Some old games are infinite, and don't have a proper end, then can roll infinitely. So as a result, you can play as long as you wish, until you reach the hardware limits, which is called the kill screen, and might be the closest thing to "finishing" a game when it's infinite.

I don't understand, what hardware limit?

In this case, the hardware limit is CPU time. NES Tetris will give you more points for clearing lines the higher the level is. For example, on level 0 clearing one line gives you 40 points but on level 9 clearing one line gives you 400 points. This is implemented in the code by adding the base number of points to the score in a loop where the counter is the level number: https://github.com/CelestialAmber/TetrisNESDisasm/blob/2bd89... NES Tetris is written with the assumption that its main game code won't run for more than one frame. When the player reaches a high enough level the "add score for cleared lines" code will take too long, and the processor will get interrupted for the start of the next frame while it's still adding the score. This causes the game to crash.

Why do they use a loop - does the 6502 not have a multiply instruction?

My non-expert googling suggests that there is indeed no multiply instruction for the 6502.

https://www.masswerk.at/6502/6502_instruction_set.html

Memory. Past a certain point, the numbers can become so high (score, level or other) that it corrupts the memory and freezes the game

The Tony Hawk's Pro Skater games can crash this way. If you cheat and enable perfect balance, then you can grind infinitely on a circular rail and eventually the score is so high it crashes the game.

That’s what I think makes the Tetris feat so incredible. There’s no cheating needed. This is being done on the base game after something like 40 years.

I think of it as "the first time anyone has not lost". Until that crash run, every game had ended with a top out.

Tetris is a deathmatch, and this is the first time Tetris died first.

It's the last thing that can happen in the game, which I think is a fair definition of getting to the end, or "beating" it.

When I made a game crash through a bug before I did not consider myself beating the game, that is why I have or had difficulties understanding. I suppose the difference is making it crash through playing as many levels, for example, as possible.

In order to get the "kill screen" on Tetris (or Pac-Man, etc) you need to play at an absolute master's level for an extended period of time

Also the Tetris player here isn't "exploiting" a bug. The game just literally hits a limitation that can only be hit via extended and nearly superhuman play

What if this "kill screen" were to happen at, say, level 30 due to bad code? Would it still be considered "beating the game"?

What's interesting to me is the "kill screen" in Tetris is (apparently?) at a variable point in play.

So, this was the first person to encounter the "bug" (hardware limitation). But, future players may try to hit that point faster, with a lower/higher score, or other combinations of game state to beat the game in a "better" (more interesting) way.

Yes, it would, and then you'd never hear about competitions in that game because it doesn't support ultra high-level play.

Well, it's definitely a judgement call.

I think most "serious" Tetris players can get to 30 easily in a short amount of time, right? Like 15-20 minutes? It'd be hard to think of that as a legitimate "ending."

Whereas a run to level 100+ or whatever for this kill screen is a major achievement. Took people 30+ years to get there.

I don't know, though. It's all subjective. If you want to think of a hypothetical lvl 30 crash as a "kill screen" or "ending" then hey... go for it.

Um, it wasn’t until 2011 before the first person achieved level 30, as the video mentions fairly early on

Thanks for the correction

no

I guess that's the difference between playing a game to the end, and exploiting a game to jump to the end or crash it. The former might be considered beating the game, the latter might not.

If it was just happening randomly and fairly commonly, I doubt anyone would care. The reason people care here is because 1) it's a game people have been playing for so long without triggering this state before, 2) the condition was known (from an AI playing, and subsequent analysis) and people had been actively chasing it and were pushing the limits on the number of levels completed at the same time.

Pushing limits in games is fun.

I've never played competitively, but I remember well my own excitement as a child whenever I managed to trigger a glitch (such as getting enough bonus lives to overflow the lives counter on Commando on C64, which was pretty simple but still was very exciting the first time it happened).

Some games don't have deliberately implemented "you win" screens and gameplay can in theory go forever unless the game has a bug that prevents advancing.

In a game with no human-authored intentional "win state", one reasonable definition of "beating the game" is "reaching a state that no other player is able to surpass". If the game crashes at a certain point, then reaching that crash is a viable state that fits that definition.

We're on hackernews. Think instead of the player triumphing over the game as a hacker.

It's an achieveable ending--one not planned by the coders, but one nonetheless. You cannot play past that point.

A couple of old arcade games have "kill screens", such as Pacman and Donkey Kong.

i reached the kill screen on Demon Attack on Atari as a kid. It was highlight of my early life

My understanding is that it's a bit of a roll of the dice exactly when the kill screen appears, so in theory you could continue playing longer. I suppose the significance, then, is that it's the first time that someone has played the game and not lost.

Because of the specifics of how this is done you actually can play past this point, but you have to intentionally dodge the crash and the conditions for the crash can be different per level.

It is technically possible to go up through level 255 at which point it loops back to 0. It's been done through a tool assisted run (carefully doing inputs frame by frame rather than playing in real time)

In this case, he even intentionally intentionally triggered the crash. As cool an achievement this is, I struggle to call it "beating" the game, since players could still theoretically go much further. Given that the glitch is predictable and well-documented, I'd consider it to be yet another (and new) obstacle to overcome, not the end. This is therefore the first time someone's ever hit the obstacle, and the first time someone's ever passed it (since he missed his first chance at glitching and had to take the second).

All that said, I feel a bit weird telling a community I don't belong to that they're incorrectly enjoying a game I don't play.

The game is man vs machine. The machine fills the space with blocks; the human clears space for the blocks. The machine wins when the human fails to make room for the next block. The human wins when the machine fails to deliver the next block.

this comment is fantastic and gives a clear answer to anyone griping about whether this is considered "beating" the game. scruti is the first human to beat the game, now other people will try to beat it in other ways on other levels.

Because the game is code.

It's impossible to do anything else within the confines of the system, and it's difficult to get to that end state without skill and intention.

Reverse engineering the NES Tetris cartridge shows that it has only so many possible states which is what allows one to find these limits.

It’s technically arbitrary and something the community sort of collectively decided was a coveted goal. Personally, I think it’s a cool milestone. There will be more similar milestones probably but few will be on true unmodified code, which I think makes this one truly special.