You’re best off forgetting your grammar lessons when it comes to creating passphrases, according to new research out of Carnegie Mellon University and MIT.
The researchers say that using grammar – good or bad – can clue in hackers about the words in a multi-word password. And they’ve built an algorithm as a proof-of-concept to show it (The team, led by software engineering Ph.D. student Ashwini Rao of CMU’s Institute for Software Research, will present its research at the Association for Computing Machinery’s Conference on Data and Application Security and Privacy on Feb. 20 in San Antonio.).
The team tested its grammar-aware password cracking algorithm against 1,434 passwords containing 16 or more characters, and cracked 10% of the dataset via the algorithm.
“We should not blindly rely on the number of words or characters in a password as a measure of its security,” Rao said, in a statement.
The researchers say that while a password based on a phrase or short sentence can be easier for a user to remember, it also makes it simpler to crack because grammatical rules narrow word choices and structures (in other words, a passphrase with pronoun-verb-adjective-noun would be easier to crack than one made up of noun-verb-adjective).
The researchers found that “Hammered asinine requirements,” for instance, is harder to crack than even the longer and seemingly clever “Th3r3 can only b3 #1!”
Passwords in general have come under increasing fire by security pros, as some of the highest profile breaches (LinkedIn, Nvidia) have been the result of password compromises or resulted in passwords (including encrypted ones) being made public.
Google’s security team is looking into ways to avoid passwords altogether for logging into websites.