Tetris Matches are Getting Too Long. What's the Solution?

this is eric icks setting a new world record in nes tetris by being the first person to ever reach level 146 in the august 2022 masters event you may notice that this level is very dark that's because he has gotten so much farther than the games programmers originally intended that the code for the game colors is actually glitching out but beyond that eric set perhaps an even more significant record in the overall scene the longest competitive tetris match ever with this one game lasting 37 minutes and 20 seconds over four and a half times longer than a typical game used to last this is the apex of a wave of players using a technique called rolling developed less than two years ago that completely turned the meta of enes tetris upside down and has had games going longer and longer this has a lot of people in the community debating is it time to change the rules of the game one so that audiences won't get bored and two so that organizers won't run behind schedule because when you're trying to run a live tournament that operates on a set schedule in a venue with limited time you can't have matches going on far longer than you can reasonably predict and the two biggest tournaments in the scene today have both put out official stances on what they're going to do this month in september the online based classic tetris monthly tournament is introducing what is effectively a line cap on level 49 by having a rom hack start dropping pieces six times a second making it virtually impossible to play on and the classic tetris world championships next month in october which are happening live for the first time since 2019 have announced that they are not going to have a line cap or score cap or any limiting cap at all so was this the right decision will the world championships be a bloated train wreck and have to be cut short after running overtime well the hard part is that historically the only universally gathered statistic about aeneas tetris matches is who won and who lost so any discussion on how much games have increased in length and what the results of changing the rules would be has up to this point been reliant primarily on intuition and guesswork but i didn't think that was good enough i wanted some actual answers so for the past nine months i've been gathering tons of data on over 1 000 games between the top players in the scene and in this video i'm going to show exactly how much games have been increasing in length along with projections of what will happen this month with classic tetris monthly's line cap of 49 and what's going to happen next month at the classic touches world championships both if they do nothing and if they try out several different time-saving measures including a line cap to finally determine what's the most effective solution to begin with though we need a little bit of context on how the scene got in this situation in the first place so let's get started nes tetris was originally released in 1989 as a single player game but thanks to a bunch of fortunate coincidences it ended up working really really well in a competitive format where people play independent games head to head and try to beat each other's score chris tang commentator for the classic tetris world championship since the beginning described one of the most important coincidences in his talk at gdc in 2019 so when a player reaches level 29 nes tetris pieces fall at a rate of one line per frame making horizontal movement nearly impossible resulting in imminent death this effectively acts as a kill screen and will instantly end the game for nearly any player the kill screen behavior is unique to nes tetris and it cannot be played forever like classic tetris variants such as game boy version or tengen tetris the results of these mechanics create an imminent death situation and round times are around 6 to 9 minutes at most in duration which are a very good length for an esports match in 2019 this was correct the two major play styles were das which uses the built-in piece movement of the game by holding the side to side buttons down and hyper tapping which uses manual tapping of the buttons to achieve faster speeds in the world championships the game has started on level 18 with pieces taking one second to fall after 130 lines the game transitions to level 19 where pieces take two-thirds of a second to fall and the game moves up a level every 10 lines until the maximum speed is reached with pieces falling every 1 3 of a second when the player hits the level 29 kill screen at 230 lines das can't move pieces to the side at the speed which effectively ends the game and even though hyper tappers could sometimes squeak out a handful of lines in a tetris or two it was not really enough to impact the length of games this was the longest game in the entire 2019 championships where hyper tapper joseph sailey lasted just seven lines past level 29 and the overall world record wasn't much better at the end of 2019 the highest score world record of 1.35 million was also by joseph sayley in a game in which he only got a few lines beyond that to level 30. over the next two years hyper tappers slowly made progress in getting better at playing past level 29 but it was still very rare to see a game ever get beyond level 31 or 32 in competition and then in late 2020 one person changed everything a player named cheese combined several techniques for button mashing that he had observed in other games to develop a more efficient technique for nes tetris where he rolled several fingers across the back of the controller to push the buns into a finger on the top this technique was called rolling and it actually made the level 29 speed much more possible to play on so a ton of world records got shattered over the following year the longest game of the 2021 world championships which happened online was by roller hufflepufficus in which he got all the way to level 41 and the world record high score by cheese at the end of 2021 was a staggering 2.34 million in which he got all the way to level 61. because remember after level 29 the speed doesn't increase so if you can play past level 29 you can theoretically play forever this was around the time in the community where people really started