Keeping Classic Tetris Interesting

In 1989, Nintendo released their take on Tetris, the concept of which had been revealed 5 years previously by Alexey Pajitnov. Some quick math says that 1989 was… a long time ago.

There have been many iterations on Tetris since 1989, but NES Tetris lives on in many circles as the definitive “classic” version of the game. How is it that this title, among all the Tetris versions you can find, still has a following? And why is it that now, in 2022, the community is facing a crisis on how to keep their beloved game interesting?

This is a story of balance, popularity, innovation, and reimagination. We will start at the beginning.

Competitive Tetris

In 1990, Nintendo’s marketing agency organized the Nintendo World Championships, regarded widely as the first ever esports event. Players competed in Super Mario Bros, Rad Racer, and Tetris. The three games were morphed into one by a custom cartridge, but due to the strategy of the scoring, almost all of the actual competition would be decided by Tetris.

After 29 stops around the USA, the tour brought the regional finalists to Los Angeles for one final competition. The winners by age group each received a $10,000 U.S savings bond, a new car, a fancy TV, and a golden Mario trophy. Those winners were Jeff Hansen (11 and under), Thor Aackerlund (12-17), and Robert Whiteman (18+).

Thor Aackerlund in particular, was on a different level from every other competitor when it came to Tetris. He clearly had the skill to win the competition, but he additionally had claimed achievements in the game which were unheard of by his peers. Two such achievements were the maximum score of 999,999, and level 30.

999,999 (The Maxout)

I started playing NES Tetris during a popularity surge in 2018 (more on that later). I got my first maxout in May 2022, joining the ranks of some 400-500 players with documented maxouts.

With three and a half years of steady improvement, I reached the skill level where such a feat was barely possible for me. Thor had it in the first year of the game existing.

What makes a maxout challenging? Well, answering that kind of requires understanding what a game of NES Tetris “should” look like.

Anatomy of an NES Tetris game

Like most versions of the game, NES Tetris gets faster as a game progresses. There’s a direct association from which “level” you are playing, to how many frames it takes the piece to go down one row. Starting on level 0, it takes a comical amount of time for a piece to hit the bottom of the grid. However, on level 0 you also only score 1200 points for a Tetris, a 4-line clear achieved by placing the line piece (long bar) into a Tetris well.

The scoring is higher on higher levels, and a Tetris in general is worth (level + 1) * 1200 points.

The game lets you begin on any level from 0 to 19. No matter what level you pick, the game eventually levels you up, in what is known as the “transition”. Before the transition, you play on your starting level, and after transition the level goes up every 10 lines.

Errata: transition timing and line count weirdness

Due to some bizarre errors / lucky accidents in the code of the game itself, the timing of the transition has a weird relationship with the starting level. Starting at level 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, the transition comes at 10, 20, 30, 40, 50, 60, 70, 80, 90, and 100 lines respectively, which all makes sense. But then levels 10-15 all also transition at 100 lines. The fundamental reason is a hexadecimal/decimal conversion error, and as a funny consequence it is more optimal for score to start on level 9 than it is to start on level 15, just by measuring the number of points you could possibly have when entering level 16.

As for levels 16, 17, 18 and 19, the transition comes at 110, 120, 130, and 140 lines.

Anatomy continued

These line counts to transition are actually pretty small. If for example you’re someone who struggles at levels 16 and above, no matter the starting level you will probably reach your limit within 5-10 minutes. This gives rise to a “one more game” effect, where you start again. And again. NES Tetris boasts impressive replayability compared to other games where you commit hours every time you press start.

Now let’s think about the maxout. Knowing that a Tetris is worth (level + 1) * 1200 points, and the level goes up in this wonky way, how do you actually get to 999,999 points?

Sparing you the actual math, the best strategy is to start on level 19, and score Tetrises nonstop, past transition, until you land your 44th Tetris on level 22. A more practical strategy is to start on level 18 (which is significantly less difficult as a starting level than 19), and score around 40 Tetrises before level 29.

