search

codeisscience / code-is-science

Scientific code needs to be open source and peer reviewed
http://www.codeisscience.com
MIT License
136 stars 32 forks source link

Convert MozFest notes to manifesto draft. #17

Closed yochannah closed 6 years ago

yochannah commented 7 years ago

Convert MozFest notes to manifesto draft.

I'd like community input on this; maybe another workshop session of some sort?

yochannah commented 7 years ago

Notes from SSI discussion: we need to include the importance of software citation here!

yochannah commented 7 years ago

Software isn't measured only by papers; also by code quality

leotrs commented 7 years ago

I like the idea of a manifesto. There are a good few gems in Best Practices for Scientific Computing for example,

  1. Write programs for people, not computers
  2. don't repeat yourself
yochannah commented 6 years ago

Update: at cw18 we worked on a manifesto draft together and have it open to public consultation until ~June 1. The draft text is here: https://github.com/codeisscience/manifesto/blob/master/manifesto.md

Please feel free to comment, pr, raise issues, etc!

JJ commented 6 years ago

OK, let me pick a bone that has troubled me through time here. And full disclosure before I go ahead: I don't like proprietary software in science. Main point is that using proprietary software makes science not reproducible, since you might not be able to use that particular proprietary software (or platform) needed. So my (maybe not so) small proposal would involve slightly changing phrasing so that we talk about "runnable" code to rule out those platforms (if that's what we want, of course. I know I do, but...)

yochannah commented 6 years ago

@JJ In essence, I agree with you that all scientific code should be open source and that proprietary software shouldn't exist in science at all (at least, I've yet to hear a good argument for highly expensive / closed software yet!) BUT I also want something realistic - and I think more people are likely to adopt the manifesto if it's accepting. so, to my mine:

1) BEST CASE: Paper is published with linked open source code w/ hash of the relevant commit, fixed dependencies, well documented, etc. 2) Not as good but still reasonable: Peer reviewed code before publication, even if the code itself isn't open. 3) Not proper science: accepting papers without actually looking at the code that produced the science.

yochannah commented 6 years ago

marking this issue as closed since the manifesto is currently in draft format.