He's done it. https://www.youtube.com/watch?v=aWJMZUCAhJ4 (Thread posted with permission. Sorry that this post is nearly a month late.) Featuring recording of the full physical laptop screen + hand on keyboard. For anyone out of the loop, Phantom Mania is a mode in Nullpomino that is essentially T.A. Death turned 100% invisible (plus TGM3 mechanics). Yes, that means surviving not 60 seconds, but a full 999 levels of invisible Tetris, all with the sadistically accelerating speeds that Death is infamous for. But this is Death's mechanics plus TGM3's mechanics, so of course survival alone isn't enough. First there's the three torikans at 300, 500, and 800 - reach these three levels too slowly and the game ends prematurely. The level 500 torikan is a little bit more lenient than original Death's, at 3:38 instead of 3:25. Second there's the grade requirements. Grades M, MK, MV, MO, and MM are straightforward enough - "simply" survive to level 300/500/600/700/800 and you've got it. Finally there's GM, the beast that zero people have managed to conquer - until now. Naturally, surviving to level 999 doesn't cut it. The grade also demands that you make a whopping 31 Tetrises in total, as well as 2 in each section 000-899 and 1 in the final section. And of course everything's invisible the whole way, so forget the location of just a few holes a minute in and you're hosed. Thought getting six grades in Ti's credit roll was hard? Other than Question_Mark, no one has ever gotten above an M grade on this mode, which is already more than ~twice as long as the credit roll. If I remember correctly, during the development of Nullpomino one of the developers had begged Zircean to make the Phantom Mania grade requirements easier. Looks like that might not be so necessary anymore. For those unfamiliar with the relevant history: 2006-07-09: Lockjaw creator Tepples documents an algorithm proving for the first time that with SRS, the Bag randomizer, Hold piece, and enough previews available, it is possible to play forever. The documented method requires 5 next pieces, low gravity, and 20 bags (140 pieces) to loop. It is mentioned that the shortest physically possible loop is 5 bags (35 pieces) since bags occur every 7 pieces and bravos can only occur every 5 pieces. However, the documented method is stuck at 20 bags because it has an STZ pattern that only loops every 4 bags. 2007-06-25: Tepples improves the Playing forever algorithm to only need 4 next pieces. 2007-10-13: Colour_thief improves the algorithm to only need 3 next pieces. 2008-02-06: Someone mentions on the Playing forever talk page that an optimal 5-bag loop is impossible, since it contains an odd number of Ts which cannot possibly be tiled into a even-dimensioned rectangle. This is quickly amended to say that any 5-bag loop would require making holes and fixing them as part of the algorithm, making it extremely difficult. 2014-12-13: More than seven years after the previous update, Question_Mark documents in a 2100+-word essay a 10-bag loop, using L and J pieces in the STZ stack to make them possible to loop on multiples of 2. This method is dubbed "Playing Forever V2". 2014-12-13: Less than a day later, Question_Mark posts a 2600+-word essay documenting the holy grail: a 5-bag playing forever loop. After a quick patch to fix a worst case scenario, this method, rigorously proven to be the shortest possible loop ever, is dubbed "Playing Forever V3.1". 2014-12-14: To add (quite a lot of) icing onto the cake, QM demonstrates that the PF-V3.1 method is doable in 20G. One remarkable technique (in my opinion) needed to pull this off requires rotating a Z piece left nine times in a row, which QM dubs the "epic flail". 2014-12-21: One week later, QM successfully uses the above technique to achieve an M grade in Phantom Mania, which requires a minimum of six loops (he pulls off seven). But the fastest parts of the mode are yet to come.... 2015-02-04: QM achieves the world's first MasterK grade in Phantom Mania. 2015-03-23: QM achieves the world's first MasterV grade in Phantom Mania. 2015-04-27: QM achieves the world's first MasterO grade in Phantom Mania. 2015-05-02: QM not only achieves the world's first MasterM grade in Phantom Mania, but also makes it to level 999 and clears the credit roll, earning him the highest possible grade short of GM. Unfortunately this is the end of the road, as he has been relying on his created Playing Forever method to conquer the invisibility that all others have yet to, and the method doesn't make it possible to clear enough Tetrises to satisfy the GM requirement. So there is no way to get any further...or is there? 2015-09-14: QM does the impossible and invents his own memory system that can consistently memorize the entire stack, documented in another 1600+-word essay and described as taking hundreds of hours of practice (full-time for three+ months) to pull off. As a result, the Bag randomizer is no longer needed. A highlight video uploaded to Youtube showcases loads of spin moves, half a secret grade, a downstacking challenge, and more, all done in invisible mode (and with the TGM randomizer). As QM so aptly puts it: "You've never seen Tetris like this before...and neither have I". 2015-09-15: QM uploads a video showing MasterO performed in Phantom Mania, without the Playing Forever method (and as mentioned above, using the TGM randomizer). 2015-10-17: QM clears Phantom Mania for the second time, earning another MasterM plus passing the credit roll, again without the Playing Forever method and with the TGM randomizer. 2 Tetrises short of GM 2016-03-07: QM achieves Grand Master in Phantom Mania...and the camera suffers a memory card error, losing the footage. Seriously. 2016-03-09: It is done. Invisible Shirase next?