Levels and Speed

As mentioned earlier, the game tracks its speed by the number of frames it takes a piece to fall by one row. Due to the limits of small whole numbers, levels 13-15 all share the same timing of 4 frames. Levels 16-18 all share the same timing of 3 frames. Levels 19-28 all share the same timing of 2 frames. Finally, every level 29 and onward shares the brutal timing of 1 frame.

The game will lock a piece to the stack as soon as it tries to fall further and cannot. There is no grace period of “lock delay” present in most modern Tetris games. You must do all your maneuvering mid-fall. Decision making is still relevant, but it is inextricably tied to movement limitations. For this reason, even the brilliant players who can place 5+ pieces a second in other versions, struggle with NES Tetris.

To put it mildly, level 19 is really hard. Even highly skilled players have to carefully balance their risks to stay alive at that speed. Level 29, by comparison, is the game-ender. It earned the name “the killscreen”.

Maxout Game Plan

In theory, from a level 18 start you can get around 750,000 points before the game transitions to level 19. From there, you can theoretically get another 750,000 before hitting the killscreen. That’s more than enough for a maxout, but the difficulty lies in keeping a high Tetris rate. The Tetris rate refers to the percentage of cleared lines, which are cleared as part of a Tetris. The comparative value of single, double, and triple line clears is so low that the Tetris rate is a very effective proxy for score potential.

The rule of thumb is that a 60% Tetris rate, sustained throughout all of a level 18 start game, will get you a maxout. More commonly, a player will have a higher Tetris rate, something like 75%, during the level 18 portion. During levels 19-28, a lower Tetris rate like 50% will suffice. This allows safer play.

The Journey

As I worked my way up to my first maxout, I started out barely able to survive on level 18, and played level 19 only to laugh at myself. I couldn’t think fast enough, I couldn’t maneuver fast enough. One mistake turned into two, three, and the board fell apart instantly. Eventually I reached the point where I could consistently reach transition from a level 18 start, although my typical Tetris rate might be something like 30%. On level 19, Tetrises were the least of my worries, I was lucky to just get a couple levels deeper.

After some more practice, I became comfortable with the delicate balance of strategy on level 19. Commit without overcommitting, adapt your placements based on board health and height, do situational “quicktaps” to move a piece outside of the normal movement limits, at the expense of the next piece’s movement.

If we go back to Thor Aackerlund, and his 1990 maxout claim, we would assume it was a different, much faster journey. And we would be correct.

Thor’s playstyle: Hypertapping

What did I mean above by “normal movement limits”? NES Tetris will shift a piece to the side every 6 frames, when you hold down the button. There is an initial delay between the first and second movement, which adds a layer of complexity and techniques. This delayed-auto-shift (DAS) playstyle is sufficient to get a maxout, but it is not the only playstyle.

A viable alternative to DAS would be to mash the left and right buttons manually. If you could get taps fewer than 6 frames apart, it would allow you to move pieces outside the normal movement limits. For a game that runs at 60 Hz, that is 10 taps per second. Grueling but not unreasonable, especially considering that the most you would ever need to do is 5-tap bursts.

This is how Thor Aackerlund played the game. Tapping has a lower learning curve, certainly, since there aren’t any special techniques required to maintain the full range of motion. The only technique is press button many time, very fast. This allowed him and other “hypertappers” to have a higher range of motion on level 19, muscling past the delicate balance.

Playing on the Killscreen

Level 29 doesn’t have to be the end of the game. A couple of things suggest that it is supposed to be. For example, level 30 displays as level “00” in the game’s UI. This is a fact that Thor learned earlier than most.

Nowadays, there are several videos of DAS players reaching level 30. This is achieved by entering level 29 with a large center well and getting pieces that don’t need to move.

