Do you want a short (1 minute) summary of this video because 44 minutes is just too dang long? Well, I made one which goes over the most important things about the experiment without going into any of the data: ua-cam.com/video/iVocj2szSok/v-deo.html . . Annotations, continued from the description: 7:10 - Dietary supplements as a whole are mostly scams and just make your pee more expensive. Some people have a legitimate need for supplementation, but if you aren’t one of those people, don’t bother. Here is an accessible video about this subject: ua-cam.com/video/TrYc0-NSZZg/v-deo.html . 10:05 - It might seem like improving ruins the experiment, but it really doesn’t. And I think it was unavoidable. Any test I designed I would have improved at with time. It just means that my data analysis method must either cancel out any improvements or be smart enough to consider and take advantage of them. By sticking to familiar games, though, my improvement was kept to a minimum. If I tried more novel games, my improvement would have been immense and might have actually started to impact the validity of the results, at least for the first several runs of the gauntlet - those being most impacted by a lack of experience. Given that I got better, it becomes important to pay attention to outliers of this trend. When did I do worse than before or did not improve? . 10:21 - For the attack helicopter, in order to limit the effect of getting it or not getting it within a game, I put a limit of 15 kills with it per match, and I could only call one in one time per match. Once 15 kills were reached, I immediately exited back to the ground, which incentivizes reaching that limit as quickly as possible. . 10:26 - I played on two different maps to help me from becoming incredibly bored of the map and a bit for my personal curiosity - which map would I do better on? I answer this in the data sheet linked above (row 600). Playing on multiple maps in a sequence has no effect on the results as long as the map selection is always the same for every gauntlet. . 10:46 - There is no way to limit their choices, and if there was I would have taken away gas grenades, smoke grenades, claymores, and the JOKR because they don’t take skill or reaction to avoid - you just die to them out of nowhere or they are disruptive without fair counter-play. With smoke grenades, bots can see through the smoke while you can’t, which is unfair. In hindsight, a game like Black Ops 4 may have been better for this test - less potential for balance updates mid-experiment and more bot options. . 11:16 - Any reaction time greater than 300 ms or less than 150 ms was thrown out as they are considered outliers. 150 ms was chosen because we could not get a time less than that without having preemptively clicked the screen before indicated to, and 300 ms was chosen because it is too high above our average reaction times to be indicative of our actual ability. I ended up getting a bit better at the reaction time test over time as well, which was unexpected. I think this is due to me getting more familiar with the test itself. The time it takes for the prompt to tell you to click the screen is random, but it seems to have a particular distribution of values which can become somewhat predictable with experience. Though we did not try to jump the gun and click before the prompt, if you know it is coming soon, you can prepare yourself to click. . 11:31 - Keeping track of both the average and median for every test I run is important because they reveal slightly different things, and when juxtaposed reveal nuances which only one data point could not convey. For example, if the median time is considerably less than the average, than there were probably some outliers on the higher end which brought up the average, but the median was barely affected. If the average and median are the same, the data likely fits more of a bell curve. “The Pit” also has some randomness - where certain targets will pop up and where the moving ones will be - hip fire is somewhat random too. Therefore, “The Pit” is not perfectly predictable (just like most video games). [The dropped frames present in the captured footage is not representative of my in-game experience. . 12:45 - It also is a good way for me to enter a flow state, or to get in the zone, which happened much more often during this test than any other. While in the state of flow, all other outside influence fall away, and it’s just me and the quiz - so the effects of the drink should be heightened as any potential environmental distractions go away. . 12:53 - The reason for splitting the MW2019 games in half like this is because I wanted to get in the reaction time tests closer to the peak caffeine concentration in the bloodstream (about 40 minutes after ingestion) and because it allows me to see if there was a dip in my performance from the first 6 to the second. If there was, that suggests no caffeine was taken because it supposedly would have reduced my fatigue. . 12:57 - All tests were done on the same computer using the same hardware (1 ms response time 144 Hz refresh rate monitor; 500 Hz polling rate mouse), except of course the Bosconian part, where I used an Xbox 360 (my brother also used the 360 for his gaming) - 60 FPS. For MW 2019, my average frame rate was around 85-95 FPS. Also, all tests were done wearing the same headphones without any outside distractions. If you are wondering, the reason I did not just do tests like aiming drills and reaction time tests in CS:GO is because I wanted to actually do testing in a real game environment. That was the entire point of the experiment - to see if you could see improvements from GFUEL in a complex field like a Call of Duty match. We still did a basic RT test to at least have one simple data point to act like as a reference point . 13:26 - The bots are excluded from using any shotguns or the assault shield because my brother was getting overly annoyed with those weapons. Particularly with the shotguns, since they are usually a one-shot kill, there can be no counter play possible. Having him do 10 games and 65 reaction times per sally makes the averages more representative of his abilities. . 15:31 - A third party couldn’t have gotten involved, since then they would know too. Considering this third party would have been either my mom or dad, the amount of contact we would be in during the testing could lead to some unwanted biases cropping in. This would defeat the purpose of the tests being double-blind - that being to eliminate any conscious or unconscious suggestions about which drink we had consumed. . 16:14 - Since I was making the drinks, and I did not trust my brother to make them correctly, when it was my turn to drink one of the drinks, I correctly labelled each drink as either GFUEL, sugar, or caffeine. And then my brother came in, and while I was away, he labelled the drinks with his own key and shifted around their positions on the table. He would leave, and I would then drink and record the letter. . 16:31 - We did this because caffeine takes around 30 minutes to take effect, while the GFUEL website claims 10-15 minutes. “Studies have found that the effects of [caffeine] are noticeable after just 10 minutes, but the peak caffeine concentration in the blood occurs after 45 minutes” (preview.tinyurl.com/ycfgkksh). Also, we made sure not to eat anything within the hour before the test and did not eat anything during the test either. The day of the test, we did not always eat the same food as that would have been impractical for us (though ideal); we did however try to eat a similar amount of a similar kind of food the morning before daytime tests and the night before nighttime tests. . 17:14 - It was important to test during both the day and night because caffeine is said to help you stay awake. So, if all testing was done during the day, when we diurnal creatures are most awake, then a big part of the supposed benefit of caffeine would have been ignored. The reason not all tests were done at night is because, practically speaking, it would have been more inconvenient to do - and, more importantly, many people do not consume GFUEL only at night; they also use it during the daytime. We therefore had to test if it helped us during the daytime as well as at night. . 18:11 - They can be considered independent because, though the half-life of caffeine in the body is around 4-6 hours, caffeine’s noticeable effects tend to dramatically decline after 4-5 hours; the tests were spaced apart by about 10 hours. Each day of testing, we felt pretty much the same before taking the drinks, so it’s not like one day we felt really crappy or something, which would create a bigger effect for the day-and-night tests than the single-test-per-day runs. . 19:28 - 50 RT makes the results a bit more accurate, and since I only did one round of the sleep deprived tests, I felt a more intensive test would be good. The reason I did not do another 35 at the end like with the night trials is that for these sleep deprived tests, I actually became more awake over time as the day went on. This is because humans tend to wake up more with light, so the more light I was exposed to the less tired I felt. Therefore, for these sleep deprived tests, I was always most tired at the beginning, not the end. This new labelling was done because I wanted to treat the sleep deprived runs as separate from the other runs, as they may be the most revealing about GFUEL’s effectiveness. If I had used the same key as before, there would be no way to guess specifically and only based on the sleep deprived runs. The reason the system was changed more drastically than it would appear was necessary is because I wanted to minimize the amount of work for my brother to do so he could go back to sleep as soon as possible. . Character limit reached, see reply to this comment for continuation.
