Hacker News new | past | comments | ask | show | jobs | submit login

leaking unsalted md5 passwords == leaking passwords



Not if it's my password. I use ~100 bits of entropy.


Hence "Strong passwords that need to be memorized" in OP's comment. Or else your memory is way better than mine (or I care way less, or probably both).


Whether they need to be memorized or not does not make the statement "it will be leaked eventually anyway" more true.

I use a password database so I don't memorize most of my passwords.


Well, it does, because memory puts some limitations on length and complexity...


It is possible to memorize a 100 bit password. I once had a 1000 word poem memorized, and could write it down flawlessly from memory.

I agree that it's not worth memorizing, you should instead use a password database. But I still maintain my original point that there's no reason to assume that your password will be leaked eventually if you use a strong password.


Could the attacker find an easier to find string that matches the same md5 hash?


The current best attack wrt matching an existing hash brings MD5's 128 bits of security down to 123. So no, that's not going to happen.


MD5 is terrible for human passwords because it's fast. But md5 is not actually broken for password storage purposes. If you use a long random password, md5 is enough.


Yes, if you add a (long - at least 32 bit) salt and something like at least 10^9 rounds of MD5 then, yeah, it's probably ok


No. I mean single unsalted MD5. You will not crack a 20-random-char password. You cannot process 2^120 guesses, and MD5 is not broken for this use.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: