Date: Thu, 20 Sep 2007 20:21:41 +0200 From: Michal Luczaj <regenrecht@...pl> To: john-users@...ts.openwall.com Subject: Re: Markov filter password generation Simon Marechal wrote: >> I just "released" an experimental support for Markov filter password >> generation. (...) >> I need some feedback on this generator effectiveness, especially with >> non french passwords. If it turns out it works well, I will release a >> clean patch for JtR. Hello there, Congratulations, Simon. I gave it a short test and I must say I was very positively surprised. I've ran your patched John twice (about 3,5h each run) against a list of 946 unique raw-MD5 hashes (of mixed Polish, English and few Japanese/romaji passwords) with the fallowing results: incremental=all, guesses: 365 incremental=alnum, guesses: 398 markov=266:0:0:8, guesses: 469 (At first I've ran only incremental=all, but realized that would be unfair in comparison with this Markov level.) It looks like your Markov filter, just as you wrote before, "is more effective /during a certain amount of time/". And, of course, incremental mode found some passwords that Markov didn't. Strengths and weaknesses of both modes are quite obvious, but it would be really good to have such an addition to John, so I'll be waiting for a clean patch. Great job! Thanks, michal P.S. Just for fun, with john.pot already filled after those incremental/markov passes, I've run markov=250:0:0:16 - 17 new guesses, ~4h. And then markov=257:0:0:30 - 2 more (9 chars long) guesses, ~8h. And that's really really nice. But I've noticed that 250:0:0:16 found some additional short (6-7 chars long) passwords that 266:0:0:8 didn't. Is it correct behavior? -- To unsubscribe, e-mail john-users-unsubscribe@...ts.openwall.com and reply to the automated confirmation request that will be sent to you.
Powered by blists - more mailing lists
Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.