However, just like level 19, the extra bit of movement makes it a little easier for someone like Thor. In this iconic video, years after his 1990 competitive victory, taken as a sort of “proof he’s still got it”, Thor starts on level 19, gets a maxout and reaches level 30, all while someone impatiently knocks on his door.

One placement in particular is completely out of the question for a DAS player. That is this T, which requires 5 movements in only 16 frames. Thor’s tapping was nearly twice as fast as DAS. Or at least could be, sometimes.

But even with this exceptional tap speed, Thor makes a mistake on level 30, and can’t recover while on the killscreen. The game ends.

2010

The next real effort to connect a global community around NES Tetris came in 2010. (At least, global in the same sense that the Nintendo World Championships were global – search across America and only America).

The past two decades had seen lots of love for the game, but in the way games were loved in that era. Individual players, and maybe families or local friend groups, would all get better at the game in isolation. A few may reach out to try to find out if they were “actually” good, but to many this just wasn’t easy to do. If you knew about Twin Galaxies, you could join their rankings, but those were notoriously incomplete.

Robin Mihara, who had also competed in the 1990 Nintendo World Championships, decided to go on a mission. He wanted to round up the best of the best at NES Tetris and crown a champion. This quest is very well documented by the film Ecstasy of Order: The Tetris Masters.

Classic Tetris World Championship

The culmination of Ecstasy of Order is the debut of a new tournament: The Classic Tetris World Championship, or CTWC. Mihara got a crew of skilled players together, from Twin Galaxies lists, from word of mouth, and in the case of Thor, from decades-old notoriety, to all play the game under one roof, and see who would win.

Each of the players had some quirks to their playstyle, the kind of things that emerge from a self-taught craft. Dana Wilcox wasn’t aware that the rotation buttons were different and only used one of them. Harry Hong played with the Tetris well on the left side, which is known now to be strategically worse than on the right due to piece rotations. But each of them were the big fish in any local gaming circles, and it took Mihara’s efforts to connect them. They shared and marveled at each other’s gameplay.

Eventually a champion was crowned, and it wasn’t Thor.

Jonas Neubauer

The final match of the 2010 CTWC was Jonas Neubauer vs Harry Hong. Both players had claimed maxout scores in the past, and had incredible control over the DAS playstyle. The final was a single game, higher score at game over wins. Jonas took the game, 530,034 points to Harry’s 517,590.

Jonas would emerge as not just the game’s dominant player, but also an incredible community leader. Charismatic, uplifting, entertaining, he would make sure that his position of visibility would lead to the game itself succeeding.

The CTWC returned in 2011, and Jonas won again. Same in 2012, and 2013. Finally in 2014, Harry Hong took down Jonas in the finals, and secured the win. But Jonas came back in 2015, 2016 and 2017, winning all 3.

By this point, the scene had attracted a wide (and actually international!) following, and CTWC partnered with the Portland Retro Gaming Expo as the convention’s flagship event.

Sustained Interest

It seemed like NES Tetris had stood the test of time. A short but sweet implementation of the classic block game, where efficiency and strategy reigned supreme. The maxout and level 30 stood as lofty but achievable goals to inspire new players to keep pushing. For players who wanted more, a Game Genie could allow you to view scores beyond the maxout, and chase the “true score world record”, which Harry and Jonas had pushed past 1.2 million. YouTube videos and Twitch streams connected players to resources to learn faster. Classic Tetris was a wholesome and well-balanced global scene with an internet following.

The following grew significantly with the totally-complete-and-accurate summary of the 2016 CTWC below:

It’s amazing how a viral video can change an entire scene. Sure this is just two and a half minutes of well-edited commentator favoritism and cliché sayings, but the internet clicked. And then began asking questions, and a new generation began learning this old game.

The next generation

By 2018, there were some much younger competitors in the CTWC. In the most watched Classic Tetris video of all time, 16-year old Joseph Saelee took on Jonas Neubauer in the CTWC finals.

