News for hackers. Stuff that matters.
Re:Diablo hackers uncovered a speedrun scandal by lcnPylGDnU4H9OF on Saturday February 15, 2025 @10:35AM (#43059219)
> its own notion of ethics, legitimate and illegitimate exploits, etc.

There are run “categories” for any given game. Super Metroid’s arguably most prestigious category used to be called “Any%”, referring to the required item collection percentage as distinct from 100% or Low%, and now it’s called “Any%, No Major Glitches” because someone discovered that you can go out of bounds with the X-Ray Scope and a door. It specifies “major” glitches because the run otherwise makes heavy use of (among others) a “minor” glitch called Mock Ball, which lets you move in morph ball form at regular form running speed. Ultimately, the community has to agree on what is major and what is minor but it’s usually moot until a new exploit is discovered.

It is interesting and it makes a whole lot of sense once you look at it from this angle. The categories actually have pretty rigorously defined rules precisely because there is so much competition involved.

Re:Diablo hackers uncovered a speedrun scandal by jordigh on Saturday February 15, 2025 @12:47PM (#43060467)
It also depends on what is "fun". Major glitches in Super Metroid make you skip too much of the game and that just isn't fun to watch.

Speedrunning is a spectator sport, so spectators ultimately determine which categories are the prestigious ones.

Re:Diablo hackers uncovered a speedrun scandal by djur on Saturday February 15, 2025 @11:37PM (#43065376)
An example I'm aware of is the Lufia II No Major Glitches/No RNG Manipulation run. That game is very easy to make essentially deterministic with RNG manipulation, but like most highly technical RPG runs it's long and exhausting. The No RNG run not only bans RNG manipulation but also certain very rare drops that RNG manipulation is usually used to obtain. Why? Because otherwise, the optimal strategy would be to grind runs 10 minutes at a time until you got the rare drop from an early boss, and that's not fun. So the ruleset is designed to be accessible for runners who aren't interested in pathological optimization.
Re:Diablo hackers uncovered a speedrun scandal by immibis on Saturday February 15, 2025 @03:14PM (#43061886)
You can always play whatever you want - you just can't rank it on leaderboards if it doesn't follow the rules of that leaderboard.

TASes break every rule and are pretty cool.

Re:Diablo hackers uncovered a speedrun scandal by NobodyNada on Saturday February 15, 2025 @04:09PM (#43062350)
> TASes break every rule and are pretty cool.

There's still rules in TASing, just a different set of rules! Specifically, you have to submit a file containing a sequence of controller inputs that completes the game when played back in an accurate emulator (or a replay device connected to the controller port of an original console running the original game).

I co-authored the current fastest TAS of Super Metroid: https://youtu.be/m-Gt57ur7OA?t=2m30s We completed the game in under 2 minutes of gameplay by exploiting a race condition in the game's sprite animation system to start an exploit chain leading to full arbitrary code execution (full writeup: https://tasvideos.org/8214S ).

We used all kinds of tooling to create the run: a recording emulator with savestates, slowdown, and frame-by-frame input editing, scripting tools, debuggers, disassemblers, and memory viewers. We used memory editing tools during development to test different scenarios without having to set them up in-game first, but the actual run you submitted is completely "pure"; just an input file that beats the original, unmodified game.

And just like real-time speedruns, TASes are split into categories too. Here's the world record TAS with major glitches disallowed, by Sniq: https://youtu.be/Rr9gdQ-VkO4 Whereas our ACE TAS is more like a CTF challenge than anything else, Sniq's run is a masterpiece of gameplay that shows how far the game can be pushed beyond the current human limits. But again, it's still a true run of the game; the only difference is that you submit an input file instead of a realtime performance of the run.

Re:Diablo hackers uncovered a speedrun scandal by snibsnib on Saturday February 15, 2025 @06:54PM (#43063739)
A bit outdated now, but Saturn TAS made a very good TAS. It optimized door transitions to get the best in-game time. Unfortunately this was right as the rules changed between tracking in-game time and real-time, so the run never made it into the leaderboard. It made him stop making TASes for Super Metroid altogether. https://www.youtube.com/watch?v=y3THRhCgCO4
Re:Diablo hackers uncovered a speedrun scandal by AJenbo on Saturday February 15, 2025 @08:49PM (#43064534)
Here is the TAS we produced for Diablo (the original goal before we got side tracked by the issue found in what as supposed to be the baseline): https://www.youtube.com/watch?v=tiwoqd4arI0
Re:Diablo hackers uncovered a speedrun scandal by tetha on Sunday February 16, 2025 @07:36AM (#43067539)
> There's still rules in TASing, just a different set of rules! Specifically, you have to submit a file containing a sequence of controller inputs that completes the game when played back in an accurate emulator (or a replay device connected to the controller port of an original console running the original game).

