The data that was released were password hashes, or cryptographic representations of passwords churned through an algorithm called SHA-1. For example, if a person’s password is “Rover” the SHA-1 hash would be “ac54ed2d6c6c938bb66c63c5d0282e9332eed72c.”
Converting those hashes into their original passwords is possible using decoding tools and powerful graphics processors. But the longer and more complicated the password — using sprinklings of capital letters, numbers and symbols — the longer and harder it is to crack.
What’s interesting about the LinkedIn hashes is the trouble experts are having at converting the hashes to their original password. Of the 6.1 million hashes, some 3.5 million appeared to have already been cracked since those hashes have “00000” at the beginning.
That leaves 2.6 million uncracked hashes, which Young and some colleagues have been working to decode. And they’re not having a whole lot of luck. They need, in short, more words.
One of the methods for revealing passwords is comparing the hashes with ones that have already been decoded. They can do that by using lists of hashes and their corresponding passwords which have come from other data breaches, such as those affecting Stratfor Global Intelligence and MilitarySingles.com.
Young said it appears the hackers have already used those lists against the LinkedIn hashes. When he ran a check of the LinkedIn hashes against those decoded from the Stratfor and MilitarySingles.com breaches, he only got about 5000 matches. The “easy pickings,” Young said, are gone.
“These are definitely the tougher passwords of the bunch,” said Young, an adjunct professor of information security at Utah Valley University. “They’re pretty complex.”
In order to crack them, Young and his team need more words and more word combinations for so-called brute-force attempts. They’ve turned to some of the world’s most famous books.
Young wrote a program that draws passphrase strings from books such as Tale of Two Cities, War and Peace, The Call of the Wild and The Land of Oz. The program takes words from those books and creates phrases and concatenations such as “lionsandtigersandbears” and “ihavebeenchangedforgood.” Both generated hits in the LinkedIn hashes.
For the passage “Tip was made to carry wood from the forest” — from The Land of Oz — Young’s program will try the hash for “Tip,” then “Tipwas,” then “Tipwasmade” and “Tipwasmadeto” and on. The program could also be configured to add numbers, symbols in further attempts to match a hash.
One of the password programs Young’s team is using, John the Ripper, is trying some 700 to 800 million word combinations a second. Other tools Young is using are trying as many as 5 billion combinations per second, an astounding figure. So far, they’ve only decoded around 50,000 LinkedIn hashes.
Young’s next idea is try to use combinations of financial and business-related words. Because LinkedIn has many business users, it may be more likely that those types of phrases would be used.
In another innovation, Young’s colleague Joshua Dustin recently wrote a program that mines Twitter for more word combinations. The script connects to Twitter and pulls 500 messages related to a desired term, then delivers all of the words in a list.
Dustin wrote on his blog that he used the method against hashes generated by the MD5 algorithm from the MilitarySingles.com breach. Of the 4,400 unique words pulled from Twitter, 1978 passwords were revealed using John the Ripper cracking program, he wrote.
“This is a very small example of what can be done to generate more relevant password lists using twitter/websites/social media to supply you with the related words,” he wrote.
The difficulty in cracking some of the LinkedIn passwords may be of some relief to users, even though LinkedIn said it doesn’t believe that e-mail addresses used to login into accounts with passwords have been released.
“These are tough, tough passwords,” Young said.
Send news tips and comments to firstname.lastname@example.org.