iiinova — where your digital self keeps evolving

My masters is in CS, and now I mainly work with academics in CS, econ, and law. And they all do things differently from each other.

010

Wes Cowley @wes.wordsbywes.ink · 8h

Many econ working papers eventually make it to formal journals, some don't. In the meantime, most of them are published on SSRN. Job market papers like this one sometimes just live on the author's site or github forever, sometime they morph into working papers and/or journal papers.

110

Wes Cowley @wes.wordsbywes.ink · 8h

I've seen similar messages recently on Windows Chrome and traced it down being triggered by sites requesting passkeys.

100

Wes Cowley @wes.wordsbywes.ink · 8h

I'm also saying you can't necessarily judge the quality of the paper by the way it's been presented and reviewed vs how it's done in CS, because they do that type of review differently in econ.

010

Wes Cowley @wes.wordsbywes.ink · 8h

Again, it works a bit differently in econ in than in CS and other fields. Working papers like this list conferences they are presented at in the acks. I'm not saying one way is right and the other way is wrong. I'm saying that's the way it's done in that field.

200

Wes Cowley @wes.wordsbywes.ink · 9h

I'm not an econ PhD, but from what I've seen, it works a bit differently. Papers get presented and discussed in various forms at different conferences and seminars, sometimes over years, before they're eventually published. At a quick glance I see at least two conferences in the acks of this paper.

100

Wes Cowley @wes.wordsbywes.ink · 9h

That's the norm for economics job market papers (I edit a lot of them). The point of the paper is to showcase a new or soon-to-be PhD's work for potential employers.

110

Wes Cowley @wes.wordsbywes.ink · 07/10/2026

I mean, at least they edited out the red squiggly line.

010

Wes Cowley @wes.wordsbywes.ink · 02/10/2026

By this metric, Excel 2026 blows GPT away.

050

Wes Cowley @wes.wordsbywes.ink · 29/09/2026

I run a feed listing academic calls for papers. This list is simply to help me filter out posters who use "CFP" for football or financial advising. Nothing negative is intended.

110

Wes Cowley @wes.wordsbywes.ink · 26/09/2026

LinkedIn bro, probably: "How I Got My Job by Bringing Iced Coffee to an Interview: I told an agent swarm to find out the interviewer's name and surveil him. The agents found his iced coffee order, so I showed up with it and slid it across the table. Buy my course and I'll show you how."

001

Wes Cowley @wes.wordsbywes.ink · 23/09/2026

What could to wrong? Well, her shoulder surgery to start with.

021

Wes Cowley @wes.wordsbywes.ink · 23/09/2026

I've been hosting my site there (mainly static html along with mp3s and and a bunch of videos) since I moved off Wordpress, and I've been very happy with them.

120

Wes Cowley @wes.wordsbywes.ink · 16/09/2026

I finally said, look, does this really need to be a book? Maybe write a series of articles, then if you want to make it into a book, we'll deal with that then. Fast forward two years and she had self-pubbed two books and was working on her third.

020

Wes Cowley @wes.wordsbywes.ink · 16/09/2026

A4: I rarely do coaching, but a few years back I worked with a client who was doing her first nonfiction book. I was like the fourth person she'd worked with (I should have noped out, but there were Reasons). She'd make progress, stall, and start from scratch. Over and over. #StetChat

231

Reposted by Wes Cowley

Caitlin Cooper @caitlinmaycooper.bsky.social · 24/08/2026

Best plural forms ranked: 1) internal (tours de force, culs-de-sac) 2) ending in -i (magi) 3) is to es (oases) 4) suppletive (people) 5) vowel changes (geese) 6) f to ves (thieves) 7) adding en (oxen) 8) no change (moose) 9) adding es, ies (boxes, babies) 10) adding s

3031876

Reposted by Wes Cowley

Wes Cowley @wes.wordsbywes.ink · 04/06/2026

New feed aggregating calls for papers, abstracts, and so forth. #cfp #academicsky bsky.app/profile/wes....

