Anyway, the thing I want are hashes. Which ones should be used? My suggestion: SHA-256.
I think MD5 would be good enough:
* It is the most common supported hash
* It is fast to calculate
* It is short as hash (32 hex digits, almost anything else is longer)
* Resistance against intentional tampering is not needed (only against random errors).
* Intentional tampering would need second preimage attack (seems infeasible to attack MD5 that way).
I think MD5 would be good enough:
* It is the most common supported hash
* It is fast to calculate
* It is short as hash (32 hex digits, almost anything else is longer)
* Resistance against intentional tampering is not needed (only against random errors).
* Intentional tampering would need second preimage attack (seems infeasible to attack MD5 that way).
It's true that we needn't to worry about malicious attempts to modify the files. But then again why not use a more modern hash? MD5 is broken. MD5 might be more common, but by the same argument we should choose plain CRC which clearly isn't desirable. About every commonly used hash is fast to calculate, firstly they were designed that way, and secondly the calculation is I/O bound. And the hash string isn't that long really.
The most general attack against MD5 I am aware of is choosen-prefixes collision. Not quite a second preimage attack, but is more powerful than ordinary collision attack.
This does not allow to corrupt files in transit without MD5 sum changing unless the original file is specifically generated to make such replacement possible.
Doing such corruption without requiring the original file to be carefully crafted would need a second preimage attack (AFAIK, currently infeasible against MD5).
Geez, I'm fine with any hash, we just need to adopt one. I still don't like that you appeal to popularity in favour of MD5. For myself, coreutils has provided sha256sum since forever. Even though this isn't about cryptography, why choose an obsolete and broken hash if others are available? I can's see any real reason.
The most general attack against MD5 I am aware of is choosen-prefixes collision. Not quite a second preimage attack, but is more powerful than ordinary collision attack.
What I meant with my question was "how is it broken such that it makes it a bad choice in this particular case". As was established earlier, the possibility of deliberate tampering is not the issue here.
How do you think about "Atlas Movies"? Will it be possible to publish them besides the SD and HD encodes?
I created an Atlas Video for [1349] NES Super Mario Bros. "warpless" by HappyLee in 18:41.70, see here http://www.youtube.com/watch?v=i8u4SyclFNs
I don't expect this particular video to be good enough for publication here. There are still many things missing for a "perfect" video -- movie dependant enemy placement on the maps, "cool" intro, object updates on the map (broken blocks, latest enemy position etc). The necessary tools for that are being developed right now. Some of them might be even useful for TASing.
Would "perfect" Atlas Videos get a chance for publication?