Annotations, continued: 19:28 - To ensure only my brother knew the order of the letters I had, I put the letter labels on the bottom of the three identical cups. Once the drinks were made, I would mix up the order of the cups and then drink one, making sure not to look at the bottom. Once my brother woke up, he could look at the bottom of the cup I drank and record the letter. The problem with this is that for subsequent days, it would be possible for me to drink a duplicate drink. So, my brother had to dump any drink which I already had. . 20:13 - This was my first time having a traditional energy drink, and at first, I liked the taste - it reminded me of a Fanta, but it started tasting worse and worse over time. Way too sweet… considering it has 54 grams of sugar. And that’s why I was testing Monster - GFUEL claims these high-sugar drinks cause you to crash and are therefore not good. I say the insane amount of sugar is bad for you and your taste buds. So that’s why I didn’t test out Monster for more than a day - it’s just pretty gross stuff. . 22:36 - Sometimes this is done by news shows/sites, blogs, etc. with malicious intent to make differences seem greater than they really are. Here, I am just doing it to make the data easier to see. Hopefully my “bar of significance” will help alleviate any unintended consequences of messing with the y-axis. . 23:10 - I came up with the values really with just a gut feeling. For the reaction times, a difference of one 60 FPS frame seems significant. For Call of Duty and other games, it is harder to quantify, but I basically asked myself, “How much better would I have to do for me to notice that I was doing better?” However, just because two data points are within that bar of significance from each other does not mean that the two points should be treated as equal. It just means that you won’t really notice a difference in your performance in game. The bar sort of acts like an error bar except it’s not for error but instead your subjective feeling of performance. If you got one more kill in a game compared to the last, you probably wouldn’t think you had improved in that game. It could have just been that you got lucky with that extra kill. I debated instead of using a subjective mark of significance to just use one or two standard deviations of the set of data (all gauntlet runs together) as an amount that is significant. In fact, the standard deviation values influenced somewhat my own mark for significance. However, almost all my judgements for significance are equal to or less than one standard deviation. . 24:44 - The daytime runs do tend to have better scores, but all it takes is one slight mess up or one fast reaction time for the entire 65 times to be thrown off. Since no averaging is being done, a single outlier can ruin the data set. Even still, the data might be of some use to you. . 25:03 - If you want to see them, see the data linked in the description. About 393 rows down if you are having trouble finding it - results were a bit inconclusive, and it’s hard to tell even if being tired increases the deviation. The daytime tests do tend to have slightly smaller deviations on average, though. Standard deviations tend to be much smaller values than the actual measurements, which means the changes between them will be smaller and therefore more susceptible to random noise influencing the results. . 26:11 - This is similar to how in blind taste tests of name-brand vs generic foods, you generally assume the better tasting food is the name brand, even though it is likely that at some point the generic will taste better. It just makes things easier if we assume the best is always GFUEL, even if we logically know that at some point it is likely the caffeine could have better results. It also happens to fit nicely into my analysis method. . 28:25 - I have since achieved a 18.55 sec time. If you are curious about the difference between mouse + keyboard and controller, I actually recorded 128 pit run times using a controller back in 2018 as part of an experiment testing if music makes you better at games. I use the slightly slower M4A1 for those tests, which I calculate results in a 1-1.5 second slower time. My average time for those runs using a controller was 29.5 seconds, compared to my average here of 21.6 seconds. That is a massive difference. My best time ever achieved with a controller is 22.2 seconds. Of course, I might have gotten better since then, meaning not all of the difference is due to the differing input methods. . 29:10 - When recording the round reached, I included the number of enemy bases destroyed in the final round if it was not fully completed. So if I destroyed 4 bases out of 8 in round 13, I said I reached round 13.5. Due to the nature of the game of Bosconian, there are two viable gameplay strategies - 1. rushing through each round by destroying the bases as quick as possible, or 2. maximize points in the round as quickly as possible. I tend to do the latter. Good ways to do this are by taking out all the turrets on each base, each one awarding 200 points upon destruction, which then destroys the base itself, rather than just taking the quick and easy route of shooting the base’s core; trying to take out the entire group of the formation attacks, which gives a hefty point bonus, rather than just quickly taking out the head ship of the formation which calls off the other attackers; and shooting as many ships as possible. You can in theory stay on round one forever, gaining points by destroying oncoming ships, but bases award much more points, so it is worth it to progress through the rounds. To get the biggest score per round while also trying to get as much score as possible, you must balance staying within a round destroying ships and turrets and moving on to be able to destroy more bases. . 32:59 - I originally planned for there to be two different ones, but I scrapped the one I call the “relative ranking method” because it is honestly just worse than the other method I came up with. But it measures a slightly different thing, so you might be interested in it - check out the data sheet for this analysis (at BB, 531). I tried to design it as a simple analysis method, as compared to the % helpfulness, which is a bit more complicated, but making it so simple ended up making it too flawed to warrant being included. . 3:42 - I eliminated these data points for consideration before compiling the data and running the analysis, which is important because if I had chosen which data points to consider afterwards, I might have introduced biases (unconscious or not) which could have favored one drink or another. Like if I secretly wanted GFUEL to win, I could have taken out a data point for which GFUEL did particularly badly. I did not do this and only considered the appropriate data points based on other impartial reasons. . 34:14 - The values used to calculate percent difference were not rounded to the nearest tenth place to help alleviate rounding errors. That’s why, for example, the K/D’s for caffeine and placebo, though previously shown as 4.9 for both, have different percent differences from baseline. Without doing this, there would be a considerable error to the percent values, which would compound later on to make other values have even more error. I am only showing one decimal place for values, and later, two for percents, because that is cleaner. All values were in fact four or more significant figures when calculating the % helpfulness. . 34:56 - The paired t-test assumed the standard deviations for both sets of data are the same, and the data follows a normal distribution. There assumptions were basically met, at least enough for the t-test to be considered relatable. I only ran the t-test on the reaction times since those I could be most confident met the assumptions, but the other data points I’m sure had relatively small differences which constituted statistical significance (for example, my guess is 2 to 3 kills on average was a statistically significant amount). . 