Tootfinder

from my link log —
Inside SPy: language semantics for a statically-typed compiled variant of Python.
https://antocuni.eu/2026/03/25/inside-spy-part-2-language-semantics/
saved 2026-05-23

Inside SPy, part 2: Language semantics
Language semantics: phases of execution, redshifting, static typing, and metaprogramming in SPy

Omni, which is building a "semantic translation layer" for enterprise data, raised a $120M Series C led by Iconiq at a $1.5B valuation, up from $650M in 2025 (Lily Mae Lazarus/Fortune)
https://fortune.com/2026/04/23/exclusive…

Exclusive: Omni raises $120 million to fix one of AI’s biggest enterprise data problems
The Craft Ventures-backed company is betting that a fully integrated, U.S.-manufactured chip can unlock a new drone supply chain.

Raiders cierran segundo día del draft sumando a línea defensiva y ofensiva https://www.raiders.com/news/raiders-cierran-segundo-dia-del-draft-sumando-a-linea-defensiva-y-ofensiva

🔀 Interleaved multimodal input: combine image text in one request — the model captures complex cross-modal relationships for more accurate semantic understanding.
📐 Matryoshka Representation Learning (MRL): output dimensions scale dynamically from 3072 down. Recommended: 3072 / 1536 / 768 — balance quality vs. storage costs.
⚡ Real results:
− Paramount Skydance: 85.3% text-to-video Recall@1
− Sparkonomy: 70% lower latency, ~2x semantic similarity scores

On a July afternoon in 2019, I found myself in a large, sun-dappled room within one of America’s great estates.
An assemblage of distinguished jurists, Ivy League professors, nonprofit leaders, journalists, and theologians sat around me in a half circle.
I was trying to be on my best behavior, but I blurted out a word dirty enough to make them blanch.
In my defense, I thought it was what I had been summoned there to do.
An independent commission had spent the previ…

Update your threat model accordingly.
The document does explicitly talk about essentially automating linguistic forensics. This is a case where the semantic ablation of an LLM could actually help retain anonymity, but that has yet to be studied (as far as I know). This is not 100%, and probably wouldn't hold up in court (if you have a good lawyer).
But the point is not that. The point is that it would be used to identify a set of high probability candidate and then use targeted attacks from there. This does seem to miss. I haven't read far enough in to see why.
It also seems to be going on a lot more than linguistic forensics. It also doesn't seem to be about deanonymization so much as connecting accounts across platforms, where accounts may not be trying hard to obscure their identity. While this is definitely the type of research that would get very specific types of funding, that doesn't mean it works exactly as advertised.

On the <dl>
It’s used to represent a list of name–value pairs. This is a common UI pattern that, at the same time, is incredibly versatile. For instance, you’ve probably seen these layouts out in the wild…
🌐 https://benmyers.dev/blog/on-the-dl/

On the ‹dl›
The semantics you didn’t know you needed.

blast from the past:
https://www.youtube.com/watch?v=MacQpu8JniQ
How are things going, whatcha think?
addenda:
https://www…

Heute für eine weitere #Radtour mit dem #BromptonElectricGLine #rausgeschafft. Dummerweise hatte ich die Sonnencreme zuhause stehen lassen und vergessen, mich einzucremen. 🙄 Sobald die Sonne …

The US FDA drops an enforcement complaint against Whoop over its blood pressure tracking tool, reversing a July 2025 warning letter; Whoop is updating the tool (Samantha Kelly/Bloomberg)
https://www.bloomberg.com/news/articles/20