Not to be an asshole, but I think QM kind of forfeited people at TC fawning at his achievements with his whole "fuck you I'm leaving TC and not coming back" act. If he wants to flaunt what he's done, he could start with addressing some of the concerns people have previously raised in the other thread (linked by @JBroms). Imo, if he wants to be snarky and quit TC as soon as people start questioning the legitimacy of his claims, then he can stay over at HD where people have no qualms blindly adoring him like a god, instead of having someone (you) post on his behalf.
seriously though what's with that video it's like a goddamn magic show or something with the music and ending
i'm not referring to the gameplay itself, but the method used to memorise what's going on. he's giving the impression with what he told us he was doing that either he is literally not human based on the speed he says he can memorise things, or he is lying or exaggerating his claims. jago (K) puts up some very interesting points if you read the thread JBroms posted. i'm not denying what he's done (if it's legit) is impressive, but his method is not what he claims it to be.
To be fair @BBQTurkeyzZ your previous post flippantly implies that the overwhelming consensus on here is the videos are bullshit, which I'd say isn't the case - at least definitely not for everyone. And as much as I personally think there are multiple fishy aspects to the whole thing (which have gone mostly unanswered), I also don't think there's anything like enough evidence to suggest that the videos are definitely fake. I'll also repeat what I said in the other thread, in that dismissing his achievements as not possibly by a human being seems kind of overly dismissive, especially considering that even moderate Tetris skills on can be regarded as 'not humanly possible' by your average uninitiated Tetris player. We're not TwinGalaxies, and we don't impose all records have stupid levels of proof to verify them. I get why some things might merit extra scrutiny, but at the same time I also think maintaining a "guilty until proven innocent" attitude to select people/records is unfair. I'm not a big fan of someone basically ragequitting in a huff and then continuing to have things on here beyond reproach through a proxy, but I do think a couple of people stepped over the line in the previous thread and there's literally no point even discussing things again if people aren't going to stay civil the second time around either. Keeping a party line of "well he's just a faking little dramaqueen" doesn't get anyone anywhere.
So, according to the OP, in the span of a year and a half (being generous), some dude supposedly mastered both discrete math and memorization, while apparently studying physics at the same time. We're skeptical that this is even possible, especially when his methods don't necessarily make sense, as seen in the previous thread. This isn't to say that it's impossible, but with this site having a number of users that have played the game for years upon years (many still not being GM in regular-ass TAP!), it's hard to see some guy achieving all this in a year and a half. Hell, the very best players can still have a hard time playing the invisible roll. That last thread was pretty brutal, I don't blame QM for leaving after that. That doesn't change the fact that we still have no way of knowing if he's for real or not. Considering what he's claiming and showing, we're going to be very skeptical about it without actual, physical proof. It's a board about Tetris, so we're going to be passionate about Tetris.