35:35 - Think of it like this - the bar of significance value is the difference from one night’s average amount of kills per match compared to the next night which would make you think you had done better that second night, and this new value is the difference from a weeks’ worth of matches to the next week. It would probably take a bit less of an improvement for you to consider it as an improvement for the week worth of matches compared to the day worth of matches because the sample size is bigger. Therefore, these new values can be smaller than the bar of significance values. . 37:05 - Removing one data point like I am doing simulates what would happen if, for that data point and drink, my brother had done equal to the average percent difference from baseline not considering that specific data point. Each of the ones I highlight in the video are outlier percent differences, so the new % helpfulness basically shows how the drink would have done if the outlier had not happened. But it did happen, so it cannot be ignored - I just thought it was good to point out the amount of sway a single data point has on the % helpfulness. . 43:01 - The flavor of GFUEL seems to be a major selling point, but plenty of drink mixes taste just as good at a much lower cost. To be fair, though, I have only had one of their flavors - maybe other kinds are so amazing, they are worth spending the extra $30. Perhaps you want to support a content creator who is sponsored by GFUEL. Merch or Patreon is a much better way to do that without having to buy a product which in all likelihood is either just a placebo or is literally 97.9% filler. But if you think GFUEL works for you, that’s fine by me. I have nothing against the company specifically.
Tyrosine boosts performance in stressful situation and when you're sleep deprived Sovern gaming: it doesn't help with anything gaming related Me playing ranked with randoms at 5am completely stressed and sleep deprived
Before I watch this, I did this experiment in high school. I had water as a control and gfuel as what I was testing. Then I had coffee, monster, and Gatorade as other tests. We would drink 1 cup, wait 10 minutes and perform 3 different tasks; all veteran FFA bots in call of duty, target practice with a football from 30 feet away, and a quarter mile race. We tried to do each day of testing about 3 weeks apart to make sure people weren’t getting better from just doing it consistently. It was a sloppy experiment admittedly as my two test subjects didn’t take it seriously. And it also didn’t account for crucial variables like the fact that some tests were after really long work days so some of us were already really tired. It didn’t account for the food and other beverages they had during the day, nor did it account for the test subjects continuing to play other games, my friends were very much into csgo so their skills may have been sharp despite being two different games. With all that said, my experiment showed that every single test had improved performances with gfuel across all three participants, while all the other drinks had varying levels of increased/decreased performance. We were juniors at the time, so 16, it could have been so much better if we put more than a couple minutes of thought into it. With all that said, I am very excited to see what a legit experiment will result in. TLDR: I tested the same thing in high school and showed it did increase gaming performance, among other things, but it was a sloppy experiment and I don’t think it has that much merit. So I’m excited to watch this video. Thanks for reading.
Yeah, my experiment has no merit other than getting a B on a project. This video was very well done though, the only thing I would say is that the experiments should be more spread out to prevent getting better and in theory it would better show how the different drinks affect performance.
Very true. If I was doing this as a real science experiment for publication, I would for sure have spread out the tests. But I wanted to get a video out in a reasonable time, so the testing was expedited. I did my best to keep in mind and control for improving over time, but there's only so much you can do.
What I learned today: G-Fuel is literally an energy drink in powder form. The only marketing point they should focus on is the fact that it's substantially cheaper than most energy drinks, 75% the cost of the low end ones. Had I known this before your video then I probably would have been drinking G Fuel by now.
24:22 but "The right input on the wrong frame can make all the difference in the world. So, wake up, Mr. Freeman. Wake up and... smell the G Fuel..." -The G-Man, _Half-Life 2_
The amount of work you put into this in every way is incredible. At some points it might be a bit too hectic and difficult to follow, but overall it's great and very informative. Keep it up!
This is really awesome and in-depth! I have no idea how you haven’t reached a wider audience yet, you definetely deserve it considering the insane amount of time and effort these videos must take.
never before has a video deserved a like this freakin much ur insane... like seriously i hope you recovered in the year after editing a 45 min video and doing all this testing. much love and respect
One problem I have with this (well the biggest one, not the only one) is that in a real life scenario, the placebo effect really does take place and you thinking gfuel will help actually does help so really what I would have done instead is to do this test on someone else you you're the one controlling everything. You just don't tell them that there is a placebo drink at all which makes it so no matter what, your participant will think everything is an energy drink thus making it so there is no bias as to knowing that one of them is a placebo or else the placebo effect wouldn't be present. I really appreciate what you did though because I am planning on doing a similar experiment and it's always good to take great ideas from others and improve on the aspects which faulter.
Your channel is literally the most underrated Channel ever. I don't get how the channels that put days and weeks into videos barely get any subscribers, but the youtube channels that clickbait and upload 5 videos a day somehow have 10 million subscribers.
I really enjoyed your video on the hardest achievement, so I think you should make a video on the hardest game. I know it's been done to death so it would be cool to see you finally answer the decades-old question.
when I drink gfuel I just get real jittery and then start getting a headache which goes away after eating but still lingers.. redbull is enough for me. I once saw a streamer chug a scoop and drop a 20 bomb on apex so fk who knows.. lol
It's more for the thought-provoking aspect (what exactly is wrong with GFUEL? etc, etc); and I do have some criticisms of GFUEL in the video, like how their formula is not backed by solid scientific research
The problem with your experiment, is that it only tests tactical skill, not strategic skills (a.k.a. problem solving under pressure). The reason people use supplements or energy drinks (or Adderall) for games like League of Legends isn't because it makes you hit your four buttons and right clicks faster. It's primarily to have your eyes be more alert and consistent in checking the minimap, making split second decisions on who to dive in team fights, deciding stuff like their kit at the macro level, etc. Doing this while stressed or tired is far, far harder than simply tapping buttons with a certain level of competency, unless we're talking at the level of high difficulty bullet hell (such as Lunatic Touhou, top difficulty DonDonPachi, etc.) or fighting games (SF, Tekken, KOF, etc.).
