I keep seeing posts about TASTE being the only moat left. It's unclear what they mean by it. Who defines what's tasteful or distasteful? Is it simply a visual matter, or character? Does it supersede functionality?

token count aside, doesn't HTML have better semantics than markdown for LLMs to reason with?

"You can't manage what you don't measure." High-leverage individuals like C-suite execs, investors, advisors, and board members are only ever effectively briefed or given SNAPSHOTS of an organization's health. The data they receive is effectively REDACTED. There's a lot of

structured-data-is-the-foundation-of-trustworthy-AI.md

"structured data, you've heard of it, SQL..."

tAsTe and vibes: most vibe-codebases come off N+1'y. They run, but, they aren't tAsTeFuL. LLMs are PROBABILITY calculators, and because the code in the response they produce will PROBABLY be bloated, the only LOGICAL conclusion is that their training data was CERTAINLY bloated

Rhys I'm not very happy with the code quality and I think agents bloat abstractions, have poor code aesthetics, are very prone to copy pasting code blocks and it's a mess, but at this point I stopped fighting it too hard and just moved on. The agents do not listen to my instructions

"It's hopeless trying to produce The Beatles. No one can do it. The four individuals are too strong individually." - Paul McCartney while George Martin sits behind him

x.com/i/article/2035…

Work's Primitives and how they used to be handled: Team Communication: Email, PDF attachments Reports and Research: Email, PDF attachments Technical Design: Email, PDF attachments

"it's like an Oura Ring for your organization, constantly checks all kinds of "biomarkers", scans for trends, anomalies, etc. lets you know what's working and what to do to improve"

what problem do LLMs actually solve? a roundtrip from English to code and back? what if neural networks are the wrong abstraction? the whole world is racing to build around a novel and experimental tech: token-based autocomplete. meanwhile, as we discover its not-so-meaningful

“A computer can never be held accountable, therefore a computer must never make a management decision.” – IBM Training Manual, 1979

"it was just a clean job that had the clout of CERN behind it" Ted Nelson discussing the World Wide Web

The recent trend/influx of high-production announcement videos should serve as reminder why Bezos banned slides at meetings and replaced them with PRFAQ. Work backwards, indeed.

Soft skills are hard. They require listening, which requires patience, which requires caring. Reassuring you've understood the problem at hand, and which steps come next, require clear communication. Can't really be taught. You can apply these to prompting, selling, product