Re:Science Is Open Software
by D-Machine on Saturday September 19, 2026 @04:02AM
(#49764403)
> it's now widely expected that a high impact pub will make all data and code available for review, and then publicly available upon manuscript publication
I am also in academia and don't think this is true at all for data, not even remotely, unless by "publicly available" you mean "available to most academics or academic labs after an often tedious and slow approval process requiring an academic email and various signed agreements". Maybe what you are saying is true in some very specific domains (e.g. machine learning research), but in general what you are saying here is IMO wildly out of touch with present realities in the vast majority of fields, but especially those involving human subjects.
> While I have some sympathy with a lot of your bitterness, this statement is insulting silliness that a quick look at the list of Nobel Prizes in physiology and medicine would prove wrong. Almost all major breakthroughs in the applied sphere stem from decades of basic research that happened just because it interested someone.
Nobel Prizes are so rare they don't speak at all to the generalizations I am making here. Also, much medical academic research is successful because it is in fact ultimately industry-funded. It is of course though highly dependent on the academic subfield, for sure, and I was painting with a broad brush.
If I had to narrow things, STEM academic research isn't so bad, so long as we exclude social science from STEM. Much social science research needs to be defunded ASAP. And I'm not claiming industry research doesn't also have warped incentives. But, on balance, I'd wager outside of pure math/physics and certain more algorithmic/pure domains in comp sci, the smartest people today are going to choose (and be found in) industry, not academia.
I am also in academia and don't think this is true at all for data, not even remotely, unless by "publicly available" you mean "available to most academics or academic labs after an often tedious and slow approval process requiring an academic email and various signed agreements". Maybe what you are saying is true in some very specific domains (e.g. machine learning research), but in general what you are saying here is IMO wildly out of touch with present realities in the vast majority of fields, but especially those involving human subjects.
> While I have some sympathy with a lot of your bitterness, this statement is insulting silliness that a quick look at the list of Nobel Prizes in physiology and medicine would prove wrong. Almost all major breakthroughs in the applied sphere stem from decades of basic research that happened just because it interested someone.
Nobel Prizes are so rare they don't speak at all to the generalizations I am making here. Also, much medical academic research is successful because it is in fact ultimately industry-funded. It is of course though highly dependent on the academic subfield, for sure, and I was painting with a broad brush.
If I had to narrow things, STEM academic research isn't so bad, so long as we exclude social science from STEM. Much social science research needs to be defunded ASAP. And I'm not claiming industry research doesn't also have warped incentives. But, on balance, I'd wager outside of pure math/physics and certain more algorithmic/pure domains in comp sci, the smartest people today are going to choose (and be found in) industry, not academia.