The Tetris the Grand Master grading system that runs from 9-1 and then S1-S9 (not all-inclusive) is clearly based on the kyu and dan ranking system found in Go, Shogi, and various martial arts, where the kyu ranks descend, from some high number (30 or less usually) down to 1, and are followed by ascending dan ranks from 1 to, usually, 9. An important difference between how ranks are assigned in, e.g., Go and TGM: Go is a 2 player game, and ranks are assigned within a pool of players, like a national association or an online server, and you ascend the ranks by winning against other players. Considering the game itself, it's only win or lose; in other words, your technique doesn't matter beyond whether it was good enough to beat your opponent. As far as the ranking system is concerned, it doesn't matter how many joseki variations you know or how deeply you read or even how many points you ended up with at the end of each of the games. Your rank ultimately is how many games you have won against similarly ranked opponents, and indeed is primarily a function for fair matchmaking. TGM, on the other hand, is primarily a single player game, and each game is graded based on your performance, as in how fast you are and how many high line clears you can make before the game is over. The similarity between the grading systems of TGM and Go ends at the superficial level. Let's examine this more.
In many martial arts, your technique and mastery of the various moves is considered when being promoted from one grade to the next. Promotion is normally a special event or test here, separate from the regular grind of practice and from competition. In TGM, at least in the context of being played exclusively in a game center, every game of Master mode is effectively a test of skill and a chance for promotion, assuming you've not already achieved the highest available grade. In TGM3, the "Promotional Exam" was added, but this more or less functions in the same way as the main game and is graded exactly the same; you're arbitrarily shooting for an average of some recent grades rather than the highest grade you can get and so is only mostly nominally different from a normal game. Also, in martial arts, the mastery of individual techniques such as certain punches, kicks, holds, and other general forms, is what determines your grade. In TGM, individual techniques are not considered, only very basic criteria measuring speed, line clears and overall consistency. The techniques you learn in the course of mastering TGM are a means to an end, rather than the point, of the activity. Here we see that the grade system of many martial arts, though hewing a little bit closer to grading in TGM in the sense that it's your actions within the activity that determine your grade, is still very different. It is, however, worth pointing out that a martial art is not a game in the sense that a TGM game's Master mode is a game with a set beginning and end, but a set of techniques employed within several different activities.
What I want to say is that TGM effects a superficial resemblance to graded non-video game activities through the use of the arbitrary signs of a kyu and dan-style grade system, or at least a modified version of it. They're arbitrary because it's only the signs that are used, how they are used (the meaning of the signs) is completely different from the two cases I've thus far discussed. Whereas your grade in Go is a matchmaking function and your grade in Judo is a judgment of how familiar you are with its techniques, grade in TGM is basically a stand-in for score (in the first game it is, in fact, based almost solely on your score). I feel it's important to mention now that I don't want to bring TGM closer to either Go or martial arts in terms of what their grades mean; it's a different thing from those two examples and would be better served by a grading system tailored to it. We could have a grade system more like Go's in the case of versus mode, which would make the grades a stand-in for an ELO system or something like that, but there is value in single player Tetris as well and there is value in having a comprehensive and sophisticated grading system for it. On the other hand, judging the presence and quality of individual techinques like wall kicks and synchros as if it were a martial art would be silly and meaningless; completely unnatural to the way the game is played. Imagine being judged in the game of Tetris on the way you kicked a T piece into a T-shaped hole. The height of stupidity! Glad that never happened.