The production quality of the CTWC was also much higher by this point, and we can see the handcams of both players. Jonas plays with DAS as he always has, simple thumbs holding down buttons. Joseph however, has a fascinating left hand placement which gives him access to hypertapping. And as the commentary remarks, Joseph has not only reached level 30, but levels 31 and 32. He seemed to be on the cusp of sustained killscreen play, a feat thought not humanly possible for decades.

Joseph would sweep Jonas, becoming the youngest NES Tetris world champion and ushering in even more interest. This is where I joined the scene.

First signs of a crisis

The premise of competitive NES Tetris, higher score wins, had always led to good games. Nobody had yet scored a maxout in competition, so the rules around it went unexplored, although eventually there was agreement to just keep counting score, either using the Game Genie or manual addition. Everyone who made it to killscreen never made it more than a few lines further, so the score was a measurement of efficiency. Accordingly, strategy focused on learning the most efficient ways to remain accommodating toward the game’s harsh random piece generation.

Hypertapping in the era of Thor Aackerlund was a way to reach the same proficiency with raw speed and less technique. Hypertapping hadn’t yet been lauded as a way to turn NES Tetris into a marathon game. But some people viewed Joseph’s rise to world champion as the beginning of the end, since he would go on to stream the first ever level 33, level 34, level 35, only to be passed by other young hypertapping players like “EricICX” who rocketed all the way to level 37.

Eric’s game was noteworthy not only for the level reached, but for the fact that he recovered fully from at least one obvious mistake while playing on the killscreen.

The game doesn’t get any faster. If any person masters the killscreen, they can leave the maxout in the dust, and gain an unrecoverable advantage over those who still die on level 29. You wouldn’t even need to score Tetrises; a million points of singles and doubles would only take around 10 minutes of killscreen play.

But we weren’t there yet.

The Rolling Revolution

DAS operates at 10 Hz. A skilled hypertapper could reach 20 Hz. The fundamental speed limit of the game is 30 Hz, the button alternating between pressed and unpressed every single frame.

Propelled by the new network of teenage players, strategies began emerging for faster ways of manipulating the controller. For true competition play, it needed to be viable on a genuine NES controller, connected to a genuine NES. The less toll it took on the hands, the better (many already were frustrated by the physical demands of hypertapping).

Enter Cheez. Classic Tetris history will remember him as the godfather of rolling. The 14 year old who, late in 2020, revealed to the world how you can consistently and elegantly achieve 30 Hz inputs on a standard issue NES controller.

The gist of it is as follows: with a thumb on top of the controller, pressing lightly but not hard enough to register an input, roll the fingers of your other hand on the underside. The motion will push the button into your waiting thumb, while simple mechanics will push it away the following frame. 5 fingers, 5 inputs. You could do exactly what Tetris asked of you.

Learning curve

This strategy wasn’t exactly easy to learn. Cheez himself had quite a head start, having practiced it for several months before it went public, and he still felt as though it wasn’t ready during the CTWC 2020. During that event, he simply used hypertapping, and made it as far as top 8 before losing to an even younger hypertapping teenager (and future world champ), known by the name DogPlayingTetris.

The biggest challenges faced by early “rollers”, as they became known, were the rotation management, and the 5-tap. Rotations were awkward as they were done by the same hand that anchored the controller, and hovered the thumb, and they were input while the controller was vibrating from below. Several rollers succumbed to pieces flipping extra times, landing wildly on the stack, and causing a game to end well before killscreen.

As people grew more consistent, and the rotation problems faded from the spotlight, the 5-tap was the new hot topic. Deep killscreen runs would frequently end on a piece that needed to go 5 spaces to the left, but only went 4. When this happened, survival required another 5-tap with an even tighter timing window. 5-taps were the most demanding of the input sequences. Two inputs landing only one frame apart would get merged together, indistinguishable from holding down the button. Most runs below end to failed 5-taps.