the "trickery" part was a bit weird and i don't know why i put it, but every person i've spoken to about this (zid, qlex, lav, drpete, muf, etc) all agree to some extent his method (not his skill) has not been explained fully. he claims to have a memory palace, but when that was challenged, he switched over to it being "more like ram". the "not human" comment does not refer to his skills, but, as kevin has pointed out, the amount of things he memorises in a given time. QM explained how he does it, though him "cheating" his own system seems kinda strange. i understand that this would work if his stack is clean, but what if it isn't, and he has holes everywhere? that would imply that using his method he would have to memorise and remember lots of things in a very short space of time (where is the hole? what does the top of the stack look like, and furthermore where do i place this piece in a way that it wouldn't impede the clearing of the hole? what does the next line look like?), which in turn leads to more questions. hence "illusion". i fully agree with you. i don't doubt the veracity of his skill, and i believe he does have some reasonable method of memorising what he's doing. either he's dumbing it down because he thinks people won't understand it, lying to us, or he can't explain it and is giving as close an explanation as possible. really? that's the first i've heard of it.
Maybe it's because I sit on this side of the fence, but aside from Jago, I didn't feel like there was really much egregious behavior. Maybe some people had more accusatory tones, but that's to be expected. QM himself said his feats are hard to believe. I just wish (giving him the benefit of the doubt) he chose a better timing to become "patient" with this community. Instead he argued back with full force during what he said is a busy period of his life then flipped the table to get the last laugh. I would like to say that while this can of worms open again and whether or not we actually believe QM, I hope this thread doesn't reflect poorly on our community like the previous thread (apparently).
I agree with this, though I do think Jago went overboard and personal enough that I wasn't surprised he buggered off. And maybe overall there could have been a bit less snark and a little more tact in calling things out. Whether they're telling the truth or not, someone's still always going to get defensive and annoyed when bluntly called out for it, though I guess there's no nice way to call someone a liar. I agree. I'd also add to this a few other circumstantial weirdnesses, such as his perpetual and critical lack of time whenever he posts, and yet he still somehow finds the time to get that good. Plus his complete narrowness of skill and apparent rejection of doing anything else besides what he's currently doing, and how instantly defeatist he became about the prospect of showing it off to someone IRL (it is a separate case in point that pretty much all of the top players here have had their skills somewhat verified through meet-ups or at least online mulitplayer, so in a sense it's not necessarily special treatment there). And none of these points are necessarily inexplicable or proof of anything, but collectively they don't help the cause imo. I was referring to OP in this thread. I don't mind people posting his results and tooting his horn a bit, though I'm also not surprised things have been called out a little.
oh i thought you meant proxy as in something like https://hide.me/en/proxy, though i get what you mean now.
Somewhat off-topic to this particular thread but, as far as playing forever is concerned, I find it quite surprising that, till now, no one has tried any finite method (could these methods be called satisfiability based methods or is there a difference?) to show (essentially prove) it (for various randomisers/playing styles). Now it wouldn't work if you weren't already certain that a given randomiser is easy enough, because the time complexity for exploring all possibilities will presumably be unfeasible. For the sake of an example consider the simple scenario of bag randomiser and any reasonably intuitive playing method (for 0G) we have in mind (leave no gaps, keep stack relatively flat, leave a line or two on the right for tetris etc.). To show that it works we keep the record of game states (stack and randomiser state) and wait for a repetition. If from the initial point (empty stack and initial randomiser state) the game states eventually stop growing in number then playing forever for that method is shown/proven. I also mentioned this (somewhat more rigorously) in a more general setting before. The main point from practical perspective in this specific case is to ensure that the number of states don't grow prohibitively large. That's why applying this for a specific playing method (could be deterministic or non-deterministic(more expensive) ) is better than generically considering all possibilities (which would presumably be unfeasible in most of the cases). Another point is that we might want to have a reasonable terminating condition (particularly important if we are in real doubt whether a given method works). Another benefit of making a well-designed program to make such calculations is that for a particular randomiser, once we seen that a given method works, if we store all the game states and transitions while making calculations, testing another method should probably turn out to be faster (given that both methods are applied on same randomiser obviously).
