Mikhail Kats

Between grants getting politically targeted and my lab’s building getting literally fucking bombed, 2025 is going great thanks

Jealous, given the state of my inbox

What is your company using that is allowing for auto-sifting emails? Copilot in outlook has not been useful for me (though I only briefly tried)

Dog on multiple pillows and blankets, set up in steps

Advanced use of pillows and blankets

Someone make a CNOT

At first glance I thought it was NTFS en.wikipedia.org/wiki/NTFS

I've been a proponent of downvotes/dislikes to counteract amplification of outrage posts, so I'm happy to see Bluesky trying it out, if in limited form

Yeah, that's old-time-y BS. Not a good argument by the editor, in general, IMO

With a minor exception that the journal may be looking to get the copyright, and Optica may have the copyright for the figures you submitted, so you may need to slightly change the figures and some of the text..

Faculty/staff at universities: what is the norm about your colleagues adding a meeting to your calendar and/or moving meetings via a calendar invite without first running it by you via email/chat or in person?

In contrast, my experience is that chatbots are still nearly useless when trying to write good reference letters (or most other serious writing), unless it's to break a case of writers block

Probably they would do better if I narrated (by voice) a bunch of thoughts and asked a chatbot to organize..

Used a two-period ellipsis ('..') and then wondered why I did. Googling doesn't seem to bring up anything definitive (eg www.reddit.com/r/writing/co..., where most people are saying '..' is not a real thing)

I think for me it avoids the possibility of sounding sarcastic? Or am I imagining the tone?

Agree with the pet peeve, though "I work on optics" works as a decent shorthand for me, so I don't have to give an elevator pitch..

"Botanical garden" must mean something different down under 😉

It's a high-quality life

Grams are right there

I'd like to trade schedules with the dog for a few days

I'm sad I never got to play around with that model

Ahh

Congratulations Sylvain!!

I tend to agree with your view but there are many people using the term 'white pigment' for scattering stuff like TiO2 particles scholar.google.com/scholar?hl=e...

I agree, though I do wonder what happens if you give the many copies of an LLM large scratchpads, a bunch of emulators, an ability to slowly rewrite parts of themselves, and have each be in its own infinite loop

In my mind only the first (mistakes of incomprehension) are actually Gell-Mann amnesia. The second (intentional simplification that looks like an error to an expert) feels like some other type of effect

Anyway, that's semantics

I totally agree*

*Not sure about "will always", but with the rest

3/3 Similarly, someone using an LLM may looked at the polished LLM output and infer that it is giving you expert-level output, even when it is not. That's a tricky failure mode of using LLM chatbots today -- this is the point I was making in the post you quoted

2/x That generalist journalist is an expert in writing (hence the text looks polished and professional) but not in the topic. It's hard for a non-expert reader to tell that there is a problem, because the heuristics (like bad writing) are missing

1/x What you have in mind (an intentional simplification for the benefit of the reader) is different from the example I had in mind, where a generalist journalist has to write a story about a topic in which they are not an expert, on a short clock

You had in mind a specific edge case that it did not address, whereas others working in the field (me and @jctravs.bsky.social, if I understood him correctly) thought it was a reasonably correct answer bsky.app/profile/jctr...

I'm sure I will learn the way in which I don't know optics, but it seems okay to me?

Figure 2. Comparisons of modeled spectra calculated from measurements of the multilayer 632 reflectors in the epicuticle of individuals with observed reflectance spectra in different elytral 633 colors (A-C: green, D-F: dark green, G-I: black). The TEM images demonstrate the pattern of 634 multilayer reflectors in the epicuticle. In each spectrum plot, the thin lines represent the means of 635 modeled spectra from different observation points (see Supplemental methods) and the grey area 636 shows the 95% confidence interval from measurements of different vertical transects, and the 637 measured spectra are denoted with colored thick lines.

A preprint to which we contributed (led by the folks at UW-Madison entomology) has just been updated www.biorxiv.org/content/10.1...

Check out these nice multilayer reflectors!

I agree, direct access to all scientific literature would make them super useful

It /does/ run into the problem of copyright and intellectual labor (where the current state of affairs is already not so good), so I suspect most people reading this thread would not be happy with your proposal...