Ugh, yes. Crypto experts, please show me only one example collision following the common combined seed scheme in dice games. And even if you find one collision, the verification for all other bets in the same series (same client/server seed but incrementing nonce) will fail. I'd say it's absolutely impossible to find a collision for two bets in a series let alone the whole series.
Sha1 and even md5 are absolutely sufficient for this purpose.
Sha1 and even md5 are absolutely sufficient for this purpose.
You're the most sort of dangerous: arrogant and clueless. If you don't lose peoples money due to malice, I'm sure you'll manage to do so with incompetence. I was going to demonstrate a practical md5 hash collision for you on your site, but your misconceptions (regarding the series) are so entrenched that your site isn't even close to provably fair, as you don't even give a guarantee of you changing the server seed. You only need to make 10^4 sha1's per bet in the series to revert it. Which means practically you can pick the outcome for the first 10 or so bets after someone changes their client seed.
Anyway, you might want to gloss over the wikipedia article on md5. And if you're still not convinced, download hashclash or something. And when you're done, check out how every other dice site proves they don't change the server-seed, and follow suite -- you obviously don't have the skills or understanding to deviate from the norm.
Wow, surprised to read this...