How it has played out

In the video here from August 2021, Cheez reaches level 44, now fully 150 lines deep into the “developer-intended killscreen”. This also accounts for over 300,000 points of his final score.

In this video from a month later, HydrantDude becomes the first ever player to break the 1.6 million point barrier, which exposes a quirk in the Game Genie code used to count scores past the maxout. This is the first NES Tetris world record to be genuinely unbeatable without scoring points on the killscreen.

1.6 million became known as the “rollover”, and was the new goal for the elite players. Maxouts were a thing of the past.

CTWC 2021

In 2021, the final pairing was hypertapping defending champ DogPlayingTetris, vs elite roller Huffulufugus. A scene that used to be dominated by the same old crew that Robin Mihara gathered in 2010, had transformed.

2021 also was the first championships after the tragic and unexpected passing of Jonas Neubauer, at age 39. Tributes to Jonas are ubiquitous in the Tetris community, with “JONAS” being the name entered in the victory screen for so many players’ accomplishments. The 2021 CTWC trophy was a J piece, rather than the classic T, also in recognition of Jonas and how he had shaped the community.

As for who won the 2021 CTWC, the most recent at the time of writing, the trophy still went into the hands of the hypertappers. Dog secured the victory, with feats including a level 35 and three maxouts in the finals alone.

Feats of Fractal

During 2022 a player by the name Fractal decided he was willing to be “the bad guy” and win without Tetris efficiency. He became incredibly skilled at rolling. Fractal started on 50 Hz killscreen, 16% slower using a European version of the game (yay framerate standards). Using this, he achieved a no-Tetris maxout, and became the first human player to reach level 138, a magical threshold where the game logic fails to look up the correct color palette, and displays colors generated from unintended memory access.

After carrying his skills over to the NTSC framerate standard, Fractal quickly became known for his patient, singles-and-doubles killscreen marathon playstyle.

In this July match against Dog, who was also now a proficient roller, Dog attempted to counter Fractal’s playstyle by setting up for killscreen Tetrises. He got 7 of them, but the last one destroyed any hope of progressing further, and his game ended on level 43, with 1.6 million points. In competition.

Dog, the current world champion, had rolled over the score, a feat done for the first time less than a year prior. He was the first competitive player ever to do this with an active opponent. The only reason this was not cause for celebration, was that the opponent was Fractal. 400,000 points behind, Fractal had all the time in the world to single and double his way to victory. 4 tense minutes later, Fractal passed Dog’s score, let the pieces fall, and celebrated his victory.

The crisis

In the beginning, the game was fun to play. With 10 minutes, maybe you played well enough to maxout before succumbing to killscreen. The competitive scene was engaging, with players pushing each other to keep a high enough Tetris rate to have a chance. Sufficiently large leads were safe, as there was no way a player could catch up if they were about to hit the killscreen.

Now we know the ways of the past are suboptimal. Innovation has thrown a wrench into the balance of the game.

Compared to the past 3 decades, the optimal strategy has become defensive, careful, and anticlimactic. It leads to games that last far longer, and which end with a player under little-to-no pressure, doing the bare minimum.

The skill on display is higher, but the competitive appeal is now lower. This is the paradox of innovation in NES Tetris, and it’s a tale that plays out in many other games as well. Those games tend to be actively maintained, and can deploy “balance patches” to change the core gameplay in response. As absurd as the idea might sound, is it time for a patch? How do we keep Classic Tetris interesting?

Solutions

Tournament organizers are always tasked with making sure their rules elevate the accomplishment they want to elevate. If the CTWC were asking “who is the best at NES Tetris”, and they then set a rule that your game had to end on level 29, regardless of whether you can or cannot play at that speed, it would no longer feel like a valid competition. Someone who can play killscreen has put in the work to get to that point.

