I’ve been trying to come up with a simple test that lets you know whether some software is on the web or if it just can be made to appear in a web browser. So here we go.

If you can hook up a piece of the app to the a piece of another app then it’s on the web.

This comes from the basic feature of linking, which is the unique feature of the web.

Every other feature that makes the web the web in my experience allows two things to be part of each other.

Comment here.

What you don't hear about AI is that it doesn't know how human minds work, what our limits are, what we can do that they can't. It has no memory. This was hard to believe, but you have to tell it to keep things in some kind of memory, usually a Markdown file, then tell it to read it from another file. These are things humans just do. I may have to say it two or three times, but you will remember it if you're a normal human.

I know there are places you can leave an instruction to read it when it starts, but sometimes it doesn't do it. It could never, in its current state, figure out how to make a product that people want to use. Nor would a human be able to easily read the code it generates unless you work hard at teaching it how that works, and if you have to do that, you might as well do it yourself.

It never looks for prior art, in some contexts -- but in others, it's encyclopedic about prior art. You might find a function named returnError in one place that takes a string argument, where most of the other instances take an object that contains a string. How do I, as a human, work in an environment like that?

The users who vibe code who think they know how to make code people will want to use are making the same mistake student programmers make, they come out of school only having done student projects, which have very limited objectives. The real world is far more demanding. The real world populated by humans that is. Just some random thoughts as I try to create usable code working with Claude.

Perhaps what's needed is a Developing Better Developers program for pseudo-humans.

After last night's game I now remember why I was so relieved last season when the Knicks were eliminated in the semifinals of the NBA playoffs. It’s an exhausting sport. And the sad truth is the Knicks are getting beaten by Atlanta. Or maybe it's not so sad, because then, after they are eliminated, I can tune into the playoffs with a detached interest, and save my kvetching for the Mets, and there is plenty to complain about there, LGM.

I almost never question a coach's decisions, because they have complicated jobs -- but -- why didn't they put their top defenders, Mitchell Robinson or KAT on the court with 12 seconds left in the fourth quarter with the Knicks ahead by 1. Instead they put in all these fast small players, as if they were planning on giving up a bucket and then quickly running down the court with what little time remained and scoring a quick one to win the freaking game. At least that's what I imagined they were doing. Why not just hold them right there and run out the clock and win the game?? It all happened so fast. (I asked ChatGPT about this theory and it says I'm wrong, they put in the small players because they can switch more easily, and that's likely how Atlanta was planning to defend their shooter.)

The Knicks were out of timeouts, so they couldn't stop the clock. So I guess kind of predictably, esp the way things were going last night, the Hawk with the Hot Hand, CJ McCollum, gets the ball, dribbles a bit, and nails the shot that wins the game and now the Knicks are down 2-1 in a series they were supposed to win handily.

I didn't and don't buy the idea that the Knicks are destined to appear in the finals this year, when you boast about it, or expect it, god has a way of goofing on you, making sure you don't get it. The Knicks don't have it, because as great as Brunson is, he sucks all the energy out of the rest of the team, and when he's having a bad night, there really isn't an alternative. It's not a good configuration. As in the Melo years, Brunson doesn't make it as the captain, imho.

Melo had the size and talent, but he's sweet labrador retriever type player, the sidekick of the engine of the team, somone like LeBron James, Steph Curry or Giannis Antetokounmpo. Look at how successful Melo was in the Olympics, where he could be #2 to LeBron's #1.

It's not just about skill and hard work, it's about does the team follow you. That isn't something you can learn, you either have it or you don't. And of course I will torture myself in this mode, wondering in vain how my Knicks will fare in this quarter, or that game -- until I get to retire from the NBA for the year, and instead get a big fat bellyache on about the Mets.

I wonder how many people are working on clones of existing software with an eye toward making a much more evolvable and customizable version with AI at the core of the model.

You can make the same software easily, with Claude's help, and if you think about the things users want to customize, you can give them a toolkit for doing exactly what they want in prompts, as opposed to code, plugins, etc.

So you don't vibe-code it, you start with an app that's designed to be beautiful on the inside, easy to understand for a new maestro of software, but something they can evolve with prompts so they can be working on something else intently.

We provide beautiful code for aspiring symphonists to learn from.

I remember when I first got my hands on the Unix source back in 1978. I was blown away by what was possible. I had largely been a Fortran programmer up till then. The pieces don't fit together so well on their own, I learned, you have to move them into place and for that a lot of trying-things-out has to happen.

Why am I thinking about this? I have friends who are not programmers who are pretty close to where I was then, waiting to see how real software is made. And they can have that experience soon. I love where we are now in tech.

BTW, on its own Claude writes some really shitty code. ;-)

I gave the new ChatGPT image maker a whirl, it's supposed to be more thoughtful and realistic.

