Sebastian Karcher

Adding alt text to figures in quarto with Claude Code Alt text is a very important thing to make sure you do right when making websites. Using AI agents such as Claude Code can help with a lot of the heavy lifting.

and if you have a billion images to work with (like me teaching dataviz lol), this by @emilhvitfeldt.bsky.social is super helpful for an initial pass at making alt text for plots

What happened to polisci???
(It'd also be funny bc I'd expect broad agreement on top 3 and complete mayhem on the other 2)

This is the actual (AI generated) reason an NEH grant got cancelled...

NEH's budget is tiny—not just compared to NSF/NIH, but compared to humanities funding in every other wealthy nation. What little was there was largely rescinded and several programs were cut in full last year. It decimated not merely individual projects but whole corners of the humanities in the US.

Go look the last half dozen of your interactions with other users and think about whether you've made this a better place. Bye.

I know. That's why getting them the right tools is important. (And my bio is public. What are your qualifications to opine in this?)

Basically, yes. More precisely, I'm asking "are pediatricians in general well trained & equipped to answer qs about vaccine safety to parents steeped in a vax disinfo environment".
I'm very much not convinced that's the case

Or, to put it more bluntly: insisting that the only acceptable strategy is to condemn parents who are antivax as bad parents is only marginally less anti science than denying the efficacy of vaccines themselves

It's a professional obligation when you work in social problems of this sort to treat people w/unconditional positive regard. That does NOT mean you approve of their behaviors or don't provide serious feedback about harm. But it is straight up unethical to decide it's your job to say who is "bad"

Do you think doctors necessarily have a good way to explain how we know they are safe to people who have heard/been fed vax misinfo?
In a perfect world this might not be necessary, but in the info environment we're in, it probably is

Application Form Senior Editor Clinical/Associate Editor Social Section Collabra: Psychology Starting from 1 July 2026, Collabra: Psychology is on the look-out for a new senior editor for the clinical section as well as several new associate editors for the social section. If you are interest...

Contribute to open science! Collabra: Psychology needs a new senior editor for the clinical section as well as several new associate editors for the social section. If you are interested, please fill out the application form before 30 April 2026. Repost please!
forms.gle/DgM3484SuLVD...

See the second link for some available 1.18 installers -- if you have an old back-up they might still work (they won't work on the 1.19 database though).

What version (exactly) of Mendeley do you currently have and do you still have the data? Zotero has pretty robust import from Mendeley, but the default version uses synced data; using local data is more complicated. See
www.zotero.org/support/kb/m... and this thread
forums.zotero.org/discussion/1...

Happy dog?
Happy dog carrying a white baseball hat in its mouth?

The 2026 APSA DDRIG cycle will open on April 1, 2026, and close on June 1, 2026. Provides support to enhance and improve the conduct of doctoral dissertation research in political science. Connect awardees to APSA’s extensive professional development and public engagement networks and resources. Supports the advancement of national health, prosperity, and welfare. The APSA Doctoral Dissertation Research Improvement Grant has made 139 awards to doctoral candidates from 53 PhD-granting institutions since 2020.

tell your friends: the APSA DDRIG application for 2026 will open on April 1. Updating the FAQ pages with some additional information now.

You'll get free advice on prompting :P

One of my go-tos: "If no one's responsible for that task, that's exactly who does it"

So much of data management (and a lot of other task-oriented stuff) is just roles and responsibilities

See, this is the AI posts I'm here for!
As a bad coder w/o AI, I'm still pretty sure Claude Code is pareto better than me, but now that it lets me do more ambitious things, I do think about this a lot.

We’re hiring — help us spread the word! @tnridout.bsky.social @mikefranz.bsky.social

I really like lefty Mormons. The same earnestness & cheerfulness, but woke.

The future continues weird. This is a nice example at the now common intersection of “It’s remarkable that this is even possible”, “You can see the vistas it opens up”, and “Pretty sweet for the company to have people use their service in this very pay-as-you-go resource-intensive way”.

Meet three scientists who said no to Epstein The warning signs included a web search, a mother’s doubts, and inklings of a “sexist attitude”

This is a weird piece, but noteworthy that 2/3 stories of male scientists staying away from Epstein involve intervention by a woman. I can see the (justified) eyerolls, but also, diversity in teams matters. You shouldn't need your mom for this.
www.science.org/content/arti...

Ha! The original Lancet article on the dangers of reading in bed is here: doi.org/10.1016/S014...

They are very good at standard proofreading, catching things that standard spell/grammar checkers don't.
I've also had good success with prompting for unclear passages. I rarely take specific suggestions on those. They often lack precision.

Finishing up Day 3 in the Paralympic Village, great to see my mixed doubles teammates starting off with a big W against a very good Latvian team! 🇺🇸🥌

Github diff isn't remotely close; you can't see edits from even three people & you don't have highlighting & editing in the same doc and not accept/reject at the line level. Meld at least give you 3-way comparison & merge, but that is line-by-line: very much not the same, and w 4 authors you lose.

I guess my larger point is that Word is somewhere between not great & bad for every individual task, but there is nothing that can replace it wholesale while keeping basics very easy to use.
People use it for a reason, not just lock-in/habit.

Is there a Quarto/Rmd editor that gives you usable track changes with accept/reject option? I can obviously diff and with a good tool that's OK, but I don't think it's close to what you get in Word (or GDocs).
Hard to see how a plaintext format can do that on the same level

No, I want to write while tracking changes w/o seeing every change in the doc as a Pollock painting of crossed out & added sentences. In GDocs you can preview "accept all" but you can't edit then

GDoc doesn't let you edit without seeing everyone else's edits. Word does. And GDoc is just missing too much functionality otherwise (no figure/table captions???)