001

Reposted by Wes Cowley

Dr Kiri Thornalley @kizzythechemist.bsky.social · 04/09/2026

If you want to have fun, every PhD awarded by a UK institution should be on here: www.bl.uk/collection/e... The most recent ones aren't because of a cyber attack in October 2023.

bl.uk

E-Theses Online Service (EThOS) update

EThOS (E-Theses Online Service) is the British Library’s database of UK doctoral theses.

2231

Wes Cowley @wes.wordsbywes.ink · 02/09/2026

Q5: I'm late to this, but it reminded me of the time I needed to reference the Simpson's fan wiki for a nonprofit's impact assessment. #StetChat

241

Wes Cowley @wes.wordsbywes.ink · 24/08/2026

Great thread about the NYT opinion piece that erroneously conflates debt default and debt forgiveness, and misunderstands Sumerian history in the process.

0117

Wes Cowley @wes.wordsbywes.ink · 23/08/2026

Another advantage of tablets: you can read from multiple vendors: kindle, kobo, google docs, generic epubs and pdfs all on the same device.

110

Wes Cowley @wes.wordsbywes.ink · 23/08/2026

And many ereaders have vendor lock in to some degree (kindle, nook). Tablets you can do most anything with in addition to books, so one less device to lug around. And they usually have bigger screens, so less page turning, especially if you like bigger text. But often more expensive than a reader.

111

Wes Cowley @wes.wordsbywes.ink · 23/08/2026

ereaders tend to be just that: you can read a book on them. They're usually smaller and lighter than a tablet and can be easier to read in some lighting situations. Because they're dedicated, you're not getting distracted by notifications or the urge to browse Youtube or bsky.

111

Wes Cowley @wes.wordsbywes.ink · 20/08/2026

Searching Bed, Bath and Beyond for wall sconces...

010

Reposted by Wes Cowley

K.B. Wagers @kbwagers.com · 18/08/2026

Looking for recommendations for a friend for freelance romance editors. Short story to novel length. Hit me up, especially if you're the editor in question. (what was once obvious now needs to be said, if you're using AI to "edit" I'm not interested)

102423

Wes Cowley @wes.wordsbywes.ink · 08/08/2026

These guys are going to finally build the AI that tells them how to solve the climate crises, and it's going to say, "replace your coal, gas, and oil plants with solar and wind farms," and they'll say, "damn, I thought that one was it. We need a bigger AI."

12210

Wes Cowley @wes.wordsbywes.ink · 28/07/2026

*opens Instacart* *orders all the popcorn*

010

Reposted by Wes Cowley

Boxo McFoxo @boxobark.ing · 19/07/2026

The answer to "if you don't adopt AI you'll be left behind" should be "yes, please go away and leave me behind, that sounds ideal"

2252

Wes Cowley @wes.wordsbywes.ink · 18/07/2026

That was my sense too, but my knowledge of those fields is just enough to know that using an LLM for them would be like using a jackhammer to drive a screw. Thanks for pointing that part out with more authority than I could.

050

Wes Cowley @wes.wordsbywes.ink · 17/07/2026

And here's a thread on the ChatGPT issues with the article. (7/7) bsky.app/profile/nano...

160

Wes Cowley @wes.wordsbywes.ink · 17/07/2026

Here are the articles linked from the quote in the first post in this thread: AlphaFold: www.science.org/content/arti... and GNoME: www.science.org/content/arti... (6/7)

science.org

Science’s 2021 Breakthrough of the Year: AI brings protein structures to all

Bounty of new structures will forever change biology and medicine

170

Wes Cowley @wes.wordsbywes.ink · 17/07/2026

And, especially in articles aimed at technical and academic audiences (but also those aimed at the public), we should be accurate in describing what we're talking about. Lumping everything into "AI" or "LLMs" is misleading and unhelpful. (5/7)

