As kids we used to have great fun knocking rocks together around sunset to get them to call back. Kinda like beetle bird calls.

My view mostly it was a confluence of poor dev experience and over-engineering that killed them.

Some of those protocols were well designed. Some were secure, all were pretty awful to implement.

It’s worthwhile calling out REST as a long term success. Mainly because it was simple and flexible.

Whether MCP will have that staying power I dunno, personally I think it still has some flaws, and the implementation quality is all over the shop. Some of the things that make it easy (studio) also create its biggest flaws.

The problem for me is that it’s not worth running these myself, yeah I may pay attention to which model is better at tool calling. But what matters is how well it does at my use case.

In my team I see examples like yours, but I also see engineers having to clean up slop when one of them got over their skis.

My _belief_ is that it’s a net positive, but we’re far from taking that as a hand wave fact. And it certainly ain’t the 5-10x people are shouting across the board.

See the MIT Nanda study, and the other one from a few weeks back on the perceived vs actual productivity increases.

There is value, but so far nowhere near as much as the people pushing AI would like you to believe has actually been delivered

Unless it’s proven to be a murder, they can’t report it as such. They can say an investigation is launched etc. they’re pretty good at following up with subsequent articles or even changing the headline once facts materialise.

I agree with your base point though. “Demand better” should be the new war cry

Vote with your feet and your wallets.

But I agree at least there’s a chance to do something worthwhile.

Performance exceeded what was available at the time, scaled super well, but not something I’d ever go back to.

- xslt is not a programming language. As soon as you start using xslt:functions you’re doomed.

- it is not a fit for interactive UIs. It’s good for static content, trying to mix in js goes wrong very quickly.

My opinion is that these days infra is cheap enough that you get all the same benefits by doing standard templating in your language of choice from your data in json or whatever. Which I know is essentially the same thing without the specialist software and dsl. And that’s kind of the point.

> Limit our brain diet the same way we try not to eat (delicious) fatty food every day to avoid the total collapse of our bodies.

Gives me hope that after our societal binge on social media and genai, we will come up with healthy patterns.

It only took us a hundred years or so after the Industrial Revolution to start adapting habits after the relative plenty that resulted.

I can see llms going much the same way.

IMO There’s a market expectation that “big tech must make cuts” and therefore they either feel their share price would be under pressure if they don’t show something, or more cynically they see an opportunity to trim a little that would look a lot less worse than if they did it another time.

It’s easier to manage when all you do is assign work, reward with money and blame silent quiting, millennials, or whatever the latest cause of people “just not working enough”.

As opposed to actually caring about your staff, making sure the work is fulfilling and that the team are bought into what they are doing at an individual level.

Leading shouldn’t be about what makes it easier for the manager.

Even while it won’t break anything you want CI to be your safety net, flagging a local setup as being wrong is more valuable than magically autocorrecting it.

From my take, an interesting way of dealing with scalable workflows and batch jobs. But I need to read more into it.