G fuel is mainly marketed towards gamers but if you read a paragraph on their website it says you can use gfuel for other physical activities and that’s why the boxer KSI has his own g fuel so I wouldn’t say the citrulline malate is completely useless
Interesting. But nothing will replace my green mint tea without sugar or my glass of Porto Cruz. BTW, I'm pretty sure I'm doing better when slightly tipsy, but that's probably BS.
I think player behavior is an important consideration as well. Your brother's higher average kills and lower average K/D shows that he's likely playing much more aggressively with g-fuel than with placebo. When looking at the data in a vacuum (i.e. reviewing the data blind instead of reviewing it with the context of which point is which) other useful behavior patterns can be determined.
If there is a compound that provable reduces muscle motor fatigue, that would be interesting as hands and fingers tend to get tired and stiff during long or intense gaming sessions.
great experiment, but this video is way to hard to follow. Waaaay too many data. You could have explained the experimental setup and showed the results (with deviation and so on) in less than 20 min and we would have got everything useful. The whole calculation process doesn't need to be explained. If you aren't familiar with statistics it won't help you anyway and if you are, you know them already. And the guessing part could have been shortened to "well, it we weren't consistently able to guess". for a sample of 2 persons this is way to precise statistics. (Still I like the effort you took to do this.)
But let’s be realistic how many gamers do you think genuinely get those vitamins on a day to day basis. The avg gamer generally does have a diet that is in need of more balance
me personally when i started drinking it i felt like i could react faster and had a lot of energy. Maybe it was really only a placebo but then again the flavors are really good so for me its worth (i also use it to workout and for the vitamins)
Hey Sovern huge fan of you channel just wondering if you were still pursuing UA-cam and if so best of luck making your latest video! No matter the content or topic I know it will be amazing quality and very entertaining.
Might be a weird thing to do but itd be worthwhile to write this in the format of a paper to be published, itd easier to present data and the testing seems exhaustive despite the sample size. Do you have a science background?
So what would you recommend is the healthiest form of energy ? Not looking for a cheat of supplement. I’d imagine exercise and healthy diet but anything specific beyond that???
See some gamers cant get into that awesome deep sleep its called sleep apnea it where your body when sleeping your throat closes and you somewhat wake up not in the sense of you wake up and you can notice what's going on your body just wakes up enough to go hey you're choking so you cannot get into a deep sleep even using a sleep apnea machine to help you sleep at night doesn't always work
now the bigger question, if it does, shouldnt it count like a steroid, because its giving the person an unfair advantage over people who doesnt/cant drink caffinated drinks?
bro, I can't believe you did almost an hr video about this too be honest I already knew it does not actually boost Gaming Performance at less there is no proof of it. It's just something they slapped on to it to make it stand out more than the other drinks like theirs. At less that's what i believe
Do you want a short (1 minute) summary of this video because 44 minutes is just too dang long? Well, I made one which goes over the most important things about the experiment without going into any of the data: ua-cam.com/video/iVocj2szSok/v-deo.html
.
.
Annotations, continued from the description:
7:10 - Dietary supplements as a whole are mostly scams and just make your pee more expensive. Some people have a legitimate need for supplementation, but if you aren’t one of those people, don’t bother. Here is an accessible video about this subject: ua-cam.com/video/TrYc0-NSZZg/v-deo.html
.
10:05 - It might seem like improving ruins the experiment, but it really doesn’t. And I think it was unavoidable. Any test I designed I would have improved at with time. It just means that my data analysis method must either cancel out any improvements or be smart enough to consider and take advantage of them. By sticking to familiar games, though, my improvement was kept to a minimum. If I tried more novel games, my improvement would have been immense and might have actually started to impact the validity of the results, at least for the first several runs of the gauntlet - those being most impacted by a lack of experience. Given that I got better, it becomes important to pay attention to outliers of this trend. When did I do worse than before or did not improve?
.
10:21 - For the attack helicopter, in order to limit the effect of getting it or not getting it within a game, I put a limit of 15 kills with it per match, and I could only call one in one time per match. Once 15 kills were reached, I immediately exited back to the ground, which incentivizes reaching that limit as quickly as possible.
.
10:26 - I played on two different maps to help me from becoming incredibly bored of the map and a bit for my personal curiosity - which map would I do better on? I answer this in the data sheet linked above (row 600). Playing on multiple maps in a sequence has no effect on the results as long as the map selection is always the same for every gauntlet.
.
10:46 - There is no way to limit their choices, and if there was I would have taken away gas grenades, smoke grenades, claymores, and the JOKR because they don’t take skill or reaction to avoid - you just die to them out of nowhere or they are disruptive without fair counter-play. With smoke grenades, bots can see through the smoke while you can’t, which is unfair. In hindsight, a game like Black Ops 4 may have been better for this test - less potential for balance updates mid-experiment and more bot options.
.
11:16 - Any reaction time greater than 300 ms or less than 150 ms was thrown out as they are considered outliers. 150 ms was chosen because we could not get a time less than that without having preemptively clicked the screen before indicated to, and 300 ms was chosen because it is too high above our average reaction times to be indicative of our actual ability. I ended up getting a bit better at the reaction time test over time as well, which was unexpected. I think this is due to me getting more familiar with the test itself. The time it takes for the prompt to tell you to click the screen is random, but it seems to have a particular distribution of values which can become somewhat predictable with experience. Though we did not try to jump the gun and click before the prompt, if you know it is coming soon, you can prepare yourself to click.
.
11:31 - Keeping track of both the average and median for every test I run is important because they reveal slightly different things, and when juxtaposed reveal nuances which only one data point could not convey. For example, if the median time is considerably less than the average, than there were probably some outliers on the higher end which brought up the average, but the median was barely affected. If the average and median are the same, the data likely fits more of a bell curve.
