too much coffee, too much stuff going in your life, additional stress etc. all of that can yield different results during different days. Not sure how much variance there will be in reaction time, but i assume there might be a substancial difference. This is also why people tend to have so called "bad days" when it comes to gaming or even LAN performance. It is hard to keep a consistency (both mental and physical), especially when you are busy (including long travels for example) and there are a lot of things going on in your life. This is also why esports teams always travel to the designed locations few days earlier before events & tournaments, get a hotel, hang around the city etc.
Chief Blur Buster wrote:I admit I had half a cup of coffee before my tests, so that may have somewhat helped accidentally.

Given the prevalence of caffeinated-drinks sponsors (hello Monster, GoFast, and all the clones) and their continuous supply directly to the players, it certainly suffices to say that caffeine isn’t generally a banned substance in esports. I recall coffee was essentially defacto banned for athletes in Olympics from 1984-2004. Not anymore though (if in moderation).
Coffee industry is a huge business and caffeine is neurotoxin, it is basically a plant poison. How these things work is basically by upregulating or downregulating certain hormones and pathways in your body. It causes oxidative stress, and your body responds in a certain way and tries to detoxify it and remove it from the body as fast as possible. The health benefits of coffee are way overblown in the media, and pretty much all of it is driven by epidemiological studies or short term clinical trials, and unfortunately none of that counts as scientific evidence (no long term RCTs has ever been done). Money from research also needs to come from somewhere, and conflicts of interest in scientific research are a common thing, especially when it comes to nutritional science. And when you add to that a famous quote: "There are three kinds of lies: lies, damned lies, and statistics." , then the reliability of health claims become even more questionable. There are also problems because you can get addicted and dependent of it, meaning when you need energy you basically tell yourself you need more coffee. It also can disrupt your sleep patterns / quality of sleep, due to caffeine half life and how long it stays in your systems. It can also mess up your electrolytes, affect your thyroid etc. And when you try to quit it, you get a withdrawl symptoms for a few days. But i digress... yes, there are obvious performance benefits to coffee. But if anything, it should be used strategically and sparingly, and not every day, few times per day, like most people are doing. When it comes to gaming, you can easily overdo the caffeine via energy drinks, and it will always reduce your performance instead of increasing it.
Chief Blur Buster wrote: In esports, the anticipation factor tends to blur into it all. A known-size level with a known-player-count, combined with known behaviours of your enemy....

....means an enemy may zoom by an opening (e.g. doorway in a specific location of a specific map) at a less-than-fully-random pattern or interval, and the slight excess of ammo in some levels, may produce anticipation-amplified events where some players will almost accidentally fire into a coincidentally appearing player, like a false start. This might be statistically meaningful.
Also, familiarity factor. When you are simply used to something and you see it everytime, after some time, certain actions and movements are done subconsciously. For example, when you are introduced to a new map in a game, it affects both your movement and aiming to a certain degree. It happens because you can no longer anticipate things happening / players appearing at moving at certain spots. You need to essentially re-learn all the patterns that will apply to that new map. This is why it is hard to get surprised by anything on your favourite map, but that can happen easily on a lesser known map or a new one that was introduced in the game.

spacediver
Posts: 505
Joined: 18 Dec 2013, 23:51

Re: Interesting project about mouse/ gamepad latency

Post by spacediver » 12 Jul 2019, 23:12

flood wrote:comment:
i think the humanbenchmark test works by turning green at a random time, but this random time is selected between 2s and 5s
which means that anyone who's done this test enough times, will (consciously or subconsciously or both) anticipate more and more if the color stays red for longer.
Maybe we're able to focus more attentional resources during short bursts of time, kind of like the zoom-lens model of spatial attention, but for time.

If so, then reaction times may improve when the (temporal) window, over which one has to prepare to react, is reduced. Can't remember if this has been studied or not.

But if it is the case, then if we're really interested in the limits of human performance, we'd want to reduce those time windows as much as possible. The downside is that this means guessing is more likely to occur, and it's more likely to produce false positives when it does occur. To solve this, you could introduce catch trials (where nothing happens) and get an estimate for the guessing rate, and its distribution as a function of elapsed time, and try to factor that out. However, the existence of catch trials may shift the criterion of responses (e.g. make observers more conservative), and reduce performance.

spacediver
Posts: 505
Joined: 18 Dec 2013, 23:51

Re: Interesting project about mouse/ gamepad latency

Post by spacediver » 13 Jul 2019, 09:11

Chief Blur Buster wrote:
In esports, the anticipation factor tends to blur into it all. A known-size level with a known-player-count, combined with known behaviours of your enemy....

