Time to crack DES? Is it a task suitable for a script kiddie yet?

Question

Already understanding that AES is the encryption method of choice, should existing code that uses DES be re-written if the likely threat is on the level of script kiddies? (e.g. pkzip passwords can be cracked with free utilities by non-computer professionals, so is DES like that?) A quick google search seems to imply that even deprecated DES still requires a super computer and large quantity of time--or have times changed?

In particular, this CAPTCHA library uses DES to encrypt the challenge string which is sent to the user in viewstate.

Answer 1

DES is broken so far as storing sensitive data, and so I would certainly not use it in anything new, and would replace it in anything used for long term storage of any information of interest (data that someone would have a profit for national security interest in stealing).

At the moment a DES message can be broken by brute force in a couple of days (or less) using under $100,000 worth of custom hardware.

But there are some key factors in that:

The hardware is custom - the chips used to quickly brute a DES key are not the general purpose processor you'd find in a PC. That being said there is probably room today for using a cluster of Playstation 3s or current generation graphics cards with a GPGPU to crack a DES message in a reasonable amount of time, perhaps bringing down the cost to maybe $15,000.

The other factor is time - a DES message can be cracked in a day, but if your CAPTCHA library has a timestamp that specifies a 30 minute timeout for any given CAPTCHA response, it would still be effective (you could scale up your hardware, but then you're talking millions).

Overall I'd say that for non-long term storage, DES is still secure against "script kiddies".

Answer 2

no, DES cracking is not suitable for scriptkiddies and won't probaly be in the near forseeable future.

it requires such enormous processing power, we're talking about a load of FPGA processors.

for example the COPACOBANA in the CHES 2006 secret key challenge took 21 hours, 26 mins, 29 secs using 108 of it's 128 processors, at a troughput of 43.1852 billion keys per second, and found the key after searching trough 4.73507% of the keyspace

now, if we look at moores law we see, that if we currently build a similar machine, it'll currently take 1/4th of the time for the same amount of money, or 1/4th of the money for the same amount of time.

Answer 3

DES is broken by the standards of the crypto community; but the time required to break it is generally large enough that it would be 'safe' to use for this kind of application. On one assumption: the DES key changes from session to session. If the key doesn't change, then it is open to attack by a very very dedicated individual. Now, the question is, is your website subjected to people that will spend 10+ days cracking DES, rather then applying lessons learned by the rest of the Spam Industry in the way of Image Recognition.

Answer 4

DES is probably still good enough for most use cases. But the point is, there is normally reason to use an algorithm (or in this case rather: a key strength) which is known to be rather weak. Wikipedia points out that even with special hardware around 9 Days are needed for an exhaustive key search. I don't think the script kiddies are likely to spend that many CPU time (even if they have a botnet) only to crack a captcha. (Actually, cracking captchas is normally A LOT easier with sufficient intelligent picture recognition...)