This point has been hammered home in a study of some 32 million passwords that were posted on the Internet after a hacker obtained them from social entertainment site RockYou last year.
In a report released on Thursday, Imperva, a security firm, analyzed the strength of the passwords people used and found that the frequent choice of short, simple passwords almost guarantees the success of brute force password attacks.
A brute force attack involves automated password guessing, using a dictionary or set of common passwords.
According to the report, "the combination of poor passwords and automated attacks means that in just 110 attempts, a hacker will typically gain access to one new account on every second or a mere 17 minutes to break into 1000 accounts."
The report reveals that 50% of users rely on slang words, dictionary words, or common arrangements of numbers and letters, like "qwerty," for their passwords.
Among users of RockYou, the most common password was "123456."
Sadly, this isn't a new problem. Previous password studies, using far smaller data sets, have shown similar findings. Imperva's CTO Amichai Shulman observes that a 1990 Unix password study reveals the same password selection problems.
A recent review of Hotmail passwords exposed in a breach also showed that "123456" is the most common password. Even though "123456" occurred only 64 times out of 10,000 passwords, that suggests that a brute force attacker could compromise one account per 157 attacked using a dictionary with only a single entry.
Jon Brody, VP at TriCipher, another security vendor, confirms that this isn't a new problem. He puts part of the blame on technology innovators for not recognizing that password policies are doomed to fail if they go against human nature. That is to say, forcing people to change their passwords every month will force them to choose weak passwords every month because that's what they can remember. Brody argues that technology companies need to create security systems that take real world behavior into account.