saul

tradcode

@thomasmurphy__ x.com/saulhoward/sta…

There's a teleological snare whereby the very expertise required to evaluate and direct the products of LLMs is hindered by LLM assistance. You can't become a senior dev or develop your own writing style without a lengthy "manual" period of time spent on the details.

The peak of "we're Palantir for x", the trough of "we're Accenture".

@thomasmurphy__ x.com/saulhoward/sta…

You don't need a gloopy humanist argument about the sanctity of the creative author to be dismissive of LLM-generated prose. The aesthetic criterion that matters is simply that it is recomposed from stock (& therefore tends towards cliché, kitsch and stagnation, like bad writers)

The millenial version

nix hints to the direction of the new infra layer. A spec for an abstraction over userspace.

The bull case for software in 2026

Slop can be "genuinely useful", it's still slop. Technical debt is a tradeoff.

The exceptions where LLM-generated code are genuinely useful don't justify the cost (loss of team trust, understanding, skill, shared learning, feel for the abstractions, cruft pain, ability to zoom in/out). For PMF-prototypes or plugins, maybe it makes sense, but for platform engineering so far it's been a road to nowhere.

User experience is commoditized. Products can have infinite features with infinite customization. UI complexity no longer differentiates. Simplicity, reliability, trust wins.

Developing on a framework shipped by an LLM Provider is like doing your grocery shopping at the gas station.

@RohitPaulK @aboodman I wish some of these companies, Linear included, would see themselves as infra. There's huge demand. But SV dynamics pushes them to growth, which means more slop features.

@aboodman For a typical SaaS company, say Linear - would you say that code is a means to an end? Or is it the product?

OK fine here is my thought. To a close approximation: - When code *is* the product, you are going to want expert human engineers carefully controlling the result to maximize reliability - When code is a means to an end, you won't Notice that this rule can be run in reverse: because humans control the result of products that are code, I theorize you'll see increased demand for "productization" of more and more infra. Basically a bifurcation of the software space: infrastructure and slop. And I don't mean slop in a disparaging way. There will be lots more quick software getting written by everyone. It won't ever be perfect, but it will get the job done. But in order for this to work well, it needs to sit on an absolutely solid foundation. More solid than the software community has produced before. This foundation will be very carefully managed by expert practitioners. It will have to be, because we will have 1000x more software running on it than we ever have before, created autonomously.

@aboodman Yes absolutely. AI slop (also not disparaging) is the new "frontend" and programming becomes "infra". And the challenges this new infra layer presents are mind boggling.

The returns to doing good work, taking care and being thoughtful have never been higher and it's just getting started.

If the LLM writes it why does it matter if it’s Rust or Zig?

Explaining to the CEO what a PhD in CompSci means: "would you hire a professor of hydrology to fix your plumbing?"

Slop is hitting hustle cultures hard. Gamblers love a cheat code.

Egregore as a service