So let's look at how the TGM series judges performance. In TGM1 it's almost exclusively by your score, which is increased most by clearing lots of tetrises, with a bonus for soft dropping and a bonus for what level you're on (so the loss of the soft drop bonus in high gravity is gradually compensated for by the increasing level bonus). It should be noted that these bonuses only apply to line clears and not to any other piece drops. Finally, only the Gm grade is subject to time considerations. In TGM2 there's a ridiculously convoluted system where your line clears increase a counter from 0 to 100, and then there's internal checkpoints that increase every time the counter hits 100, and then the counter will reset to 0 with no remainder, and the counter slowly drains based on the active time of each piece you play, and there's a points multiplier for every line clear based on what quarter of the game you're in, your grade, up to S9, is based on which of the 31 internal checkpoints you're on. It's really incredibly complicated for no reason and it amounts to another kind of scoring system based mostly on making high line clears, with the side effect that with the lack of rollover from one checkpoint to the next means you can waste some those high line clears by clearing them at, say, 90 points when, e.g., the tetris would've been worth 40. You wouldn't know you were doing this because while your grade isn't hidden during the game, everything that contributes to it is. Now, it doesn't really matter that much in the end if you're already aiming for M and Gm, but otherwise you can ascend most of the grades more efficiently by not going for tetrises all the time, because a lot of the time it doesn't matter if you clear an internal checkpoint with, for example, 2 tetrises and a double (100 points exactly) or 3 tetrises (120 points), because it amounts to you having 0 internal points left over in both cases. On top of this is a complicated system of time checkpoints for each section all of which determine whether you can get the M grade that comes after S9 or not. Gm you get by getting M and surviving the credit roll with invisible pieces. In TGM2 we have a grade system that's hidden from the player but mostly amounts to "make lots of high line clears", and time is only a real consideration when going for M. In effect it's a needlessly more complicated way of judging the same thing that TGM1 is judging. TGM3 is basically the same thing but it expands a modified version of the section time checkpoints into their own set of grades and expands the grade system of the invisible credit roll, which isn't really worth getting into for the purposes of this essay.
So setting aside the invisible credit roll challenge for now -- well, hold up, let's not set that aside. Why is that there? It's just a dumb add-on gimmick, right? Well, it actually makes sense for TGM2 and the context of playing the game in a game center exclusively. Arika wanted to enforce consistency for Gm, as in they didn't want players to be able to fluke it. They had to earn it. So in an environment and culture where, presumably, there is no way to practice the roughly minute-long credit roll in isolation, you would just have to achieve M, here analogous to the Gm grade from the first entry in the series in terms of what the game expects of you, over and over and over again so that you encounter Invisible Tetris so many times that you can eventually survive and get the Gm grade. When TGM3 introduced the Promotional Exam system, Invisible Tetris becomes a bit more gimmicky because it's no longer required to enforce consistency across games, but it's the sequel to TGM2 so how could it not have it? Here, along with the insane hidden grade system of both 2 and 3, it's necessary to point out that these are, first and foremost, arcade games. Japanese arcade games by and large are very fair, but arcade games they are nonetheless, and they are absolutely designed to take all the money they can from you. What better way to make money in a game center than to design an extremely mechanically refined version of one of the most addictive games ever created and add to it a mysterious and intriguing grade system and a ridiculous final challenge to engage the best of the best players? But, sad to say, arcades are dying even in Japan. Thank COVID-19 for a good bit of that but really the writing has been on the wall for years. Should Tetris at this level die with it? Must we be stuck with casual modern Guideline games and mere clones forever? Or is there just a little room for high-performance single player Tetris that follows in the footsteps of and even exceeds what Arika has done with the TGM series, a.k.a. the greatest 3 Tetris games ever produced? Given that Arika has spent over a decade trying to get TGM4 out without any success I'm not holding my breath for them to continue pushing the Tetris envelope forward as they had been closer to the turn of the century, so the best I'm able to do is figure out what TGM grades are aiming to measure, and what makes them work and not work, and come up with some ideas as to how others, either in an official or unofficial capacity, can implement more sophisticated grade mechanics into their falling tetromino games.
