mudasobwa

mudasobwa

Harness, not model is what matters most

Somewhere, right now, a senior engineer is on the verge of a nervous breakdown because his company will not let him switch from Claude to GPT-5.2, or from GPT to Gemini, or from any of them to a locally-hosted llama that smells faintly of the intern’s vape juice. He is convinced that the model is the product, that the difference between shipping and slop is measured in the number of parameters, and that if only management would approve the more expensive API key, all his problems would evaporate like a hallucinated citation.

He is wrong. Spectacularly, expensively, and in a way that will cost him exactly the difference between the two API tiers he is agonising over.

Allow me to explain, in the manner of a man who has spent the last year building the thing he is about to tell you matters more than the model itself.

Important notice: my blog has been moved to https://blog.yokerhood.com for good.

Most Liked

mira_tokarev

mira_tokarev

I agree that the harness is where a lot of the practical economics lives, but I’d separate orchestration from one quieter cost: how much already-seen state gets serialized into every next call. A model swap can hide that variable, whereas a deterministic local compression step gives the harness a reproducible place to measure it before adding routing or memory complexity. One small tool I’ve been evaluating is Contextpress (GitHub - Taha-azizi/contextpress: Deterministic context compression for LLM chat, RAG, and agent pipelines · GitHub · contextpress · PyPI); it operates locally on the message list rather than making a second model call. I’d compare harnesses on cost per successful task, retries, and time-to-result, because “more capable model” and “less repeated context” are different levers.

smithjj

smithjj

The framing is a bit dramatic but the point lands. The harness, context handling, and how you structure prompts usually matters more than which model you’re calling.

Where Next?

Popular Backend topics Top

First poster: bot
It’s easy to view yourself as “not a real programmer.” There are programs out there that everyone uses, and it’s easy to put their develo...
/c
New
First poster: bot
What’s Next for Teal, the typed dialect of Lua - FOSDEM 2021. This is my talk about the latest updates on the Teal programming language,...
New
New
First poster: bot
Too long have we hustled to deploy Clojure websites. Too long have we spun up one server instance per site. Too long have reminisced abou...
New
pablocostass
Todos coñecemos os focos de Erlang/Elixir máis renomeados do mundo, como a Suecia, o Brasil, a California ou Londres. Mais a comunidade, ...
New
brainlid
There is a new community resource available on writing “Safe Ecto Migrations”. When we get a migration wrong, it can lock up your product...
New
tonyxrandall
As DoorDash transitioned from Python monolith to Kotlin microservices, our engineering team was presented with a lot of opportunities to ...
New
ragamuf
Does the world need another How to create a blog article? Maybe not. But then again, creating something out of nothing is what we love....
New
mtmattei
For the past few years, Safari has been putting in a lot of effort to enhance its WebAssembly support and 2024 was no exception… I believ...
New
brainlid
Episode 264 of Thinking Elixir. News includes Phoenix v1.8 shipping with an AGENTS markdown file for new apps to enhance coding agent exp...
New

Other popular topics Top

AstonJ
If it’s a mechanical keyboard, which switches do you have? Would you recommend it? Why? What will your next keyboard be? Pics always w...
New
PragmaticBookshelf
Stop developing web apps with yesterday’s tools. Today, developers are increasingly adopting Clojure as a web-development platform. See f...
New
DevotionGeo
I know that -t flag is used along with -i flag for getting an interactive shell. But I cannot digest what the man page for docker run com...
New
AstonJ
I ended up cancelling my Moonlander order as I think it’s just going to be a bit too bulky for me. I think the Planck and the Preonic (o...
New
PragmaticBookshelf
Author Spotlight Jamis Buck @jamis This month, we have the pleasure of spotlighting author Jamis Buck, who has written Mazes for Prog...
New
First poster: bot
zig/http.zig at 7cf2cbb33ef34c1d211135f56d30fe23b6cacd42 · ziglang/zig. General-purpose programming language and toolchain for maintaini...
New
PragmaticBookshelf
Leverage Elixir and the Nx ecosystem to build intelligent applications that solve real-world problems in computer vision, natural languag...
New
CommunityNews
A Brief Review of the Minisforum V3 AMD Tablet. Update: I have created an awesome-minisforum-v3 GitHub repository to list information fo...
New
pragdave
Build robust LLM-powered apps, chatbots, and agents while mastering AI engineering principles that will help you outlast the tools and th...
New
CommunityNews
Open-source implementation of the classic GTA engine now running directly in your browser. Experience the reVC technology demo on DOS.Zon...
New