....means an enemy may zoom by an opening (e.g. doorway in a specific location of a specific map) at a less-than-fully-random pattern or interval, and the slight excess of ammo in some levels, may produce anticipation-amplified events where some players will almost accidentally fire into a coincidentally appearing player, like a false start. This might be statistically meaningful.
This is absolutely the case. I often fire prediction railshots, and on occasion have a hit. These are purely predictive, with no reactive component. I've spoken to at least one world class quake player who used this technique also (czm).

mello
Posts: 251
Joined: 31 Jan 2014, 04:24

Re: Interesting project about mouse/ gamepad latency

Post by mello » 14 Jul 2019, 03:23

spacediver wrote: This is absolutely the case. I often fire prediction railshots, and on occasion have a hit. These are purely predictive, with no reactive component. I've spoken to at least one world class quake player who used this technique also (czm).
Same with using rocket launcher in quake, you make predictions based on players movement. And not only in a combat, but even when enemy player is moving through the map from a bigger distance, players have learned how to predict what others will do simply by looking if they are going in certain direction on the map. And as far as team games are concerned, when you are getting information about enemy location your anticipation factor is improved, which translates to better focus, aim and reaction time in that particular situation. But it is also double edge sword, because sometimes you may get a wrong or no longer accurate information from your teammates which can cause you being caught off guard.

nnnn20
Posts: 11
Joined: 27 Aug 2019, 19:53

Re: Interesting project about mouse/ gamepad latency

Post by nnnn20 » 18 Sep 2019, 02:29

I'm a bit late to this but I'd like to add my input as someone who has been closely following the humanbenchmark scoreboard for at least 3 years. It still exists btw! It is just not linked anywhere on the site anymore: https://www.humanbenchmark.com/tests/re ... eaderboard

However, in the beginning of 2019 for some moronic reason the owners of the site decided to remove the 100ms lower limit, which was the easiest way to catch casual cheaters, aka those who weren't dedicated enough to use software to cheat, which were the overwhelmingly majority given how low the incentive to cheat in a test like this is. It used to trick users into uploading their scores which had clicks below 100ms because for yourself they appeared to count towards the average, but when uploaded any clicks below 100ms were discarded, revealing people's real averages. But now any click counts, making it much easier to cheat... If the scoreboard was already almost meaningless before, it certainly is now.... Someone desperately needs to make a reliable replacement for that site, as it would be very interesting to look at data like this without cheaters ruining everything.


Over the years I have followed the site I have seen at least 10 users who seemed to be legitimately capable of consistently scoring averages in the 120ms-135ms range, myself included (never seen anyone averaging below 120ms who wasn't an obvious bot). Of course there is a slight chance that some of them could be cheating, but it certainly did not seem like it given their comments and interaction on the site overall. And well, at least I can guarantee I have never uploaded any cheated clicks myself.

If paying enough attention, I can easily average between 130-136ms without any missclicks, very often with clicks in the 125-130ms range. The best average of all time (in 5 clicks) was 125ms, and the lowest legit single clicks I have been able to achieve are in the 115ms-120ms range (probably 0.5% of my total clicks, very difficult), nothing less than this, so it seems to be around my limit for this setup. (g402, benq 144hz, win7 without aero)
my profile with results over ~2 years with decent hardware for the curious: https://www.humanbenchmark.com/users/5c ... 42df407cd4



I could provide proof if necessary but I feel like my results are not that remarkable for this to be necessary, the lack of scores such as these in the scoreboard is most likely caused by very few people possessing good enough hardware to score in such ranges. Claiming that scores below 150ms are not possible is just ridiculous, I'd guess the realistically achievable average currently possible with consumer hardware to be closer to 110ms. If you analyse the old data of the site, you'll see how much worse the overall scores have gotten over time, as a result of laggier hardware/software. For instance, the owner never updated his claim in the "About the test" section ("The average (median) reaction time is 215 milliseconds"), which is a value he got back in ~2009 (using the waybackmachine you can see that this line remains unchanged ever since the earliest snapshots of the site, which were made in the XP/CRT era). Here is an old screenshot which shows how much the results have changed ever since 2007: https://i.imgur.com/eGdvBHA.png (2007 average: 220ms vs 2019 average: 280ms). Also, a comparison of the percentiles of 2012 vs 2018:

2012:
110ms = 00.22%
136ms = 00.91%
160ms = 03.88%
165ms = 04.81%
192ms = 21.11%
202ms = 27.85%
204ms = 30.55%
207ms = 33.03%
250ms = 67.63%

2018:
135ms = 99.1%
153ms = 99%
170ms = 99%
180ms = 98%
185ms = 97%
195ms = 96%
200ms = 95%
207ms = 94%
214ms = 92%


In the span of a decade the average results basically became the top 10% results, a gigantic change.

Sparky
Posts: 682
Joined: 15 Jan 2014, 02:29

Re: Interesting project about mouse/ gamepad latency

Post by Sparky » 18 Sep 2019, 03:42

I don't see how humanbenchmark reaction times are useful anyway. Maybe it's okay for comparing yourself to your past self, but only if you keep the same hardware and software configuration. There are a whole lot of uncontrolled variables that prevent different peoples' scores from being comparable, and the inconsistency in the human element makes it useless for testing hardware or software.

Post Reply