talking about the potential for future matches to get too long and too boring so in december 2021 i decided to start keeping track of a ton of data points on every game being played in the classic tetris monthly masters event which features the 16 best qualifying players in the world playing best of five matches in a single elimination bracket this event was basically the perfect tournament to track for good reliable data because one it happened consistently once a month two the prize pool had grown to several thousand dollars which heavily incentivized all the best players in the world to play every time and three it was online so all the best players in the world could actually play every time and so now nine months later i finally have all this data that shows the trajectory of how games got longer in the past and how much longer they might get in the future let's dive in first one of the things i tracked was what playstyle players used past level 29 on each individual game as you can see the dash play style was already nearly extinct in this tournament but the hyper tapping playstyle was still pretty prominent at the beginning with 65 percent of games still using it in january but at this point it has been almost entirely replaced with 94 of the games using rolling by august playing past level 29 used to be a negligible part of tournaments making up only 3.5 percent of the total run time in december but now with the rise of rolling it's ballooned up to 19 in august the average game length has been a bit bumpy but also undeniably shows a clear trend upwards having also increased from 7.5 minutes in december to 9.7 minutes in august and finally here's how many games got to every level past level 29 as we can see there used to be a lot of games that ended very quickly past level 29 but now it's more spread out further to the right so where's eric's game on the chart oh wait this chart only goes up to level 49. some games are missing let's double the length of the chart okay now it goes up to level 69. eric's game still isn't on here the chart now includes every game except for eric's let's see where eric's game is let's keep going out going out going out ah there it is level 146. now this chart should tell you something eric's game was a massive massive massive outlier and the thing is it didn't even need to happen his opponent hufflepufficus only made it to level 66 and eric had already beaten his score by level 70. eric continued to play another 76 levels after he had already won this is what's known as a mullen it used to be taboo to do this in the early days of the scene because live events couldn't afford to waste time but the term got coined after a commentator encouraged ben mullen to keep playing during this match in 2017 because he was close to getting the first ever 1 million point score in a live competition the practice of mulling has become a lot more common in recent years because most matches happen online where there's less of a strict time constraint eric had kept playing for similar reasons to ben mullen he was already most of the way there to the world record in his own personal best so he decided to just keep going and going and this is a crucial factor when it comes to predicting the future because we can turn that level chart into a percentage chart which shows us exactly what percentage of games reach a certain point for example 33 of all games in august made it to level 35. if we apply a linear projection to these numbers we can expect it to rise to 39 in october and if we take all these percentages and use them to simulate 1 000 best of five matches we get this a projection of what the length of matches is likely to be for the top 16 in the world championships next month the 1 000 simulated matches are sorted from longest to shortest so over here we can see a drop-off with a small amount of matches that barely last 20 minutes and on the other end an exponential rise with a small amount of matches that last over an hour with the very longest matches possible in the simulation lasting just over an hour and a half overall the average match length is 37.2 minutes if we compare this to an identical simulation based off the results of the top 16 in the 2019 world championships the difference is huge matches never got any longer than 40 minutes and the average match length is only 22.4 minutes so that's a big deal matches are lasting nearly twice as long on average except here's the thing the world championships have actually put a big limit on how long games can go it's called eliminating the mullens they've announced a return to their policy in the past where mullens were strictly prohibited and this actually changes a lot i also tracked where every game gets to if you take out the mullens which in turn allows for a projection to be built without the mullens and this gives the result of you guessed it what the world championships would look like without mullens an average game length of 33.2 minutes a full four minutes shorter with many of those outlier match lengths completely eliminated this actually is more of an effect than adding a line cap at level 49 which would only drop the average game length to 35.7 minutes but what if the world championships eliminated the mullens and added a level 49 cap like the masters event well we can simulate that too the average game length would be 32.7 minutes only half a minute shorter than just eliminating the mullens what if they eliminated the mullens and moved the cap down to level 39 it'd bring the average down to 32.4 minutes only 20 seconds shorter than a level 49 cap as you can see there are diminishing returns here no matter how you slice it matches are still going to be much longer than they were in 2019. this is because there is another factor that currently has a much bigger effect on the length of matches than playing past 29 and well it's the matter of getting to level 29 in the first place in the 2019 world championships there were a lot of very early top outs probably a lot more than most people remember because they're unremarkable games so they fade away into memory since it takes more than 8 minutes to get to level 29 every early top out that instead goes to level 29 is the equivalent of a game that went to level 29 instead