Tom MacWright:

The churn in the ‘state of the art’ feels tremendous: first we had tool-calling, then MCPs, then tool calling writing code to call MCPs: it’s hard to tell if this is fast progress or just churn.

Tom MacWright nails it: LLM results don’t work for you, they work for their makers.

You ask OpenAI for a product recommendation, and it recommends a product that they’re associated with, or one that a company is paying them to promote. Or maybe some company detects OpenAI’s web scraper and delivers customized content to win the recommendation. You just don’t know.

This is obviously going to happen. Google promoted its own products in search. Amazon recommends its own products, eagerly ripping off the branding and terms used by other companies. Microsoft promotes its own AI, Copilot, when you use Microsoft’s search engine, Bing, to search for Google’s AI, Gemini. This kind of stuff is not illegal enough to attract enforcement in the US and it’s obviously good for business, so companies do it with gusto, even when it’s totally obvious to everyone.

So in this new world of “AX” — Agent Experience — it’s worth asking: agents for whom?

LLMs are a “lubricant for the crushing weight of complexity”.

Say what you will about the mountains of technical debt at your average startup, but that technical debt was hard-won, written by hand over the course of years. With modern tools, we can greatly accelerate that process.

You thought npm made it easy to quickly pile on complexity? Let’s see where code-assistant LLMs get us.

Tom also poses a good question: where does all this accelerated productivity take us?

Here’s my question for companies using LLMs: when will I see your productivity gains and cost savings reflected in my bill?

I’m going to keep writing and making things because it brings me joy, and I might as well find some way to do so without grumpiness.

I think the honest answer is that most people can’t gain perspective and moderation and maturity by reading someone’s advice online. The wise 35-year old dads on Twitter can follow their own advice about work-life boundaries because they’ve suffered the consequences. There’s no shortcut to perspective: you have to acquire it by experiencing bad things and suffering consequences.

There’s no shortcut around experience — tech will never solve that.

Advice on the age-old debate about over working:

If you can figure out the difference between busy-work that only benefits your employer, and the kind of work that makes you as a person feel like you’re making progress and becoming more skilled, then you’re ready to learn.

The perspective being: it’s not about work or not work. It’s about cultivating your enthusiasm and following your curiosity. If you’re doing that, it’s not work. And you take the results with you anywhere you go, at work or not.

As my aunt Grace, who lived in the Ozarks, put it, “I get what I want, but I know what to want.” - The Joy of Being a Woman in Her 70s

When you’re advertising or talking about a company, naming competitors is as much about choosing as it is about observing. If you want to make your product sound valuable, you compare it to a competitor with an expensive product, not a free one.

The reality is that most people don’t have the time or need to understand the differences between different companies and products. If the first thing they try works, then great - even if it isn’t the perfect, most efficient way to do it. This is, after all, what advertising is: a way of telling people that some product is the thing that they need.

This article is the detailed, technical overview of Mastodon I was looking for.

What I built isn’t an ActivityPub system as much as a Mastodon-compatible one. I think this is the key contradiction of the ActivityPub system: it’s a specification broad enough to encompass many different services, but ends up being too general to be useful by itself.

The contrast of ActivityPub to RSS is pretty stark. Read this and, damn, you gotta love RSS!

• You can implement an RSS feed with basically any system. A static site generated by a static site generator like Jekyll? Sure! You can even write an RSS feed by hand and upload it with FTP if you want.
• Your RSS feed doesn’t know who’s reading it. If you have 1 million people subscribed, sure, that’s fine. At most you’ll need to use caching or a CDN to help the server serve those requests, but they’re just GET requests, the simplest possible kind of internet.
• RSS has obvious points of optimization. If 10,000 people subscribe to my RSS feed but 5,000 of them are using Feedbin, those 5,000 can share the same GET request that Feedbin makes to pull the latest posts.
• An RSS feed reader only needs a list of feed URLs and an XML parser. It doesn’t need to have its own domain name or identity in the system. A feed reader can be a command-line script or a desktop application.

While I didn’t agree with necessarily everything in this piece, I really loved the way it started:

The web’s evolution over the last decade has mirrored the American economy. All of the essential indicators are going “up and to the right,” a steady stream of fundamental advances reassure use that there “is progress,” but the actual experience and effects for individuals stagnates or regresses.

Nothing helps you think you’re on the right path like seeing a graph that goes up and to the right representing any aspect of the thing you’re doing.