Enter TASBot: https://www.youtube.com/watch?v=rha61Y3_Voc

It's a really cool project. In case "just" running a TAS on a real console through some wired up controller... Nintendo DS is also supported: https://www.youtube.com/watch?v=MHQCJX0RzS8

It's such a funny and rewarding escalation of the really impressive world of TAS.

Re:Diablo hackers uncovered a speedrun scandal by recursive on Saturday February 15, 2025 @03:59PM (#43062277)
I mean there can still be rules. Like using a hacked ROM or spliced video. Of course you wouldn't splice to get good performances, but different RNG states or something.
Re:Diablo hackers uncovered a speedrun scandal by immibis on Saturday February 15, 2025 @06:41PM (#43063631)
You can still do that, just don't submit it to a normal or TAS leaderboard. Put it on Youtube and say here's this cool thing I did. Even Doodong's Cavern is a meme.
Re:Diablo hackers uncovered a speedrun scandal by zerocrates on Saturday February 15, 2025 @04:26PM (#43062495)
Another good example is the Zelda Ocarina of Time any%, which progressed as new techniques allowed skipping more and more of the game, but which was still pretty broadly popular even when it was a warp to the final boss in under 20 minutes, is now way less popular now that they discovered a arbitrary code execution method that credits-warps in under 5 minutes.

This process had already been going on before, but you have this progressive sequence of the speedrun getting harder to play and harder to watch. You see some spliting out to restricted categories but even those aren't all that popular, but also something else that's recently become really common particularly in the communities for these games that have been optimized to death: randomizers. Shuffle around the positions of the necessary items, or the map itself, and you have something that's fresh for viewers and people who've played the game forever, an extra aspect of chance, usefulness of skills and familiarity with parts of the game that wouldn't otherwise be relevant... but it does split the community even more.

Re:Diablo hackers uncovered a speedrun scandal by jknoepfler on Sunday February 16, 2025 @05:34AM (#43067043)
And they are oh-so-contentious. If anyone wants to go down a fascinating rabbit hole, the recent conversation/debate around whether Bomb Torizo skip (a very difficult, borderline-human-impossible trick) should be allowed in Super Metroid runs was actually fascinating and ended up getting into deep research about inconsistencies between hardware polling rates vis-a-vis what runners could reasonable/humanely expect from one another... that's a thin gloss from a non-expert, but it's a great concrete example to work from.

(The debate was sparked by a runner proving that the trick was humanly possible, raising the spectre of someone actually pulling it off in a world record run, which would force all subsequent world record would-bes to replicate a trick that would kill 99.99% of runs, execution of which depending on the vagaries of how different controllers poll inputs in addition to nearly inhuman levels of sustained, frame-perfect inputs).

Re:Diablo hackers uncovered a speedrun scandal by rcxdude on Sunday February 16, 2025 @07:12AM (#43067431)
And there's generally a combination of principles involved, and different speedrunners and communities will prioritise different ones:

Fairness: generally speaking the competition wants to be fair. Part of this is just having a consistent set of enforced rules, but it also comes down to decisions about timing and allowing or disallowing certain tricks: if how fast part of a game is depends on how fast your hardware is, for example (whether by a trick working or simply lag), then most communities will try to eliminate this effect by banning hardware-dependent tricks or adjusting timing rules to remove loading times.

Competition: At the end of the day speedrunning is a competition around a game, so there's a general resistance to modifying the game itself, or e.g. use of external tools to assist the player beyond what was intended by the developer. But that will sometimes be compromised if there's some aspect of the game which badly affects other goals.

Fun: At the end of the day, speedruns are something that should be fun and rewarding to play and watch. So if the optimal run of a game becomes unfun to play or watch, then generally something will be done about it, whether by banning the trick which is unfun or by allowing something which alleviates the pain point (sometimes this even verges into health concerns: some pokemon games, for example, have very long runs with a huge amount of button-mashing to get through dialogue quickly, and RSI is a very real concern which means that turbo controllers which automate that are sometimes allowed). This can go as far as patching the game to remove particularly unfun or RNG-heavy parts, though it's rare.

And of course you can have different categories which will focus on different aspects of this, especially as different runners and watchers will have different ideas on what is fun: apart from what counts as 'beating the game' (any%, 100%, 'true ending', whatever), there's different levels of what other compromises are allowed. And it can get pretty contentious: from hardcore 'play the game as intended' purists to arbitrary-code-execution TASers, everyone has a different idea about what they enjoy in the competition, and different games can have strengths in each area of focus.