I've spent some time recently arguing the narrow point that "journal publishes only preregistered studies" is NOT a good measure of transparency. So let's also argue a broader point: that there is no "One true way to science" and prereg is important only for some ways 1/N

Preregistration seems to get at least part of its strength from the idea that making predictions - especially risky ones - is the strongest way we can confirm a theory. This is sensible, but not necessarily the whole picture. 2/N

One can make a good claim that (at least in natural sciences) explaining previously known facts is (at least sometimes) stronger support for a theory than predicting new facts. The typical example is explaining the precession of Mercury by relativity. 3/N

Why? Because a) the fact was well established and beyond dispute and b) other theories tried and failed to explain the fact. Contrast with the novel prediction of light bending by sun which could be explained in Newtoninan terms and was seen in only one experiment. 4/N

Requiring prereg would mean this (IMHO) strong evidence in favor of relativity would be inadmissible. I don't doubt similar examples from other fields could be found, although I can imagine this case being rare in psychology. My source: https://www.journals.uchicago.edu/doi/abs/10.1086/psaprocbienmeetp.1994.2.192926 5/N

Closer to where I work, bioinformatics as a science has a big engineering component - doing is an important goal. Should we discount the achievements of the Human genome project (or the recent work on assembling wheat genome) because it was not preregistered? 6/N

Similarly I don't think people preregistering "We want to resolve the crystallographic structure of hemoglobin" should increase our trust in the structure they report. 7/N

You can find similar important counterexamples for many sensible scientific practices - just that something is sensible some/most of the time doesn't mean it applies always. And that's my beef with TOP or the reform more broadly: lack of acknowledgement for this plurality. 8/N

And lack of humility in lobbying for unconditional adoption of a practice without much more than anecdotal evidence. To quote "methodological reforms should be held to standards that are at least as rigorous as those we expect of empirical scientist" https://twitter.com/zerdeve/status/1255168642938880001?s=20 9/N

Great case for plurality in science (which does NOT mean "anything goes") was made by Hasok Chang - see my summary thread: https://twitter.com/modrak_m/status/1304421375654928384 10/N

Much of my thought on this was formed by @zerdeve and @djnavarro (thanks) which also motivated further reading. I am no expert on phil. of sci. - many better thinkers than me are mentioned here: https://twitter.com/zerdeve/status/1234906668036608000 But hey, I have confidence and attention :-D 11/N

I fully agree that prereg is a good tool in a scientists' toolbox. It probably should be used more than it is. But I find the jump from "this tool is useful" to "everybody has to use this tool to count as good/transparent/... research" dangerous, misleading and full of hubris.

Also remember that the the original point by @zerdeve was different than mine and maybe much more worthy of addressing. https://twitter.com/zerdeve/status/1326248784175620096 13/13