1161

Wes Cowley @wes.wordsbywes.ink · 17/07/2026

I'm not saying there aren't issues with the systems being discussed; there may be — in particular, overblown hype and expectations and misattribution of credit to the "AI" rather than the researchers. But they aren't the same issues. (4/7)

1150

Wes Cowley @wes.wordsbywes.ink · 17/07/2026

Not all systems using deep learning and neural network–based prediction techniques are LLMs. Why does the difference matter? Besides the accuracy issue, LLMs come with (rightly earned) baggage over ethics, copyright, etc. that don't necessarily apply to these other systems. (3/7)

1193

Wes Cowley @wes.wordsbywes.ink · 17/07/2026

That quote links to articles on AlphaFold (and variants) and GNoME, neither of which are large language models. AlphaFold shares some concepts with LLMs (attention and transformers); GNoME not so much. (2/7)

1193

Wes Cowley @wes.wordsbywes.ink · 17/07/2026

While this article in Science is getting dragged for leaving ChatGPT trackers in links, there's a subtler issue. The author states "[T]he ability of large language models (LLMs) to predict protein structures and to accelerate the discovery of new materials has revolutionized both fields." (1/7)

23313

Wes Cowley @wes.wordsbywes.ink · 17/07/2026

Fwiw, it's fairly easy to tell which tags are needed: take them off and try it. For the trackers like utm, I've never seen a case where they are needed.

081

Wes Cowley @wes.wordsbywes.ink · 17/07/2026

1) ChatGPT is not a search engine. 2) Why on earth would you and (more importantly) your copyeditors leave the utm tags on the links in a published article? That's just sloppy (no pun intended).

455813

Wes Cowley @wes.wordsbywes.ink · 16/07/2026

You could probably replace the whole list with "leaders need to get their shit together", though that might not meet the style guide's tone guidelines.

021

Wes Cowley @wes.wordsbywes.ink · 15/07/2026

Soooo many examples like that.

010

Wes Cowley @wes.wordsbywes.ink · 15/07/2026

A5: I'm late to this one, but here's an real example I've used before when talking with a client: A table reports results from a regression where the effect is in percentage points. The discussion and conclusion reports the same numbers as percentages. Is an LLM going to catch that? #StetChat

142

Wes Cowley @wes.wordsbywes.ink · 15/07/2026

Are the references accurate and formatted correctly (and lately, do the cited papers actually exist)? Are figures and tables clear and readable? Are there logic jumps? Are spelling, grammar, and punctuation right? All of those (and more) are things I regularly clean up or flag. (2/2) #StetChat

0102

Wes Cowley @wes.wordsbywes.ink · 15/07/2026

Q1: I mainly work on academic papers/books. Lots of points where editing adds value: is the writing clear for the reader? Is it accurate (names spelled correctly, narrative matching tables/figures, math consistent and correct, etc.)? Does it meet word-count limits? (1/2) #StetChat

2102

Wes Cowley @wes.wordsbywes.ink · 08/07/2026

This addendum brought to you by yet another case study that boasts huge typesetting and production speed-ups but says nothing about quality. (6/4)

010

Wes Cowley @wes.wordsbywes.ink · 08/07/2026

I should add that a human review of the typeset proof is even more important now that publishers are starting to use LLM-driven typesetting and publishing flows. (5/4)

110

Wes Cowley @wes.wordsbywes.ink · 08/07/2026

I don’t see those as things that can be reliably handed to a text generator. Even at proofread, there’s no substitute for someone (author or editor) looking at the paper from the points of view of the author (does it say what they meant to say) and the reader (does it say it clearly). (4/4)

120

Wes Cowley @wes.wordsbywes.ink · 08/07/2026

Others issues will be from partial previous edits (oops, we’re referencing table 4 when it got cut for space). Others will be visual: how do the text and graphics fall on the page? How do the tables look? Are there poor hyphenation choices? (3/4)

100