Classic Tetris Monthly‘s approach

Debuting next month, for September’s monthly online Masters Event, the organizer is implementing a “double killscreen” at level 49. This is a custom bit of hardware that will make pieces fall twice as fast, which functions almost identically to the original killscreen. Games will end, quickly if not instantly, on level 49. Suddenly there’s a finish line again, and efficiency will matter. Is this the solution?

DAS-only scene

In parallel, there is a competitive DAS scene emerging, in which players must play DAS to compete. This has given rise to some absolutely incredible matches (though not without some controversy over whether the restriction is being followed properly…!)

HydrantDude’s experiments

Recently, HydrantDude set up an impromptu bracket with a handful of skilled rollers, to play killscreen with a twist: start on level 29 (using custom hardware again), and reach 300,000 points in fewer lines than your opponent. If neither succeeds, then the higher score wins. A competitive ruleset where there was a target score, instead of a game-over level, was an interesting take. For the scene as a whole, would this be a good format, maybe setting the target score to a maxout? A rollover? Something else?

From Below

A very familiar looking game developed by Matt Hughson for the NES, solves the killscreen marathon by reducing the playable space. One row from the bottom will become unclearable, for every level advanced at the highest speed. Eventually you are crowded out. The “From Below killscreen” is yet another plausible balance change, that can make NES Tetris competition more interesting.

Singleplayer

Not only are the competitive games in this uncomfortable spot, but the elite players are now struggling to derive meaning from their own high scores. Once upon a time, the 100k point barriers were a nice progress tracker – if you were an 800k player, your goal would be 900k. If you were a maxout player, your next goal is 1.1 million. But now, 100k points is just a few extra minutes in the marathon, which may feel like a measurement of statistics and not of skill.

To put it into perspective, here is a slice of the world record progression:

All of the pictured records felt like meaningful improvements over the former – a player brought something new to the game, either new efficiency, creativity, or ability. And then it just rockets off into space.

The current world record is 6,492,500 points by EricICX (now rolling). He reaches the glitched colors, using the NTSC framerate standard. He is the only person so far to have done this, although Fractal and Huffulufugus are plausibly close to the same feat. The game lasts 40 minutes, and ends during a color that is practically too dark to play.

The only things left to do now on the original cartridge are to literally crash the game, or to perform ridiculous (TAS-only?) crash-avoiding tactics beyond a certain level.

Romhacks

Fractal has designed what I think is the most elegant solution to the singleplayer problem.

A version of the game, which runs on an NES, but allows for a whole host of logical framerates. Now instead of measuring improvement by how long you last on the same 60 Hz killscreen, you can select an 80 Hz killscreen, and test your rolling there (with improved hardware polling speed too, allowing 40 Hz rolls if you are precise enough!) The speedhack goes all the way out to 6x speed.

This is a great direction to take it, where the games can remain short and sweet but still push the limits of skill and innovation.

The Right Decision (and why it hasn’t been made yet)

CTWC 2022 is coming up, and I’ll be there. I probably won’t make a bracket, since there are hundreds of players who can beat my scores with either hypertapping or rolling. But I want to meet the community in person, and I want to watch.

The organizers have announced that the game will not be patched for CTWC 2022, and no additional rules will prevent marathon games.

On the one hand, I always give the right to any tournament organizer to declare whatever ruleset they want. The winner will be the person who plays those rules best. I don’t blame CTWC for delaying a rule change. However, it is probably going to leave a scar. The 2022 CTWC will have marathon games, probably not in any early rounds, but definitely during the top 16.

The title of Classic Tetris World Champion is shaped by what it means to have achieved it. For Fractal and Eric, it may mean who can stay focused for longer (and perhaps, who can make out the horrid colors of level 146 while playing on a stage).

Hopefully all of these ongoing experimental formats help give signal toward what works best. I believe that within the next year, we will gradually all coalesce around a ruleset more interesting and thrilling than the current status quo.

Until next time.