Score:0

Ensure that a speedrun (or video recording) was done in one continuous hop (and not spliced together from many parts)

in flag

In speedrunning video games, one records a game being played and beaten in one continuous attempt. However, what can be done to cheat is to do multiple attempts, and splice together clips of the best segments to make one fast speedrun that wasn't done in a single continuous hop. This splicing isn't hard, as e.g. loading screens always look the same, so you can swap the video at those points without being noticed.

My question is: is there some way people can verify that their video was recorded in a single hop?

The reason I believe there's hope in solving the problem is: one mario 64 speedrun was caught splicing because the ambient background noise in the audio had a discontinuity at the loading screens (indicating the video was swapped there). So I imagine that some method of artificially injecting noise might be the key. The answer might also involve using live generated random values, to ensure the overall timing throughout the speedrun is consistent, as opposed to different parts being recorded at separate times.

I understand that there's lots of ways to cheat a speedrun, but I'd like some way to address this particular problem of splicing a video that should be done in 1 continuous shot.

Note that because it's a speedrun, you can't ask the player to perform a random action to verify something is done live, as performing that action would waste time, which one doesn't want to do during a speedrun.

fgrieu avatar
ng flag
I don't see how crypto can help without some trusted hardware. That would make the question off-topic.
in flag
The question is whether it can be done, and I suspect it can through a cryptographic scheme. My thought is a solution would look like injecting some cryptographic signal into the video that would verify the time at which each part of the video was taken. If necessary, a trusted-third party could be part of the scheme (to generate and verify time stamps).
fgrieu avatar
ng flag
"injecting some cryptographic signal" requires some trusted hardware so that it can't be done after the captures are made and sliced. Short of having what runs the video game itself within that trusted hardware, I don't think anything will be secure. On the other hand a trusted videogame that outputs timestamped and signed video captures is feasible.
Maarten Bodewes avatar
in flag
Could you edit this question so that you ask for a cryptographic resolution? If not it might be more at home at [gamedev.se].
Score:1
ng flag

A cryptographic solution would, by definition, have a public design, with only a key possibly secret. If it was an add-on to an external game console or PC program, e.g. timestamping and signing frames made externally and fed to an input, it would be vulnerable to splicing at it's input.

The only thing I can think of that would work robustly is a trusted box, setup by a trusted authority, that both runs the game and signs the outcome, e.g. timestamps and signs all frames in the capture. The capture would then become publicly verifiable by anyone trusting the authority, it's ability to set up secure devices, and the integrity of it's public key.

This compares to the practice of having a referee setting up the game and attesting a record; only automated and working even when the referee is absent.

mangohost

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.