“The Pit” also has some randomness - where certain targets will pop up and where the moving ones will be - hip fire is somewhat random too. Therefore, “The Pit” is not perfectly predictable (just like most video games). [The dropped frames present in the captured footage is not representative of my in-game experience.
.
12:45 - It also is a good way for me to enter a flow state, or to get in the zone, which happened much more often during this test than any other. While in the state of flow, all other outside influence fall away, and it’s just me and the quiz - so the effects of the drink should be heightened as any potential environmental distractions go away.
.
12:53 - The reason for splitting the MW2019 games in half like this is because I wanted to get in the reaction time tests closer to the peak caffeine concentration in the bloodstream (about 40 minutes after ingestion) and because it allows me to see if there was a dip in my performance from the first 6 to the second. If there was, that suggests no caffeine was taken because it supposedly would have reduced my fatigue.
.
12:57 - All tests were done on the same computer using the same hardware (1 ms response time 144 Hz refresh rate monitor; 500 Hz polling rate mouse), except of course the Bosconian part, where I used an Xbox 360 (my brother also used the 360 for his gaming) - 60 FPS. For MW 2019, my average frame rate was around 85-95 FPS. Also, all tests were done wearing the same headphones without any outside distractions.
If you are wondering, the reason I did not just do tests like aiming drills and reaction time tests in CS:GO is because I wanted to actually do testing in a real game environment. That was the entire point of the experiment - to see if you could see improvements from GFUEL in a complex field like a Call of Duty match. We still did a basic RT test to at least have one simple data point to act like as a reference point
.
13:26 - The bots are excluded from using any shotguns or the assault shield because my brother was getting overly annoyed with those weapons. Particularly with the shotguns, since they are usually a one-shot kill, there can be no counter play possible. Having him do 10 games and 65 reaction times per sally makes the averages more representative of his abilities.
.
15:31 - A third party couldn’t have gotten involved, since then they would know too. Considering this third party would have been either my mom or dad, the amount of contact we would be in during the testing could lead to some unwanted biases cropping in. This would defeat the purpose of the tests being double-blind - that being to eliminate any conscious or unconscious suggestions about which drink we had consumed.
.
16:14 - Since I was making the drinks, and I did not trust my brother to make them correctly, when it was my turn to drink one of the drinks, I correctly labelled each drink as either GFUEL, sugar, or caffeine. And then my brother came in, and while I was away, he labelled the drinks with his own key and shifted around their positions on the table. He would leave, and I would then drink and record the letter.
.
16:31 - We did this because caffeine takes around 30 minutes to take effect, while the GFUEL website claims 10-15 minutes. “Studies have found that the effects of [caffeine] are noticeable after just 10 minutes, but the peak caffeine concentration in the blood occurs after 45 minutes” (preview.tinyurl.com/ycfgkksh). Also, we made sure not to eat anything within the hour before the test and did not eat anything during the test either. The day of the test, we did not always eat the same food as that would have been impractical for us (though ideal); we did however try to eat a similar amount of a similar kind of food the morning before daytime tests and the night before nighttime tests.
.
17:14 - It was important to test during both the day and night because caffeine is said to help you stay awake. So, if all testing was done during the day, when we diurnal creatures are most awake, then a big part of the supposed benefit of caffeine would have been ignored. The reason not all tests were done at night is because, practically speaking, it would have been more inconvenient to do - and, more importantly, many people do not consume GFUEL only at night; they also use it during the daytime. We therefore had to test if it helped us during the daytime as well as at night.
.
18:11 - They can be considered independent because, though the half-life of caffeine in the body is around 4-6 hours, caffeine’s noticeable effects tend to dramatically decline after 4-5 hours; the tests were spaced apart by about 10 hours. Each day of testing, we felt pretty much the same before taking the drinks, so it’s not like one day we felt really crappy or something, which would create a bigger effect for the day-and-night tests than the single-test-per-day runs.
.
19:28 - 50 RT makes the results a bit more accurate, and since I only did one round of the sleep deprived tests, I felt a more intensive test would be good. The reason I did not do another 35 at the end like with the night trials is that for these sleep deprived tests, I actually became more awake over time as the day went on. This is because humans tend to wake up more with light, so the more light I was exposed to the less tired I felt. Therefore, for these sleep deprived tests, I was always most tired at the beginning, not the end.
This new labelling was done because I wanted to treat the sleep deprived runs as separate from the other runs, as they may be the most revealing about GFUEL’s effectiveness. If I had used the same key as before, there would be no way to guess specifically and only based on the sleep deprived runs. The reason the system was changed more drastically than it would appear was necessary is because I wanted to minimize the amount of work for my brother to do so he could go back to sleep as soon as possible.
.
Character limit reached, see reply to this comment for continuation.
Annotations, continued:
19:28 - To ensure only my brother knew the order of the letters I had, I put the letter labels on the bottom of the three identical cups. Once the drinks were made, I would mix up the order of the cups and then drink one, making sure not to look at the bottom. Once my brother woke up, he could look at the bottom of the cup I drank and record the letter. The problem with this is that for subsequent days, it would be possible for me to drink a duplicate drink. So, my brother had to dump any drink which I already had.
.
20:13 - This was my first time having a traditional energy drink, and at first, I liked the taste - it reminded me of a Fanta, but it started tasting worse and worse over time. Way too sweet… considering it has 54 grams of sugar. And that’s why I was testing Monster - GFUEL claims these high-sugar drinks cause you to crash and are therefore not good. I say the insane amount of sugar is bad for you and your taste buds. So that’s why I didn’t test out Monster for more than a day - it’s just pretty gross stuff.
.
22:36 - Sometimes this is done by news shows/sites, blogs, etc. with malicious intent to make differences seem greater than they really are. Here, I am just doing it to make the data easier to see. Hopefully my “bar of significance” will help alleviate any unintended consequences of messing with the y-axis.
.
23:10 - I came up with the values really with just a gut feeling. For the reaction times, a difference of one 60 FPS frame seems significant. For Call of Duty and other games, it is harder to quantify, but I basically asked myself, “How much better would I have to do for me to notice that I was doing better?” However, just because two data points are within that bar of significance from each other does not mean that the two points should be treated as equal. It just means that you won’t really notice a difference in your performance in game.
