I guess this sounds like FOMO (fear of missing out), or a rant.

Sorry about that.

Premia

I was reading Morgan Stanley’s 2025 surprises report when I noticed their use of “premia”:

Investors might worry less about deficit growth and term premia

And I thought, no wonder they’re at Morgan Stanley. They know the proper word to use. It’s almost like Kairos, the mode of persuasion.

Now, “proper” is ill-defined here; grammar changes with time and place, and I’m sure my use of singular “they” or royal “we” would annoy someone. But to the extent that I can have a coherent sense of self and grammatical correctness, I’d reckon that one should use the proper words.

So I am still one to use these proper words:

  • naïve
  • radii
  • résumé
  • indices
  • syllabi
  • crises

Maybe this is also why I think lawyer Latin is cool, like prima facie and stare decisis.

Like come on, I was in the IB, and they use ab initio in course names.

Plus people use a priori and de facto all the time. (Math people, I guess.)

And, when states come to the courts as parens patriae, you, as the state’s child, get to enjoy having things fixed for you.

The em dash

But people are lazy, me included.

(Here, I am reminded of this video of Mark Zuckerberg saying “I was human”.)

I used to use Microsoft Word, but now I use LaTeX or Typst or Markdown.

So I get lazy, and use a hyphen - in lieu of an em dash.

And while people say LLMs (large language models) are sort of average, but they use em dashes correctly.

Did you notice how the New York Times uses em dashes correctly too? It’s how you’re supposed to do it, folks. Thus my (tiny, but still there) frustration about people using em dashes to “spot AI”.

It’s just not how it works, you are supposed to use em dashes.

AI

Also, LLMs are “AI”, but “AI” is more than LLMs.

Here is another pet peeve of mine, calling things by their correct names:

  • Artificial lntelligence: anything from chess to statistical learning, to LLMs too. Probably using plain Python.
  • Machine learning: from naïve Bayes to LLMs. Probably using scikit-learn.
  • Deep learning: deep neural networks, probably involving graphics cards. Probably using pytorch.
  • Natural language processing: language. Probably using nltk or konoha.

And finally “generative AI” which I guess is fine, for those who don’t really care about the difference between Transformers and Diffusers.

Then there’s the naming problem, from Bun to Hono and React, Lovable to Fly.io, for Transformers and Diffusers too; but I digress. I guess it started from Java or something? At least “Vercel” doesn’t seem to mean anything else, I’ll give them that.

Prima facie

The second time I came across prima facie was in law; the first time was Double Effect, a song sung by ALSTROEMERIA.

I know English is not Latin, and law students study really hard, but for (Anime) J-Pop, it is perhaps more impressive that they learn English and Latin to write lyrics like that. Perhaps they have a dictionary to help them?

“Grammar Nazi”

I hope I did not sound too condenscending, or like a “grammar Nazi”.

I also don’t think we should be using this term at all. Wikipedia says this and I agree:

Critics argue that the term trivializes the historical atrocities associated with Nazism.

I only mention it to offer you this comedic sketch:

Grammar Nazis

(I use Oxford commas btw. Just to annoy some of you.)