best64.rule contest
rockyou for sure, but linkedin was uniq'd so it's no good for this purpose.
Yeah uniq the hashlist wasn't good. Btw, there is also "10-million-combos.txt" from Mark Burnett. I've replaced many of my "rockyou.txt" tasks with this list already.

+ real people passwords
+ made for research
+ nearly same size as rockyou.txt
+ from different sites not just one
+ not gaming sites, so maybe more serious passwords
- not a leak, cracked passwords
Have you taken into account any of the issues cited in "A list of flaws in the data set_10millionpasswords" at
No that's actually new to me, thanks! I've gone through the list and pulled out what could be a problem if we would use it for a contest:

- used cleanup scripts (don't this to your wordlists unless you really know what you do)
- email addresses
- default passwords tend to skew lists
- weighted criteria
- hashes in wordlist
I'm not sure if uniq'ed wordlists pose a problem for this contest (linkedin). I guess the difference between total number of cracked passwords vs unique cracked passwords is relatively small because commonly used passwords usually follow weak rules (or none at all). Contrary, non-uniq'd lists might push up random spam bot passwords.
The list from the previous contest was unique on purpose.

There are two reasons why you find duplicates in any dump: Simple passwords, and site-specific passwords. Neither of which are useful to build a stronger ruleset.
I strongly disagree, James. Duplicates are essential for sorting rules by probability. Just as you'd never generate an hcstat file with a wordlist that's been uniq'd. By removing duplicates you are skewing the stats.
There is advantage and disadvantage in both variants.

It would be nice to add more people from the password cracking scene (like team-insidepro and jtr-users) for this contest, as everyone would benefit from it.

@mastercracker & @magnum You guys interessted?
I would like to participate but don't really have the time. I will give it a shot if I have some spare time when you run the contest. If you want to make it a bit more challenging, you can make the contest about the best wordlist + rule combination. The winner being the one who will crack the most passwords using a maximum of x words and y rules. X could be around 0.5 to 3 million and Y around 50 to 500.
This opens another question. Are the plaintext passwords for the hashes known or not.