The bar sort of acts like an error bar except it’s not for error but instead your subjective feeling of performance. If you got one more kill in a game compared to the last, you probably wouldn’t think you had improved in that game. It could have just been that you got lucky with that extra kill. I debated instead of using a subjective mark of significance to just use one or two standard deviations of the set of data (all gauntlet runs together) as an amount that is significant. In fact, the standard deviation values influenced somewhat my own mark for significance. However, almost all my judgements for significance are equal to or less than one standard deviation.
.
24:44 - The daytime runs do tend to have better scores, but all it takes is one slight mess up or one fast reaction time for the entire 65 times to be thrown off. Since no averaging is being done, a single outlier can ruin the data set. Even still, the data might be of some use to you.
.
25:03 - If you want to see them, see the data linked in the description. About 393 rows down if you are having trouble finding it - results were a bit inconclusive, and it’s hard to tell even if being tired increases the deviation. The daytime tests do tend to have slightly smaller deviations on average, though. Standard deviations tend to be much smaller values than the actual measurements, which means the changes between them will be smaller and therefore more susceptible to random noise influencing the results.
.
26:11 - This is similar to how in blind taste tests of name-brand vs generic foods, you generally assume the better tasting food is the name brand, even though it is likely that at some point the generic will taste better. It just makes things easier if we assume the best is always GFUEL, even if we logically know that at some point it is likely the caffeine could have better results. It also happens to fit nicely into my analysis method.
.
28:25 - I have since achieved a 18.55 sec time. If you are curious about the difference between mouse + keyboard and controller, I actually recorded 128 pit run times using a controller back in 2018 as part of an experiment testing if music makes you better at games. I use the slightly slower M4A1 for those tests, which I calculate results in a 1-1.5 second slower time. My average time for those runs using a controller was 29.5 seconds, compared to my average here of 21.6 seconds. That is a massive difference. My best time ever achieved with a controller is 22.2 seconds. Of course, I might have gotten better since then, meaning not all of the difference is due to the differing input methods.
.
29:10 - When recording the round reached, I included the number of enemy bases destroyed in the final round if it was not fully completed. So if I destroyed 4 bases out of 8 in round 13, I said I reached round 13.5. Due to the nature of the game of Bosconian, there are two viable gameplay strategies - 1. rushing through each round by destroying the bases as quick as possible, or 2. maximize points in the round as quickly as possible. I tend to do the latter.
Good ways to do this are by taking out all the turrets on each base, each one awarding 200 points upon destruction, which then destroys the base itself, rather than just taking the quick and easy route of shooting the base’s core; trying to take out the entire group of the formation attacks, which gives a hefty point bonus, rather than just quickly taking out the head ship of the formation which calls off the other attackers; and shooting as many ships as possible.
You can in theory stay on round one forever, gaining points by destroying oncoming ships, but bases award much more points, so it is worth it to progress through the rounds. To get the biggest score per round while also trying to get as much score as possible, you must balance staying within a round destroying ships and turrets and moving on to be able to destroy more bases.
.
32:59 - I originally planned for there to be two different ones, but I scrapped the one I call the “relative ranking method” because it is honestly just worse than the other method I came up with. But it measures a slightly different thing, so you might be interested in it - check out the data sheet for this analysis (at BB, 531). I tried to design it as a simple analysis method, as compared to the % helpfulness, which is a bit more complicated, but making it so simple ended up making it too flawed to warrant being included.
.
3:42 - I eliminated these data points for consideration before compiling the data and running the analysis, which is important because if I had chosen which data points to consider afterwards, I might have introduced biases (unconscious or not) which could have favored one drink or another. Like if I secretly wanted GFUEL to win, I could have taken out a data point for which GFUEL did particularly badly. I did not do this and only considered the appropriate data points based on other impartial reasons.
.
34:14 - The values used to calculate percent difference were not rounded to the nearest tenth place to help alleviate rounding errors. That’s why, for example, the K/D’s for caffeine and placebo, though previously shown as 4.9 for both, have different percent differences from baseline. Without doing this, there would be a considerable error to the percent values, which would compound later on to make other values have even more error. I am only showing one decimal place for values, and later, two for percents, because that is cleaner. All values were in fact four or more significant figures when calculating the % helpfulness.
.
34:56 - The paired t-test assumed the standard deviations for both sets of data are the same, and the data follows a normal distribution. There assumptions were basically met, at least enough for the t-test to be considered relatable. I only ran the t-test on the reaction times since those I could be most confident met the assumptions, but the other data points I’m sure had relatively small differences which constituted statistical significance (for example, my guess is 2 to 3 kills on average was a statistically significant amount).
.
35:35 - Think of it like this - the bar of significance value is the difference from one night’s average amount of kills per match compared to the next night which would make you think you had done better that second night, and this new value is the difference from a weeks’ worth of matches to the next week. It would probably take a bit less of an improvement for you to consider it as an improvement for the week worth of matches compared to the day worth of matches because the sample size is bigger. Therefore, these new values can be smaller than the bar of significance values.
.
37:05 - Removing one data point like I am doing simulates what would happen if, for that data point and drink, my brother had done equal to the average percent difference from baseline not considering that specific data point. Each of the ones I highlight in the video are outlier percent differences, so the new % helpfulness basically shows how the drink would have done if the outlier had not happened. But it did happen, so it cannot be ignored - I just thought it was good to point out the amount of sway a single data point has on the % helpfulness.
.
43:01 - The flavor of GFUEL seems to be a major selling point, but plenty of drink mixes taste just as good at a much lower cost. To be fair, though, I have only had one of their flavors - maybe other kinds are so amazing, they are worth spending the extra $30. Perhaps you want to support a content creator who is sponsored by GFUEL. Merch or Patreon is a much better way to do that without having to buy a product which in all likelihood is either just a placebo or is literally 97.9% filler. But if you think GFUEL works for you, that’s fine by me. I have nothing against the company specifically.
I like these long vids
You know mat pat from game theory? you are the new mat pat.
