Cosmology seminar: Dag Sverre Seljebotn

Hashdist - yet another desperate attempt at fixing scientific software distribution

How many researcher-hours are wasted every year editing makefiles and
trying to get software to build on a new cluster? Wouldn't it be nice
if somebody just solved the problem?

We (me and some others you don't know) haven't done so yet, but we
believe we have some new ideas and that we will make some real headway
on this problem. We've only burned 1/3 of the funding so far and
nothing is ready for consumption, but I'll describe the ideas and
our path forwards.

I'll probably touch upon things like:

 - Cryptographic hash functions
 - Software and reproducible research
 - Why the others are wrong and we are right
 - How our ideas build on previous ideas (in software design), and
   couldn't have happened 10 years ago.

Note: This is not part of my PhD; I've taken 2 months leave from my
PhD to work on Hashdist with funding from the US Army Engineer
Research and Development Center. All our code is open source.