going to level 60 and players have gotten a lot more consistent in the masters event only 29 of games were making it to level 29 last december with no mullens and by august that number was up to 76 compare that to the 2019 world championships where only 10 of games made it to level 29 and it's not hard to see why match times have increased there are a lot more early top outs now going to level 29 than there are level 29 games now going to level 60. but the absolute biggest factor in how long a best of five match goes is perhaps unsurprisingly whether it's a sweep goes four games or goes all five over the past nine months in the masters event 46 percent of matches went three games 28 percent went four games and 26 won five games this is the data that i used to help run the simulations but the important thing to keep in mind is that these simulations just produce an average the world championships could end up with a bunch of games on the upper or lower end of the simulation for example when i showed the data earlier about the average game length in each masters event it suggests that the overall tournament is getting longer and longer with the most recent one being the longest but when i actually tallied up how long each tournament was it wasn't august but february that was actually the longest clocking in at approximately 8 hours and 42 minutes of game time and that's because february had the fewest sweeps and most five game sets out of any other month so how do you reduce this wide range of potential outcomes well it's simple instead of a best of five set you make it best of three if we retroactively make every master's event match best of threes 58 of them would have lasted just two games and 42 percent would last all three if we do a world championships 2022 simulation with no mullens and best of three the average match length plummets to just 21.2 minutes and there is much less variance and this is exactly what the world championships are doing in the early rounds they've announced they're returning to best of three like they used to in the old days in a nutshell if time is the main concern we can look at exactly what effect each time saving measure would have in isolation and see that the world championships have already implemented the two best time saving strategies next year might be a different story but for this year the world championships is absolutely fine without a line cap they just need a budget about 50 more time per game to accommodate the increased consistency of the players and heck you want to know what the most effective solution would be if consistency is the problem just eliminate the longest slowest part of every game level 18 that's where the majority of time is spent in every tournament there's a mode on the tetris gym rom hack called transition trainer that allows you to skip the first 130 lines and just start on the level 19 transition if we simulate this out for the 2022 world championships with no mullens the average match length even in a best-of-five set is only 15 minutes a whole six minutes shorter than full length best of three matches players fractal 161 and dog actually did a quick test game on this mode during down time at the lone star call this summer even though it turned out to be a super outlier game that took until level 49 to determine a winner the total run time was only 9 minutes and 20 seconds right in line with the maximum game lengths of old in the far future if organizers want to have as many games and matches as possible this might be what they choose to do because it just blows every other option completely out of the water and effectiveness and finally as for classic touches monthly as you probably would have guessed at this point that data suggests that a level 49 cap won't have much of an effect at all this month in the projections only two percent of games are expected to reach the cap without mullens which translates to only a single game across the entire to 60 game tournament and in the first couple matches we saw with the cap implemented sure enough none of the matches reached the cap except for this one game by gerald where his opponent miles topped out at level 35 and gerald mullen the rest of the way there the cap does seem to be affecting some players psychologically causing them to play more aggressively out of fear of a hard limit but at this moment in time it's pretty much just a placebo if 98 of games aren't reaching the mark the best strategy would be to simply play as if it isn't there so what's the big picture conclusion here am i saying that line caps aren't needed well not exactly this is a question i have been struggling to answer for the better part of a year what you just listened to is actually the fifth version of the script i wrote for this video i wrote different versions in january february july and august as i had more and more data to look at and the reason i rewrote the script so many times is because so much of the initial debate over line caps was over whether games would be going on too long so that was the question i was trying to answer and going back to the two original reasons people had for concern the answer is yeah games are getting a lot longer which is a problem for organizers but line caps are the least efficient solution and there are many other superior methods to cut down on time and if we look at the audience perspective what i've realized is that focusing primarily on the time question is actually a big distraction from the real issue games don't become boring just because they're going on for a long time a long game could theoretically still be exciting and tournaments could rework their structures to accommodate this if it were the case the actual best reason to implement line caps has to do with how it is affecting game strategy and that's the subject i'll be diving into in the next video where i interview fractal 161 right after he set a new world record and if the video is up by the time you're watching this you can click right on screen to check it out [Music]

Channel: aGameScout

Views: 192,627

Keywords: agamescout, nes tetris, classic tetris, agamescout tetris, classic tetris world championship, ctwc, line cap

Id: bDsaQsCWbw4

Length: 17min 5sec (1025 seconds)

Published: Thu Sep 15 2022