dmazzoni

(This is from August 19, it was widely reported at the time.)

So then they should be pickier about when they show results or which model they use based on the question.

Nobody asked for an LLM response for every single search.

They used to detect certain types of queries and offer direct answers when the query matches. In my opinion that’s how Gemini in search results should work.

I can forgive them for having a bug in their sandbox.

I can't forgive them for failing to do any monitoring whatsoever.

I don't think it was intentional or marketing, but I think it was criminally negligent and they should be held responsible.

They gave powerful models with no guardrails access to the Internet and didn't monitor it.

Even the slightest bit of monitoring of their outgoing Internet activity would have immediately given it away and they could have shut it down.

They were asleep at the wheel, and that's just plain negligence.

No, most people don't cheat. The world would fall apart if they did.

In a lot of occupations the skills you learn in college are real and meaningful. If you don't have the skills, you can't do the job. You can't fake it 'till you make it if it's painfully obvious to your coworkers that you don't understand basic concepts.

When Chrome 1.0 came out in 2008, it only ran on Windows, and it couldn't do basic things like print or export to PDF, didn't have any accessibility, didn't work in RTL languages, and didn't have any graphics acceleration, among other limitations. Don't get me wrong, it was a marvelous piece of engineering - but it was extremely incomplete. By the time it was what we think of as a modern, complete browser, they had many hundreds of engineers working on it.

Also, you're comparing the number of engineers working on one product, to the number of engineers working at Shopify across their entire company today, which includes far more than just one app.

You're entitled to your opinion that Shopify is a poor app or that it's overengineered, but clearly it's not simple. That's just a fact. If it looks like a simple app to you, that's because you're not seeing most of it.

It's like looking at the YouTube mobile app as someone who watches videos and saying that the app frontend looks simple. You're only seeing 5% of the app. The other 95% is for video creators and advertisers, and it is decidedly NOT simple. I think that's the same here.

This is based on my understanding from several pilots who have weighted in.

The cockpit voice recorder transcript has the copilot repeatedly saying "we're coming in too fast" but the pilot never clearly acknolwedges.

The copilot ought to have been more assertive and said "We need to go around. NOW."

The pilot shouldn't have needed that. He should have made the call himself, but even if he was unsure, once his copilot called that he should have done the right thing.

Neither of them will ever fly again. The only question now is if either of them will go to prison. There's already a civil wrongful death lawsuit against them.

In practice that’s a lot of the value MCPs are providing.

Technically it’d work just as well to have a thin API wrapper or cli tool the agent is allowed to access.

I think that’s true when you compare to the average white collar professional who does a lot of writing: better at writing, not better at communicating.

But compared to the average adult? I think you forget just how bad at writing the average person is.

As opposed to what?

If you ask a non-programmer to install Claude Code, just installing it will be a challenge, then opening the shell and interacting with it. Things as simple as copying and pasting can present roadblocks if you've never used a shell before, and things intuitive to programmers like using up-arrow to go back to a previous prompt would never occur to someone in the field.

Claude Code seems so simple and natural of a UI to programmers, it's easy to forget how much it builds on.

When has 20% projects ever been about bypassing every launch process and just posting your product publicly?

Google may be a big bureaucracy now, but launch approvals and processes are there for a reason.

My initial thought there is that you'd have an imbalance. Many token patterns would almost never come up with the assistant tag on them, for example words with typos in them.

255 characters ought to be enough for everybody, right?

Save As works fine for simple websites with static content.

Let's say you have a site that fetches content from a database. If you Save As, then at best you'll get a local copy of an HTML page with JS that loads the content from the same remote database. It might not work (since the local copy has a different origin), or if it does, it requires you to be online, which defeats half of the purpose.

What this project, and SingleFile, both do is save a snapshot of what the rendered page actually looks like at that moment in time. The scripts are stripped out so it runs locally and has no external dependencies.