The primary aspect of TGM grading across all three games is the regular achievement of high line clears, with the tetris being the primary target of course. The game of Tetris is making tetrises, after all. And as the game increases the mandatory speed that the player has to match to survive, the player's ability to do so and even overtake the game's enforced speed naturally becomes another consideration. TGM and TGM2 only really award speed for the highest grades, namely M and Gm, and while TGM3 requires a basic standard of speed for each of its sections as part of its general grading system, they're quite lenient at the baseline, but change based on how you did in the previous section; the M requirements in TGM2 levels 500-999 work a lot in the same way. So what TGM3 is measuring here is not really speed so much as consistency. A very cheap way to see if the player messing up a lot is to check for significant time loss in the current section compared to the previous one; this is, I believe, the rationale behind TGM3's relatively lenient baseline requirements for the section time grades and the changing of the requirement to what the player actually passed the section with for the next section. Though cheap and easy to implement, it's also very clumsy and leads to the player sometimes stacking very high in one section and then downstacking rapidly in the next, and then not stacking so high for the next, and thus setting an unrealistic standard for that next section and missing its time requirement despite not really slowing down that much in their actual play either in speed, line clears, or general quality of stacking. This also leads to the players in the know of this mechanic intentionally gaming the system, i.e. intentionally playing slowly so as to not set the standard for the next section so high, so that they can continue achieving these grades. I believe that the internal grade checkpoint system utilized in TGM2 and 3 is designed as a consistency check as well, the quarterly multipler and lack of rollover between checkpoints adding a drag in the player's achievement of grades, meaning that it doesn't matter that the player got 7 tetrises in three or four particular sections so much as they got 3-4 in every section. This makes sense in the way that it solves the problem of TGM1 awarding too many grades too early, especially if you end up getting a tetris bravo or something like that. The problem with it is that achieving all tetrises actually awards grades slower, up to around S4 in TGM2's grading, than getting a mix of tetrises and doubles and triples due to a lot of those tetrises getting rounded off, which doesn't make sense any way you slice it. It means that depending on where your tetrises are getting rounded off, and you don't know this because it's all hidden and very difficult to mentally keep track of, you can unknowingly waste lots of high line clears and end up slogging through the early and middle internal grade checkpoints, which if you're particularly unlucky can really add up and have an effect on your ability to hit S9 late in the game. Alternatively, you could get lucky and land grade 1 early into the 400 section, which sets you up to get S1 earlier, and so on. It's not that it's necessarily unfair, it simply sometimes varies in a way inconsistent with the player's performance, which I think is an undesirable feature in a comprehensive grade system. Make tetrises all of the time and you don't have to worry about it because you're sure to get S9 that way, of course, but sometimes you just have more leeway to mess up for no better reason than pure luck.
So let's boil it down: Go fast, go for tetrises, don't mess up. That's what TGM wants you to do, broadly speaking. How can we implement this in a simpler way? Let's assume we're starting with, say, TGM2, and we're keeping all of its mechanics except for the internal grade system and the M requirements. We want to track speed, line clears, and consistency, and ideally we want each of these systems to be as simple as possible, and have the overall system be as simple as possible by having each one significantly contribute to the player's overall grade. For speed, we can simply have reasonably strict static time limits for each section that altogether add up to the 8:45 time requirement for M, because if we're measuring consistency in some other way then there's no need to change the limits according to the player's performance; we merely need to hold them to a minimum standard of speed. Here since the time limits add up to 8:45 anyway, there is no separate time requirement specifically for M, and by making these time limits a more significant part of the total grade system instead of just requirements for one grade we necessarily add the drag in grade achievement that the original grade system's rounding off and point multipliers brought, because you only get the grade effect from this system once per section. It wouldn't even have to be that harsh; we could have a secondary, more lenient time checkpoint that awards less so it's not all-or-nothing. For line clears, we can assign a small static number of points proportionally to each type of line clear that accumulate over the whole game like a score. Again, it being merely a component of a larger comprehensive grade system (though we probably should make it the most significant component) naturally makes it impossible for an overwhelming number of early tetrises to increase the player's grade too much; it's simply a question of balancing it.
