This topic is for the purpose of discussing #7373: p0008874's Uzebox 2048 in 00:46.29

Uzebox huh? Long time no see. Could you explain how this game works? All I see are numbers rapidly changing.

Uzebox huh? Long time no see. Could you explain how this game works? All I see are numbers rapidly changing.

It's based on a flash game that went viral in 2014. The best way to learn how it works is to play it here: https://play2048.co/

Uzebox huh? Long time no see. Could you explain how this game works? All I see are numbers rapidly changing.

It's based on a flash game that went viral in 2014. The best way to learn how it works is to play it here: https://play2048.co/

Thanks I will check it out.

I count 948 moves in this submission. Heh, my TAS attempt finished the game in 881 moves. But I want to see someone manage to manipulate RNG enough to get the minimum, 519 moves.

What a weird ending. There is a score counter and you went for "first 2048 tile" as an ending point?

What a weird ending. There is a score counter and you went for "first 2048 tile" as an ending point?

Yes.

What a weird ending. There is a score counter and you went for "first 2048 tile" as an ending point?

That is exactly the goal of 2048. To get to the 2048 tile.

https://en.wikipedia.org/wiki/2048_(video_game)#Gameplay

The game is won when a tile with a value of 2048 appears on the board. Players can continue beyond that to reach higher scores.[9][10][11] When the player has no legal moves (there are no empty spaces and no adjacent tiles with the same value), the game ends.[2][12]

edit: I've just wrote a lengthy post and got error 404 400 GG. Not going to repeat my write, so here's a short list - TAS should go for a high score - TAS very similar to that western reaction based game TAS which was rejected because all it differed from an unassisted playthrough is the speed

You can also continue to play Donkey Kong to get a high score even past the ending. There is no need to get so nitpicky with an April Fools Submission.

Fools or not, I find it pretty impressive to see a run of this. It probably won't get 2048 yes votes but it gets mine.

In case anyone really is fantasizing about a max score / max move TAS: A number of these console homebrew versions, including those whose randomness can be manipulated at will, are incapable of going beyond 8192 for no apparent reason other than the developers not willing to create tiles with 5 digits on them. (For reference, the original version goes as far as possible, which is up to 131072, the max tile value possible.) So if you really want something like a max score / max move TAS on a faithful port that is also manipulatable, I'm not actually sure if such a thing exists out there. (By the way, such a TAS is going to last hours. Not that length of TAS has stopped anyone.)

I count 948 moves in this submission. Heh, my TAS attempt finished the game in 881 moves. But I want to see someone manage to manipulate RNG enough to get the minimum, 519 moves.

I've been messing around with the game a bit (mostly out of curiosity). While RNG can easily be manipulated to get a new tile spawn WHERE you want it, I have yet to determine exactly how the game decides WHAT value to give to the newly spawned tile. Thus far, when a 4 spawns appears to be based more on the number of moves than it does which moves are taken or when they are performed. For example, waiting a frame or two will change where a new tile spawns, but it will not change what value that newly spawned tile has. Even delaying moves earlier than the one prior to the new spawn doesn't seem to change the value of the tile; at least in my testing. Unfortunately, I've yet to find a simple move counter of any type in RAM that determines 4 spawns. I tried to look into the source code to see how new tiles were determined, but I'm not a programmer and couldn't figure it out. Perhaps someone with more coding experience can. Having a tracelogger would likely help in determining how the game decides on 4's or 2's. For what it's worth, freezing the 4-byte value at RAM address 0x109 with a value of 0x56E509FE will always produce a 4 on the next spawned tile. If any one of the 4 bytes isn't frozen, new spawns may have a 2. Having this value frozen at the beginning of the game does not yield two 4s. One of the original spawned tiles will still be a 2. Regarding other RAM addresses (in case anyone decides to investigate things further), the value of the individual squares on the board are stored in 2-byte RAM addresses from 0x0A54 through 0x0A72. The sequence of addresses goes from 0x0A54 in the upper left corner, then downward in the column, then top down in the next column, and so forth with 0x0A72 in the bottom right corner. TL:DR Unfortunately, I don't forsee a way to manipulate enough newly spawned tiles to be a value of 4 in order to make that 519-minimum-move run even possible.

Actually, the given source code is missing some critical files. I was going to dig into it to see how the game uses RNG, and it turns out the code is simply not there. In addition, BizHawk does not have a trace logger for Uzebox. I managed to poke enough values to determine that the RNG (memory address 0x109, 4 bytes long, little endian) is sequentially determined as follows (but I was unable to determine how the game uses the value to determine whether the tile is a 2 or a 4): * The initial value is 1 * The next value is calculated from the current value N by: (0x41A7*N) mod 0x7FFFFFFF (unless N is 0 in which case the next value is 0x1F0CCE42) In any case, the RNG does not appear to change, except when determining the value of a new tile. So the tile value is unmanipulatable. Too bad. If you want to manipulate tile values, you'll have to use a different homebrew version of 2048. I tested some other versions some months ago: this NES homebrew version, as well as this Mega Drive / Genesis homebrew version both have manipulatable tiles (as well as manipulatable positions).

This movie has been published. The posts before this message apply to the submission, and posts after this message apply to the published movie. ---- [4640] Uzebox 2048 by p0008874 in 00:46.29

So if you really want something like a max score / max move TAS on a faithful port that is also manipulatable, I'm not actually sure if such a thing exists out there.

The original 2048 game is manipulatable (well... I mean there is randomness) and can go over the first 2048 tile. edit: and the android port does the same. When I wrote why this TAS didn't does it, I've got response like "this is just an april fools TAS" and "this is exactly the goal". But thanks for the editor or whoever who wrote the information to this TAS that there is no going over 2048 tile. I didn't knew that. edit: nevermind, looks like my point still applies. You can still go for high-score, but the author choose the first 2048 tile and it's OK for everyone else. case over.

When I wrote why this TAS didn't does it, I've got response like "this is just an april fools TAS" and "this is exactly the goal". But thanks for the editor or whoever who wrote the information to this TAS that there is no going over 2048 tile. I didn't knew that.

Even in versions where you can go beyond 2048, just getting there is a perfectly reasonable goal. Its right in the title of the game after all.

So if you really want something like a max score / max move TAS on a faithful port that is also manipulatable, I'm not actually sure if such a thing exists out there.

The original 2048 game is manipulatable (well... I mean there is randomness) and can go over the first 2048 tile. edit: and the android port does the same.

By "manipulatable", I mean in the sense that the tile values can be manipulated at will so you can get, for example, all 2s, or all 4s. (There was talk earlier about manipulating the minimum 519 moves, which requires all 4s.) This is really what is meant, when we talk about extreme goals such as max score / max move / min move / etc. As far as I know, there is no evidence that the original 2048 game, or its most faithful ports, can be manipulated to do such a thing.

Yeah I've thought we are having a communication problem. Manipulation... the original 2048 game which probably works the same on Android as I played both online and that port on my mobile phone changes the starting seed which gives you a few tiles of different numbers each time. FractalFusion meant about the level of manipulation being "anything at wish", the ability and possibility to manipulate any tiles to be the desired value. I doubt it's possible to do that especially after starting the game. But it's only depends on the developers' code of the particular game we are talking about.