News for hackers. Stuff that matters.
Re:Psychology’s Replication Crisis Is Real by MaxBarraclough on Saturday December 01, 2018 @07:40PM (#18580023)
> 6% included code for the papers' algorithms

This is completely ridiculous.

In pure mathematics, everything you need is right there in the paper. If instead you're in the physical sciences, you obviously can't include your lab in the paper. For software, it's perfectly possible to include the lab in the paper, so to speak, and there's no excuse for depriving the rest of us of access.

Not interested in releasing the source? Ok, just don't brand your project as 'computer science'.

Including full source-code (and ideally data-sets, unless there's good reason this can't be done) should be a basic requirement for serious publications. It's disappointing that this isn't (yet?) the norm.

</rant>

Re:Psychology’s Replication Crisis Is Real by ianai on Saturday December 01, 2018 @09:49PM (#18580506)
Agreed, at least pseudo-code can be included.
Re:Psychology’s Replication Crisis Is Real by MaxBarraclough on Sunday December 02, 2018 @06:58AM (#18581974)
I have to strongly disagree here. The source code itself - and not some proxy for it - must be made available.

Pseudocode is just another description of the approach the authors took. It doesn't allow another researcher to carefully recreate the exact experiment that the authors ran.

Software is famously hard to get right. This can be exacerbated by certain research problems where you don't know the result to expect (modelling, say). If you're going to publish results, the source you used should be available for inspection, for similar reasons to why mathematicians have to publish the proof, not just the conclusion plus a promise.

The upsides here are many: protecting science against software bugs, easier replication of experiments, better protection against academic fraud, and helping other researchers extend the work.

(I realise this last point may be at odds with toxic academic careerism. All the more reason for the publication process to insist on it.)

Re:Psychology’s Replication Crisis Is Real by tnecniv on Sunday December 02, 2018 @03:07PM (#18584247)
Most papers include pseudo-code when it helps clarify things. This tends to be a very high-level description of the algorithm though, and is not really the same as including actual source.
Re:Psychology’s Replication Crisis Is Real by justinpombrio on Saturday December 01, 2018 @10:43PM (#18580656)