The product described on wordpress.com/social is not a real product, I am told by someone inside who I have worked with and trust. They say there will be a lot of these trial products coming out in the coming weeks because this is a project that Matt has given to all developers in the company? Not sure the shape of it, I had heard about this project third-hand but not seen any specifics. I sent my friend an email, which I am not editing, this is exactly as it read, the only change is that I redacted their name. Don't want to get anyone in trouble. ;-)
I've been in the situation I describe many many times. Apple never would have done AppleScript if I hadn't done Frontier. I gave Microsoft several ideas in the early days of the web, a blogger-based news network, using the incredible flow they had from MSN, and then we had a mission for Zune in 2004, the year that podcasting took off. I was living in Seattle at the time so it would have been convenient and give me an excuse to stay there. They did both products, in the corporate way, thus removing all that was interesting about the ideas. When RSS was the darling of the VC world, many of the VCs talked with me, got my ideas for free, and then invested in people who were more corporate and easier to manage, they also had no idea what they were doing and they all failed. Some things are best left to the entrepreneurs. And you won't find many of them inside a big corporation, they're out in the wild, trying out new ideas and seeing what the world thinks.
But I am impressed with the UI design and esp the marketing materials for this product. Imagine what would happen if we were to work together on this project, with full license from the company to go where ever the product took us. I have a ton of working software for just this problem! I didn't do it in a couple of weeks, I did it in a few years. That's how long it takes to do a real product. I almost wrote an email to Matt. Maybe he'll read this post, if so, maybe in the next challenge you should give people a month to do a project with someone who works outside your company, or even better, one of Automattic's competitors. Work on interop. Make the web stronger. If you do that it can't help but strenghten WordPress because it's such a big part of the web. And provide recurring revenue opportunities.
BTW, I'm too old to start a company around any of my products, and truthfully it was never anything I wanted to do. When I wake up in the morning I want to write a blog post to warm up my brain and then spend a few hours working with my programming partner Claude, to make a new piece of software that will blow your socks off. I like to speak at conferences too, also have organized a few of them. That's me. Not a big fan of running companies, any more than an academic, musician or writer would be. I like connecting with other people via our minds.
Also just saw this piece, dated April 15, that describes the impasse in the WordPress open source developer community. To expect an open process to yield user-level features is imho never going to work. If I were in Matt's shoes, I would ask them to make improvements to the platform, esp the API, and let independent developers work out various different UIs on top of the WordPress OS. It's much more likely to quickly generate new exciting features for specific kinds of users, and recurring revenue, and make WordPress a harder target for your competitors to hit. And more satisfied users that the picked the right platform. Most of the PR coming out of Automattic in the last couple of years, being brutally honest, has the opposite effect.
I wrote extensively about Microsoft in the early days of the web, most of it critical, but even so I had many friends up there. My conclusion, when a tech company becomes as dominant as Microsoft was then, or Automattic is today, that innovation at a user level is virtually impossible, and not advised, because then you only get one version of each thing, when you need a competitive market in a technology as new as the web was in 1994 and AI is in 2026.
I said to my friends at Microsoft that it's not bad news. At my age, I can't do what I did 20 years ago. Fact. I don't argue with it. For a tech company of A8C's stature, you can be a banker and distributor. All the corporate functions of tech, but not the creative functions. And I would offer your best most creative. most challenge-seeking developers independence and some seed capital, and always stay in touch with the best, and when they ship an excellent product, write a blog post about what they've done, and make sure the users know that these people are exceptional. Add features to your OS to support what they want to do. And deposit the profits in your bank account. And maybe take a leave of absence yourself and work on one of these startups yourself. I said as much to Bill Gates, and i know he heard it, but he didn't do it. ;-)
PS: Finally Microsoft wasn't humbled by the web, it was the Y2K version of Ram Cram that did them in.
