OtherShrezzing

This has recently been a fairly wild journey to watch from the outside.

Best of luck to Graham. I've got no idea how they'll make CSS-Tricks work as an indy blog in this era, but I hope they do make it work.

>This is in fact very similar to one of Anthropic's guiding principles: underpromise and overdeliver."

If this is one of Anthropic's guiding principles, then I must assume that they've only ever promised five 8s of reliability.

>Anthropic is not in the model business. They were smart and they are building productivity ecosystem business... It's at least a Google Docs killer if not an M365 killer.

Google Workspace revenues are something like $25bn/q, and that's the entire workspace division, (including enterprise gmail sales, so the productivity suite is meaningfully smaller than that figure). It's taken Google a long time to carve that slice out of Microsoft's entrenched market position. It's not small money, but they'd basically be going to war with two of the largest and most profitable companies in history on their turf.

I don't think they can credibly take that business plan to the market and say it'll be worth $1.75tn. They're all-or-nothing on being in the model business, and everything else is superfluous.

> For any application at scale it'd be way more accurate to have a dedicated classifier using the right architecture

There’s lots of applications where it’s not being deployed at scale. Also lots of cases where someone who knows they need something classified does not know what a classifier is.

My reading here is that they've gone through their logs retrospectively, because you can see these attempts to exploit pretty clearly in an observability platform, and the 17th is only five days ago, so you can assume they've got a lot of fully detailed logs over that period.

I think this project is a very good way to explore the question:

>What's the difference between a model that predicts intelligent output, and intelligence itself?

It doesn't directly attack the question, but gives a neatly constrained thought experiment which you can see running.

The same difference between getting rained on, and prediction of getting rained on.

>PHP 8.1 and older no longer receive updates from the PHP project. Some distributions and LTS vendors provide backported fixes, so contact your vendor for details, or upgrade to a supported version.

I get that maintenance teams need to have conviction in this approach, but this is definitely going to cause catastrophically large data breaches downstream.

Beds as opposed to seats mean fewer passengers per train carriage, so the fixed cost of operating the train is shared among fewer tickets.

You can make one trip per day with a night-train, whereas a seated train runs 16-20hrs per day. So again, fixed cost of the train existing is shared between far fewer ticket sales.

Then you need to store the train during the day, and you need to securely park the train when it arrives at its destination at night (a normal train yard isn't secure enough if you have a hotel full of people, and a train station usually needs all of its platforms).

GDPR doesn't cover unrecorded human memory, so you're fine to overhear whatever in a day-to-day setting. If you then do anything that looks like digital processing or cataloguing that information though, then you're making it a GDPR issue.

I think “covered well” is being used here as a synonym for “covered at all”.

For example, the “social workers & care -> counsel parents with child rearing problems” category is considered “well covered” because there are 13 prompts.

It’s self-evident that the spectrum of work carried out by a social worker undertaking that work cannot be well represented by thirteen prompts.

I suspect if you really strictly defined “covered well”, to the extent that any normal person would use that phrase, the headline would be above 99.9%.

It’ll be a few months further down the line. The developer has said their first obligation is to fix any bugs found by the people who bought the game, so they wanted to leave space for that process.

I think C# wrapped by .NET does hit all of your requirements. Might need a bit of work bringing compilation time down on enormous codebases, but for 95% of projects 95% of the time, compilation takes only a few seconds, which is fine in the context of an LLM which takes dozens of seconds to reason.

Error messages are mostly useful; compilation is quick (relative to an LLM); memory is not really optimal in most cases, but it's "good enough" and the language puts up a lot of guardrails around the programmer to keep them on the path to accidental success in this regard; the training data is full of .NET code; there's a well defined set of ways of working & best practices with .NET; types are explicit.

Interestingly, Claude has generated a LICENSE file which demands commits & changes are GPG signed. But the repo's license file has no GPG metadata (nor does the rest of the app), so it's impossible for anyone to commit to, extend, or meaningfully change this code without violating the repo's license.

In short, nobody can commit to, change, or meaningfully adjust this repository while adhering to its license.

I think the poster is saying that adding an MCP feature to it is trivial, which doesn't say anything about the complexity in building the tool in the first instance.