With speed and line clears taken care of in these ways, how do we add another dimension of consistency to this grading system? When we look at how TGM2 and 3 handle section times, that is dynamically according to the player's performance, the apparent idea is to punish having to slow down to take care of mistakes, in other words making holes and struggling to fix them by making lots of low line clears for an extended period of time. This additional time spent fixing is what ostensibly causes players to miss the dynamically set section time requirement. It could also be caused by, say, the player slowing down and not manually locking most pieces because they can't keep up with the game's enforced speed for that new section, and for this we can just set good strict time limits for that particular speed and section. But for the consistency aspect, why not just penalize making holes in the first place? Not too harshly, but we could add some kind of point drain mechanic for every active frame that holes are present, perhaps an additive effect for multiple holes (not multiple cells in one hole). Perhaps we could also include a penalty for particularly deep wells numbering beyond the one used for making tetrises. There's actually hidden mechanics in TGM3 that measure general stack cleanliness like this that add to the hilariously vague Play Data info screen at the end of the game; why they didn't factor any of this into a better grading system I have no idea; time constraints maybe... so anyway, we could have static line clear point gains, static section time limit point gains, and a point drain effect for holes and other large irregularities being present in the stack all add up to some numerical value that is judged when the game is over and converted into the player's final grade (so the grade would ideally be hidden until the end of the game). Again, these are the types of systems we could use in tandem; there is just the question of balancing them. I don't want to get too deep in the weeds in terms of how exactly to implement these three systems and weigh them against each other and how to add them up into grades, and we now have a wealth of footage from Masters and Grand Masters to average from and determine what these limits and proportions could be, but I hope to have shown that we can precisely measure exactly what TGM2 and 3 want to measure without also introducing inconsistencies or other jankiness.
Alright so let's assume we have a well-balanced system of measurement for line clears, speed, and stack cleanliness that perfectly determines the player's performance for each game. Let's think again: why are we grading each game, though? In the case of TGM1 and 2 there's no persistent user accounts so that's kind of how it has to be, but what about 3? Here we can chalk it up to historical reasons, i.e. 1 and 2 worked the same way, and for the fact that, again, we're talking about arcade games. They have leaderboards, and there needs to be a way to judge games against each other. But for a new PC game with no such hard ties to the series or arcade gaming in general, we could perhaps give a general run-down of how the player did in all aspects of the game at the end of each game played, with hard and obvious values for the aspects being judged ("this is your percentage of singles and doubles and so on, these are your section times, you had stack irregularities for this much of the game," etc.), and promote the player when their average total game score, dictated by the system sketched out earlier, reaches a certain threshold (and demote them when it falls below some lower threshold). What I'm trying to say here is that we could perhaps divorce the concept of grades from ranking the outcome of individual games against each other; we could use a single overall score to rank games against each other in a way slightly different from grade, level and time, and your overall grade would simply be a different thing altogether. We could have an online leaderboard of individual games, and an online directory of players by grade. This approach isn't all that different from how TGM3 does things anyway, but the object here is to de-mystify and provide objective data for exactly how your games are being judged. The downside of all this would be that it could possibly make the overall game a little less exciting and more clinical or sports-like, so in the end I can't say if this approach would necessarily be better than the more arcade game-style one that TGM brings. It's merely a different perspective.
As much as I've slammed certain aspects of the grade systems in TGM, specifically 2 and 3, they're extremely refined at a level unseen in any other Tetris game, much like every other mechanic in the series, and they work remarkably well most of the time. I should also mention that 3 in particular softens some of the hard edges of 2's internal grade system and section time requirements, the former by allowing you to get the invisible credit roll with only an S7 instead of requiring an S9 (in TGM2 terms) and thus make up grades during the credit roll, and the latter by judging section times closer to the middle of the section than at the very end so getting caught at the level stop isn't as devastating. Maybe along the same lines TGM4 anticipates all of these complaints and more and corrects them in its own way! Hopefully we'll know some day. I'm not really out to accomplish anything with this essay, I've just been playing this series for 10+ years and I needed somewhere to dump all of these ideas. Maybe I'll go back to working on the clone I put together years ago and experiment with some of them.