Bro you got me messed up if you think I'm about to read all that
Just say yes or no .... lol
Tyrosine boosts performance in stressful situation and when you're sleep deprived
Sovern gaming: it doesn't help with anything gaming related
Me playing ranked with randoms at 5am completely stressed and sleep deprived
Before I watch this, I did this experiment in high school. I had water as a control and gfuel as what I was testing. Then I had coffee, monster, and Gatorade as other tests. We would drink 1 cup, wait 10 minutes and perform 3 different tasks; all veteran FFA bots in call of duty, target practice with a football from 30 feet away, and a quarter mile race. We tried to do each day of testing about 3 weeks apart to make sure people weren’t getting better from just doing it consistently. It was a sloppy experiment admittedly as my two test subjects didn’t take it seriously. And it also didn’t account for crucial variables like the fact that some tests were after really long work days so some of us were already really tired. It didn’t account for the food and other beverages they had during the day, nor did it account for the test subjects continuing to play other games, my friends were very much into csgo so their skills may have been sharp despite being two different games. With all that said, my experiment showed that every single test had improved performances with gfuel across all three participants, while all the other drinks had varying levels of increased/decreased performance. We were juniors at the time, so 16, it could have been so much better if we put more than a couple minutes of thought into it. With all that said, I am very excited to see what a legit experiment will result in.
TLDR: I tested the same thing in high school and showed it did increase gaming performance, among other things, but it was a sloppy experiment and I don’t think it has that much merit. So I’m excited to watch this video.
Thanks for reading.
Yeah, my experiment has no merit other than getting a B on a project. This video was very well done though, the only thing I would say is that the experiments should be more spread out to prevent getting better and in theory it would better show how the different drinks affect performance.
Very true. If I was doing this as a real science experiment for publication, I would for sure have spread out the tests. But I wanted to get a video out in a reasonable time, so the testing was expedited. I did my best to keep in mind and control for improving over time, but there's only so much you can do.
What a nerd
@@louisdg7 'what a nerd'-🤓🤓
I’m super impressed by the effort put into this, good job dude the video is awesome!
Delta Predatorr Mad Respect
Coming soon: Does Ketamine Actually Boost Gaming Performance?
If you've done it, you know the answer
Considering it's horse tranquilizer first, and medically supervised depression treatment second, I'd wager probably not.
The amount of effort you put in your videos, definitely makes you a quality channel. Can’t wait for you to gain more subscribers!
Thanks for the kind words!
It baffles me that you don't have more subscribers, you make really high quality content, and put actual effort into your videos. Keep making them.
GREAT VIDEO, love this nerdy vibe where you pull out tons of stats/studies and the long video time makes it great as well.
What I learned today: G-Fuel is literally an energy drink in powder form.
The only marketing point they should focus on is the fact that it's substantially cheaper than most energy drinks, 75% the cost of the low end ones. Had I known this before your video then I probably would have been drinking G Fuel by now.
In the end, the true percent helpfulness was the friends we made along the way
How does this video not have millions of views this is super impressive the amount of effort and time put into this
24:22 but "The right input on the wrong frame can make all the difference in the world. So, wake up, Mr. Freeman. Wake up and... smell the G Fuel..." -The G-Man, _Half-Life 2_
The amount of work you put into this in every way is incredible. At some points it might be a bit too hectic and difficult to follow, but overall it's great and very informative. Keep it up!
Thank you!
Im impressed by the detail and dedication
We miss you Sovern! Hope all is well.
he died because he had to much energy drinks
This channel is such undiscovered gold! Every video is an absolute banger and easily deserves million views.
This is really awesome and in-depth! I have no idea how you haven’t reached a wider audience yet, you definetely deserve it considering the insane amount of time and effort these videos must take.
never before has a video deserved a like this freakin much ur insane... like seriously i hope you recovered in the year after editing a 45 min video and doing all this testing. much love and respect
the amount of effort put into this deserves like 1000 times the amount of recognition that it will probably recieve
One problem I have with this (well the biggest one, not the only one) is that in a real life scenario, the placebo effect really does take place and you thinking gfuel will help actually does help so really what I would have done instead is to do this test on someone else you you're the one controlling everything. You just don't tell them that there is a placebo drink at all which makes it so no matter what, your participant will think everything is an energy drink thus making it so there is no bias as to knowing that one of them is a placebo or else the placebo effect wouldn't be present. I really appreciate what you did though because I am planning on doing a similar experiment and it's always good to take great ideas from others and improve on the aspects which faulter.
Crazy how in depth this.
This deserves more views
where have you gone? we miss you
This is amazing. I'm so glad you put the working for this.
"Rarely if ever consumed caffeine." Are you sure you are even human?
I'll never drink it because of all the awful obnoxious "influencers" that promote it
Their marketing strategy is pretty obnoxious, but it seems to be working considering their sponsorship budget
G-fuel is like the raid shadow legends of energy drinks
Holy good video. Appreciate all the effort you spent on this. Great testing and vid.
Yall should come to the science field. I'm impressed with level of attention to details you put on this. Good job.
Your channel is literally the most underrated Channel ever. I don't get how the channels that put days and weeks into videos barely get any subscribers, but the youtube channels that clickbait and upload 5 videos a day somehow have 10 million subscribers.
It's really sad that he hasn't posted in 6 months
Very good and very interesting video. I really admire that you do all of the testing and experimenting. Keep up the good work :)
Play the story mode of a random Lego game without dying.
I really enjoyed your video on the hardest achievement, so I think you should make a video on the hardest game. I know it's been done to death so it would be cool to see you finally answer the decades-old question.
One nostril gfuel one nostril adderall
*LES GOOOOOOOOOOO BOYYYS*
😂😂😂😂
If you snort it, yes it does boost my gaming needs.
when I drink gfuel I just get real jittery and then start getting a headache which goes away after eating but still lingers.. redbull is enough for me. I once saw a streamer chug a scoop and drop a 20 bomb on apex so fk who knows.. lol
You should change the thumbnail, it look like a publicity :)
I thought someone would say that. I’ll see if I can come up with something better
@@SovernGaming With the cross out removed it would make more sense because this video doesn't seem to bash on or praise gfuel
It's more for the thought-provoking aspect (what exactly is wrong with GFUEL? etc, etc); and I do have some criticisms of GFUEL in the video, like how their formula is not backed by solid scientific research
@@SovernGaming By bashing I mean just to do so ( Example: because its popular) ofcourse some criticism is fine since its backed up with statistics
It did lmao
I see the love you put into your videos keep it up bro.
Waiting for my gfuel to come in, never tried it. First game I'm gonna play after drinking it is Geometry Dash.
Dose it realy increase focuse?
@@lx1429 ehh, not to me.
This needs wayyyyyyy more views
I swear this guy is Drif0r’s long lost brother, he sounds and looks just like him 🤣
Lol my actual brother says the same thing
@@SovernGaming 🤣
@@SovernGaming where have you been?
this video derseev WAY more views
This is so good! Made sure to share around, keep it up please!
