I think it's just a different focus on what matters.
There's clearly enormous (human) work that goes in to building and tuning their models.
The software they develop around their model is, relatively, trivial. It's not clear that enormous amounts of (human) energy need to go in to that code.
It seems largely a function of applying human energy towards the things that are marginally most important to generating you money and putting you in front of competitors.
I mean a harness around a model is pretty trivial, I’ve also built one (heavily relying on LLMs to build it).
With a few thousand lines of code over the course of a week or so of my limited evening energy, I have agentic loops, ide integration, and a limited selection of custom tools which are all very safe (no unintended changes), along with session management and awareness of git.
It works great, is customizable, and allows integration with any provider.
If they need 30k changes per day, it’s not due to the nature of the problem!
I think claude code is just way, way, way more than "just a harness" at this point.
You have to remember all of the surrounding features it works with now, like the infra behind the cloud agents, or how remote control works, the mobile apps, their connectors, team skills and mcp servers, on and on....
You probably don't, but that's true of all software. 99.99% of users are going to use a small minority of the functionality but it will be a different minority
Oh, I mean I can't disagree there. But they've dug themselves into a hole where they're implementing custom ray tracing in the game engine that claude code runs on or whatever.
> implementing custom ray tracing in the game engine that claude code runs on
Haha, I don’t doubt it! Pretty damning overall for long-term maintainability of vibe-coded projects, even as the models are great at a lot of individual tasks
I mean, sure we can assume that (although the Anthropic article from the other day say that claude mostly builds Claude) but this is the Claude Code team which, I again am assuming so I could be wrong, focuses on the TUI/CLI interface to those models, not necessarily the models themselves. That coupled with the fact that this is how a large portion of their clientele interact with their moat, it would be odd to me if they didn't apply significant work towards this thing to keep the hype or productivity train moving
Perhaps? But a confounder is the strengthening or weakening of social ties. It's not clear that what seems to increasing loneliness is doing well by this next generation.
Connecting your comment to another about commercial model, seems the potential win here is selling useful tools to radiologists that may leverage AI rather than to end customers with the idea to replace some radiology consultations.
This seems generally aligned with AI realities today: it won't necessarily replace whole job functions but it can increase productivity when applied thoughtfully.
The mechanism that can shift supply to less polluting sources of electricity and shape demand for computation/electricity is pricing.
Measurement, pricing and reducing pollution externalities is a major challenge, but I don't think that's as unknown of a problem as the article seems to suggest.
Timely but I dare ask: I need to rework my thinkpad, still in good shape but with an older ubuntu distro. Reasonably fluent linux user but not expert, primarily dev workflows - what's the recommendation?
- Before 35: Mostly ubuntu, latest version (install after a couple of months of general availability)
- After 35: ubuntu LTS versions (install after a couple of months of general availability). Since 16.04 I've been on LTS only; should have done that earlier.
unless you really need a very new version of a piece of software, the safest bet will always be the latest LTS version of Ubuntu. You could perhaps just upgrade your laptop OS with the software package GUI?
Fun fact: You don't really need to install any of the "flavors" of Ubuntu to get the various desktop environments (KDE, GNOME, Cinnamon, etc.), you can just install them as a package! (`apt install kde-full`). You can switch between them when logging in.
Start with goals. “With all this stuff I hope…” or think what you hope to be true in three years. Work back from that. The default (which is great if you like) is I’d like to spend a lot of time shuffling digital resources.
One nice improvement is applying a constraint. Bard will now give a valid answer for "give a swim workout for 3000m" that correctly totals 3k, while chatgpt does not.
GCP organization structure is such a breath of fresh air after dealing with AWS. AWS Orgs and all the complexity with VPC and DNS management at a scale of hundreds of accounts is just a complete pain in the arse. GCP makes it far easier with Shared VPC and org/folder/project structure.
There's clearly enormous (human) work that goes in to building and tuning their models.
The software they develop around their model is, relatively, trivial. It's not clear that enormous amounts of (human) energy need to go in to that code.
It seems largely a function of applying human energy towards the things that are marginally most important to generating you money and putting you in front of competitors.