[bitcoin-dev] Time to worry about 80-bit collision attacks or not?

[Jorge Timón]

On Fri, Jan 8, 2016 at 4:50 PM, Gavin Andresen via bitcoin-dev
<bitcoin-dev at lists.linuxfoundation.org> wrote:
> And to fend off the messag that I bet somebody is composing right now:
>
> Yes, I know about a "security first" mindset.  But as I said earlier in the
> thread, there is a tradeoff here between crypto strength and code
> complexity, and "the strength of the crypto is all that matters" is NOT
> security first.

If the crypto code is properly encapsulated, the code complexity costs
of choosing one hashing function over another should be non-existent.
You made the space argument which is valid, but in my opinion code
complexity shouldn't be a valid concern in this discussion.

As a maybe uninteresting anecdote, I proposed the asset IDs in
https://github.com/ElementsProject/elements/tree/alpha-0.10-multi-asset
to do the same ```ripemd160 . sha256``` choice that Mark Friedenbach
had proposed and I had approved for
https://github.com/jtimon/freimarkets/blob/master/doc/freimarkets_specs.org#asset-tags
. More humble than me, he admitted he had made a design mistake much
earlier than me, who (maybe paradoxically) probably had less knowledge
for making crypto choices at the low level. In the end I was convinced
with examples I failed to write down for documentation and can't
remember.

That's not to say I have anything to say in this debate other than
code complexity (which I do feel qualified to talk about) shouldn't be
a concern in this debate. Just want to focus the discussion on what it
should be: security vs space tradeoff.
Since I am admittedly in doubt, I tend to prefer to play safe, but
neither my feelings nor my anecdote are logical arguments and should,
therefore, be ignored for any conclusions in the ```ripemd160 .
sha256``` vs sha256d debate. Just like you non-sequitor "sha256d will
lead to more code complexity", if anything, sha256d should be simpler
than ```ripemd160 . sha256``` (but not simpler enough that it matters
much).