The prompt: “A turkey in plastic with supermarket label with text and fine print. The branding is Really Simple brand Turkey. The motto, ‘turkey for simple people.‘“

A really simple brand turkey from ChatGPT.

I'm creating new software in ways I never would have conceived of last year. I see solutions to one of the most significant code management problems we used to have, our inability to remember how our code works, unless the app is very small and they never seem to be. This is why you don't see much inspiration in app updates after a couple of years. Everyone has moved on and no one who remains can figure out how it works. But that has changed now, bigtime.

The next product I develop should be able to continue to evolve much more easily thanks to Claude Code.

"Hey Claude, people want this feature to be optional (or configurable). Here's how that should work."

In a human-based development organization, even if you ran the show, you might wait a very long time and it might never come. With Claude, I can have the new functionality before I know what to do with it.

I called my second company UserLand. The idea was that we'd develop software for users, always be thinking of them, and listening and give them more and more power to shape the way their computers worked. It was what I felt was missing from software in the 80s, a focus on the users creating their own future. So back then we designed the software for them. We were hoping they'd get to implement it too.

Now we're going to try again. 😄

A few months ago, some guys from Netscape came out with a way to get AI to pay called Really Simple Licensing or RSL.

I was really put off by the way they took the name that everyone knows came from me with no help from anyone. Maybe I could have done something about it, instead I found a way to have some fun.

With Really Simple versions of granola, spaghetti, ketchup, cola, ravioli, baby shampoo and books. And it's time for another one!

Introducing the all-new Really Simple Coffee Mug!

BTW, I don't think RSL was a good name. Simplicity is not its main selling point, it's money! It's about making you rich. If I'm not mistaken.

But syndication, in 2002 -- was getting much more complicated than it needed to be. So "Really Simple Syndication" was supposed to be a little funny because it was not like the acronyms that tech usually comes up with.

It didn't pretend to be anything but a syndication format. Complexity is the enemy of good software. You have to work at making something no more than it is. It's a struggle. And you don't add complexity wihtout having a really good reason for doing it.

I've been designing and developing software like WordPress for over thirty years. I have stong opinions about where the product should have gone, but mostly I've not been talking about that, because I don't want to interfere with what Matt is doing.

I've known him since he was a teenager in Silicon Valley, a boy wonder to whom the web has always been there, whereas to people my age, it was a miracle that came along to put down all the dominant BigCo's who made it impossible for individuals to create.

But I've never believed in open source the way Matt does, as I explained last week. I think there needs to be competition in the writer's UI for WordPress, and in all other areas of the user interface. I think that's what it suffers from. There isn't enough diversity. Creativity is crowded into a very small space, plug-ins. Because there's an API that covers the full functionality of product, there's no technical reason it has to be this way. I believe in competition, because it encourages listening. People don't listen to their friends, I've discovered, but people do listen to their competitors.

The community is paralyzed, it can't fix basic problems that have been there forever. Gutenberg was a good idea for a site designer and a not-good approach for writers. But it should always be a choice for writers, if they like Gutenberg. There should be no single recommended editor for WordPress.

Imho, there are ways to navigate this landscape, but it's going to require immediate and radical restructuring.

WordPress is not the last hope of the web and the web is not going to disappear in our lifetimes. Everything is built on it. People who say it's about to disappear are alarmist purveyors of clickbait. You'll still be able to ship apps for the web five and ten years from now. But WordPress is an important part of the web, and I don't mean because it runs a certain percentage of all the sites, which is imho a meaningless stat. It's a uniquely valuable API and an implementation that's debugged and scales and is profitable, and can sustain a large organization supporting it. It's one of those things we could lose, but we'd be much poorer if that happened.

WordPress is unique in the products that came to us from Silicon Valley. It's universally useful and it doesn't lock you in.

If a product like EmDash were to be the successor it claims to be, well there goes all the open stuff, because I don't think they have it in their blood the way Matt and WordPress do.

My conclusion after being a software developer since the early days of Unix and personal computers, and at times being part of the Silicon Valley -- there have to be a variety of UIs for WordPress, where all our work is compatible, regardless of what our tools look like, the approaches for users could be radically different. It's been a monoculture, and imho that's the problem. Break it apart, yet retain the compatibility -- that's the most powerful position possible in tech.

PS: After writing this piece, I looked for early references to Matt on my blog, and came across this piece he wrote in 2006. He totally understood what was going on in RSS land. Here's another Matt post from 2010, which after reading I concluded that he saw WordPress as I saw it and still do, as the rightful heir to the legacy of Twitter. And Matt and team did develop the API he talks about in 2010 as hypothetical, it's there now, ready to lead us out of the darkness.