If I have a mobile number for a PO, is it okay to send a quick "Thinking of you, thanks for your service" message? Or does that risk getting them in trouble somehow? Not sure if it's a work-owned device or not.

Kathleen Lonsdale's notebook used while determining the crystal structure is also rather wonderful - such elegant handwritten calculations.

lol hope this lands and enough of your students lie at the intersection of critical AI thinkers and historians of the far side’s most inscrutable joke!

I think that might've happened. For what it's worth, in the old days POs used to tell me that NOSIs were really only used for keeping track of interest in various topics (ie, they didn't increase your chances). I assume they have more urgent issues to deal with now

This is my read too

Congrats Eartha!!!! Getting the bioRxiv out is the best milestone. The results look great!

Not to say that i think the current system is working well at all. The system I've imagined: Censor journal names from grant applications, remove all requirements for peer review, but allow authors to attach optional & non-anonymous peer reviews to their preprints.

While very creative, this proposal would nearly require publishing in a gov-run journal, and I think right now is evidence of the danger in that. It wasn't that long ago (~2015) that people seriously pitched the idea of Twitter replacing journals, and I think we all see the danger in that now too.

Sounds like they could have sharpened up their operational def, but I don't see this as evidence that ODs aren't desirable. The main alternative - non-operational definitions - can lead to imputing all kinds of cognitive constructs willy nilly, like a mouse model of a case of the Mondays or whatever

With only 1 month of data, you have fewer pairs to correlate at +/- 20 days. This can show up as a decrease unless you normalize, which then produces high variance. If you repeat many times, does it reliably decrease, or do the error bars get wild at the long latencies?