4 Comments
Sep 30, 2022Liked by Alexandros Marinos

When I was still in academia some 10 years ago, and when asked to review papers, I would unwaveringly not recommend publication, for any work that did not intend to publish both code and data that would reproduce all claims and figures; preferably in a single click, not some hastily zipped up folder of incomprehensible spaghetti. I thought the time was kinda right for that; even though boomer editors of journals charging like $40 per view would think that the complexity of hosting a few 100mb was somehow an insurmountable problem. There used to be some genuinely hard problems, like how to get others to actually compile your legacy fortran dependencies. But then docker came along and distributing those kind of dependencies also was sort of a solved problem, if you knew what you were doing.

Id like to think I put my drop in the bucket. But seeing posts like this, in the year 2022, talking about 'the trial of the year', perhaps 'a drop in the desert' would be less presumptuous.

Expand full comment

“our effect size does not change importantly (relative risk [RR] 0·74 [95% CI 0·56–0·98].”

#GoldStandard

Expand full comment

They are bleeding your Twitter followers hard now. Just FYI. :(

Expand full comment