LinkedIn's negligence in failing to adequately secure user passwords

As most of you are aware, LinkedIn's site has apparently been hacked, and 6.5 million passwords of users were exposed (if you weren't aware, change your password); the likely attacker operated out of Russia. Take all I say with a grain of salt, as LinkedIn has recently tweeted "[o]ur team continues to investigate, but at this time, we're still unable to confirm that any security breach has occurred. Stay tuned here." But, I doubt that this is a false alarm, and for the uninitiated, let me translate that tweet in honest technology speak - "We've realized a breach occurred, we are panicking in a board room and attempting to spin this in the least damaging light possible."

In this day and age it is unsurprising that a large site has been owned by hackers; I think most would agree that this has become commonplace. But, it appears that corporations are failing to evolve based on the failures of their compromised brethren. While LinkedIn should be applauded (quietly) for their use of SHA-1 hashes to store passwords, they should then immediately be criticized for failing to also salt the passwords, or use a more cryptographically strong algorithm such as SHA-256, or SHA-512.

A quick explanation will make their negligence clear. Let us assume that the chance of disclosure of passwords is merely a function of exposure to the internet, multiplied by the traffic of (aka attacks on) the company, divided by the security measures in place to prevent data disclosure. The equation can be noted as EXP * TR / SEC = DISC(%). That equation is of course not scientific, but it helps to explain the current atmosphere of the internet. The variables EXP and TR are hard to control by any company that is out on the internet, and in fact, most companies interested in making a profit want those values to increase. The key to business viability, trust of the consumer (industry respect), and meeting the responsibility placed on you as a data steward is the company's SEC value. I would also argue that the more vital the service you are offering on the internet is, the more responsibility and obligation you have to increase your SEC value.

By using unsalted SHA-1 hashes, LinkedIn essentially conceded that the value of DISC would be enormous, and it did so by negligently failing to salt those passwords. I say negligently because it is commonly understood in the industry that use of a salt makes cracking password significantly harder. Take for example the NIST Enterprise Password Management Guide, which states:

The use of salts also makes cracking more difficult—for example, using 48-bit salting values effectively appends a 48-bit password hash to the original password hash, assuming that the attacker does not have access to the salting values and that the salting values are well-chosen. So a salted password might have the same effective length, and therefore be roughly as time-consuming to crack, as an unsalted password  that is several characters longer. Also, salts typically use the full range of possible values, unlike passwords that have limited character sets, so salts can strengthen the effective password complexity. Policies for password expiration, length, and complexity should take into account the use of salts.

The use of salts defeats, or at least slows down the use of "rainbow tables," which are tables of already calculated hashes of passwords. So, if I know that your site uses SHA-1 hashing, I take a wordlist of X number of words, and hash all of those into a database. Then, when a Russian hacker discloses all of your passwords, I merely correlate the values disclosed with the values in my table to discover passwords. I may not get all of the passwords, because the dictionary file originally used normally does not have every word or possible combination of letters, numbers, and symbols used by individuals, but I am guaranteed to get a large portion because users typically have bad passwords (or shall I say weak/predictable passwords).

The use of salting defeats rainbow tables because the hope is that the potential "cracker" of the passwords is clueless on the salt used to hash the passwords by the particular site, so a traditional rainbow table is useless. Thus the hacker would need to create a rainbow table for every possible iteration of the salt - an extremely time consuming task, and wholly not worth it. In all of these password cracking scenarios, there is a race condition going on. Specifically, that the number of entrants to the race decreases exponentially as the complexity and difficulty of the passwords that could be cracked increases (the value of SEC increases). As an internet company you need not outrun the bear behind you that is attempting to expose your security weaknesses, you merely need to be running faster than the others around you.

It is no argument for LinkedIn to assert that they could not have feasibly implemented a salt on their SHA-1 hashes, nor is it an argument for them to assert that others are using SHA-1 hashes. It is widely known that SHA-1 has been significantly weakened, and SHA-2 (256, 512) algorithms are better alternatives - the federal government urged federal agencies to stop using SHA-1 in March, 2006, and a competition has been running since 2007 to come up with SHA-3.

We must assume that password hashes are going to be disclosed because of the plethora of weaknesses in software currently implemented worldwide. What we shouldn't assume is that the stewards of our data are failing to exercise due diligence in protecting our information. The driver of an increase in the value of SEC is the real world accountability for preventable security failures.

Update: As expected, LinkedIn has confirmed the breach.