Also this is a lot of dedication for a vid that only got 6k views.
Jesus you went all out on this
I love the thought put into the design of the Gauntlet, top notch!
Haha iced instant coffee go brrrr
Where have you gone
The problem with your experiment, is that it only tests tactical skill, not strategic skills (a.k.a. problem solving under pressure). The reason people use supplements or energy drinks (or Adderall) for games like League of Legends isn't because it makes you hit your four buttons and right clicks faster. It's primarily to have your eyes be more alert and consistent in checking the minimap, making split second decisions on who to dive in team fights, deciding stuff like their kit at the macro level, etc. Doing this while stressed or tired is far, far harder than simply tapping buttons with a certain level of competency, unless we're talking at the level of high difficulty bullet hell (such as Lunatic Touhou, top difficulty DonDonPachi, etc.) or fighting games (SF, Tekken, KOF, etc.).
Ok I just watched the vid and basically gfuel does help you with your gaming experience.... by ALOT. And I want my like now
Gfuel does help, I drank some and dropped a 20 bomb at a high school which is much higher than my 10 kill average.
In game of course.
Real life is a game :o
I can't stop thinking about how much effort you put into this, it's incredible
G is not GAME
G is GAMMA!
From the amount of research you’ve done you deserve more views and likes
You deserve wayyy more subscribers!
*after hearing the ingredients* at this stage it sounds like glorified caffeine
stocking up on gfuel for when cyberpunk 2077 comes out
Where did you go?
He's been active he replied to a guys comment 2 months ago
I hope he's okay and doing well. I really miss his videos.
Can you please do more is it possible to beat without touching a single stud!
Bro I just tried my first Bottle of gfuel.and you caught me peeing.
Can you test some of those mental performance enhancers? Looks like there are a few options
The caffeine helps me be better at gaming because I have ADHD
G fuel is mainly marketed towards gamers but if you read a paragraph on their website it says you can use gfuel for other physical activities and that’s why the boxer KSI has his own g fuel so I wouldn’t say the citrulline malate is completely useless
Interesting. But nothing will replace my green mint tea without sugar or my glass of Porto Cruz. BTW, I'm pretty sure I'm doing better when slightly tipsy, but that's probably BS.
Take a shot every time he says "my brother".
I think player behavior is an important consideration as well. Your brother's higher average kills and lower average K/D shows that he's likely playing much more aggressively with g-fuel than with placebo. When looking at the data in a vacuum (i.e. reviewing the data blind instead of reviewing it with the context of which point is which) other useful behavior patterns can be determined.
This is more scientific than most food theory episodes
That's some dedication
If there is a compound that provable reduces muscle motor fatigue, that would be interesting as hands and fingers tend to get tired and stiff during long or intense gaming sessions.
Apparently caffeine helps out with this, but to what extent I don't know
great experiment, but this video is way to hard to follow. Waaaay too many data. You could have explained the experimental setup and showed the results (with deviation and so on) in less than 20 min and we would have got everything useful. The whole calculation process doesn't need to be explained. If you aren't familiar with statistics it won't help you anyway and if you are, you know them already. And the guessing part could have been shortened to "well, it we weren't consistently able to guess". for a sample of 2 persons this is way to precise statistics. (Still I like the effort you took to do this.)
Yeah I thought about all that. It was a learning experience making this video
What if someone wants to know how these are calculated, who doesn't know it yet? :3
But let’s be realistic how many gamers do you think genuinely get those vitamins on a day to day basis. The avg gamer generally does have a diet that is in need of more balance
The answer to a question no one asked but everyone wants to know
Came from Star Wars lego vid. How do these videos not have more views…
Nice work !
Bro post mo you’re dope
me personally when i started drinking it i felt like i could react faster and had a lot of energy. Maybe it was really only a placebo but then again the flavors are really good so for me its worth (i also use it to workout and for the vitamins)
I'm going to have to re watch this. So much science.
We had to do a lot of gaming to counteract the small sample size. Next time you need a small sample size hit me up.
This man needs a Nobel Prize in science.
Using gfuel may not increase your gaming skills, but it WILL decrease your ability to get laid, so it’s not like it does nothing.
Defo should have at least 100k
Moral of the story? Brushing your teeth makes you a better gamer
Hey Sovern huge fan of you channel just wondering if you were still pursuing UA-cam and if so best of luck making your latest video! No matter the content or topic I know it will be amazing quality and very entertaining.
Yes, I am still making videos. Just took a bit of a break. A bit busy with grad school too. I plan to have another video out by Thanksgiving hopefully
@@SovernGaming Can't wait!
I see you made it over the 10 minute mark
Crazy how many people in these comments can say the same thing in different ways
Might be a weird thing to do but itd be worthwhile to write this in the format of a paper to be published, itd easier to present data and the testing seems exhaustive despite the sample size. Do you have a science background?
Yes, but I don’t think this is worthy of publication. Did you see my comments at the very very end of the video?
Holy fuck, this is a lot of studies done
Good job man
Good video! Respect the effort.
So what would you recommend is the healthiest form of energy ? Not looking for a cheat of supplement. I’d imagine exercise and healthy diet but anything specific beyond that???
See some gamers cant get into that awesome deep sleep its called sleep apnea it where your body when sleeping your throat closes and you somewhat wake up not in the sense of you wake up and you can notice what's going on your body just wakes up enough to go hey you're choking so you cannot get into a deep sleep even using a sleep apnea machine to help you sleep at night doesn't always work
Wh…what!?
So... almost all the ingredients basically mean nothing if you eat a balanced diet? WOW, WHO WOULD HAVE KNOWN!?!
Im going to do this for my science project wish me luck!!!
now the bigger question, if it does, shouldnt it count like a steroid, because its giving the person an unfair advantage over people who doesnt/cant drink caffinated drinks?
24:10 he is on controller so thats why the reaction time test was odd
bro, I can't believe you did almost an hr video about this too be honest I already knew it does not actually boost Gaming Performance at less there is no proof of it. It's just something they slapped on to it to make it stand out more than the other drinks like theirs. At less that's what i believe
This video is so underrated
They prolly drank a lot of fluoride
Me who's a kid damm I only can drink is gatorade I'm sad😟😟😟
I think that you could of tested something repetitive to test the jitterness
my guy there is no way u just put gfuel in a tall boi cup and used a spoon to...
this is REALY game theory
awesome video 👍👍