Lifetimes of cryptographic hash functions | Hacker News Reader

Lifetimes of cryptographic hash functions

valerieaurora.org

54 pointsluu15 comments

This has been discussed before, about 9 months ago: https://news.ycombinator.com/item?id=6123720

And this page was largely written in 2007, although it's been updated a few times since.

I'll also note that Val's "Compare by hash considered harmful" is not a conclusion which is widely shared. One can also worry about two people randomly (assuming quality random number generators) choosing the same RSA public key. Or we can worry about all of the oxygen molecules in the room simultaneously randomly deciding to concentrate into a corner of the room such that you die of asphyxiation. Sure, these things are _possible_; but whether you should therefore say that "life on earth is clearly broken because we are only probablistically correct and not provably correct" might be considered by some to be a somewhat insane conclusion.

Don't miss the table at the bottom: "Reactions to stages in the life cycle of cryptographic hash functions". Reach each row. Be entertained.

I'm surprised SHA2 is labelled as "weakened" since there's no real reason to believe so.

I agree. There was some concern around 2005-2009 about whether the attacks which broke earlier hashes could be extended to SHA256, but at this point it looks like those fears aren't coming to pass.

A very naive question by a non crypto expert: where is scrypt and bcrypt? is that because they are simply iterating another hash function or something? this is why I don't see PBKDF?

They are different things. I think the proper term for scrypt, bcrypt, etc. is a key derivation function. They take a password, add a salt, and produce a key that can be used for authentication.

http://en.wikipedia.org/wiki/Key_derivation_function

The hash functions in the OP are cryptographic hash functions.

http://en.wikipedia.org/wiki/Cryptographic_hash_function

The 'cryptographic' part of the name means that it is hard to find two files that produce a collision (hash to the same value).

I believe that key derivation functions will include cryptographic hash functions.

Short answer: Those are key derivation functions, and people who call them "hash functions" are abusing terminology in a dangerous way.

This was asked in the previous discussion of this item. Here is the link to the question and discussion:

https://news.ycombinator.com/item?id=6124825

If you want to see more of the discussion, follow the "parent" links from there.

Please notice Valerie, she is absolutely brilliant.

http://valerieaurora.org/

This is my personal favorite: http://valerieaurora.org/tcpip.html

What about Whirlpool?

Whirlpool is unbroken. It uses the same design principles as AES, which both makes it a little more complicated than it could be, but also demonstrably secure against the attacks that plagued the MD4 family of hashes.

There are lots and lots of hash functions lacking widespread adoption or standardization. I don't remember the exact timeframe, but djb once told me he came up with CubeHash in some number of days.

Whirlpool though is standardized (as ISO/IEC 10118-3) and it is one of the three available hash functions (along with SHA-512 and RIPEMD-160) for use in TrueCrypt.

The "Breakout" chart would be awesome if the color/stage change cell linked to the relevant result. Does anyone know o a resource like this?

According to the chart, RIPEMD-160 is deprecated. Does anyone know of a citation?

Reply on news.ycombinator.com