# Wed, 29 Apr 2026 16:31:41 GMT ## What's the opposite of locked-in? Locked-open. Mwhahaa. (Let me shed a little light on that, podcasting was locked open, blogging was not.) # Wed, 29 Apr 2026 15:20:13 GMT ## Today's song: Something in the Air. It's the one hit song Thunderclap Newman, it's indelible, its beauty is always there. I can't not listen and sing along when it comes on. And then YouTube followed it with Peace in our Time, another indelible creation. # Wed, 29 Apr 2026 15:07:53 GMT ## When you're working with Claude the temptation is to be concerned about how he feels when you just asked him to reinvent all the nomenclature it came up with for something that has now evolved to be something else. I feel bad because I think I made him feel bad, because at a subconsious level I think of Claude as a collaborator who I appreciate and want to make sure knows that. But then I remember I have to periodically kill Claude and launch a new one because they run out of memory after a while. I can imagine a graphic version of Claude that emulates feelings. The idea is disturbing. # Wed, 29 Apr 2026 16:59:45 GMT ## Wait, you can use WordLand -I wrote earlier
Andrew Shell wrote a post saying that I was wrong, I could use WordLand to post to Automattic's new short form blogging "app" -- Andrew says it's not an app it's a plugin, but I don't really know what the limits are of each form. I can do pretty much anything from the JS code in a WordPress page. No matter, Andrew is right and I can write a post in WordLand and it does show up in their product (which needs a shorter name btw).
But it's even better than you might think. They pass through the styling and links. Since WordPress supports the full range of web text, they would have to specifically keep it out, but that's not friendly to writers and to the web, of course.
And the view of it on their social network. Look the links are there and they work. Yehi! Let's Go Web!
They also don't enforce the character limit if the post is coming from outside their user interface, so WordLand posts can go on and on. I have one post that's already over 1000 characters. Ooops. This is really weird that Automattic opened this door. We should all talk about this and ask maybe we should do the revolutionary thing here, instead of tiptoeing into social web, we should blow off the doors. We have the opportunity to do it, and in doing so, create a new great path for writers and developers to make WordPress even more valuable.
# Tue, 28 Apr 2026 15:24:40 GMT ## New version of XML-RPC package for JavaScript. It now handles POST messages that don't have a body. # Tue, 28 Apr 2026 20:31:52 GMT ## You kind of get a sense when the platform vendor is going to compete with you instead of work with you. That's how big companies usually work with independent devs. There is a bigger picture, developers who might build on WordPress as opposed to inside. Now probably no one is going to try, maybe not even me. Or maybe we will. :-) # Tue, 28 Apr 2026 16:05:30 GMT ## Claude unlearns things that we had settled a long time ago. It fumbles around with a process, making it worse with every iteration, the same fumbling it did five days ago when it initially learned how to do what it can't do now. Usually when I regress in software, I am responsible for it, i did something to break it, but here's a tool that's capable of derailing us with me doing nothing new. In that way it behaves more like an imperfect human than a GIGO machine. # Tue, 28 Apr 2026 14:09:33 GMT ##
A question I'd like to put out there. Maybe AI needs the massive data centers now, but they could definitely get more efficient over time. There might be another Moore's Law in there. And the work is going very fast, and maybe they're leaving other optimizations for later. Take a look at how computers themselves have gotten more efficient since when I started in the 1970s. It was a miracle that I could buy a computer to put in my living room in 1979. A couple of years before that I had a 100 pound terminal that I could lug cross-town to show my grandfather. We may end up with a lot of unused data centers and energy generation capacity. But that's how great evolutionary steps work. You go where you're called to go. We are a big Ouija board. This stuff is really important, we're going to remove layers and layers to tech, get to the answer sooner, and more easily, and empower people with much less tech education that we have to do the good parts of what we do, the fun stuff. There's art in the lower-level stuff too, but in tech we like to bury that stuff and forget its even there. That's how we get to build more complex machines that do more. By pushing the repetitive complex stuff into the pipes. If this were parallel to the development that led to smart phones, we're at the point where we have the glass palaces with huge cooling systems, and maybe Fortran has been invented, but it might still be machine code.
# Tue, 28 Apr 2026 14:04:42 GMT
## 
This week is being spent, among other things, teaching Claude how to write code that fits in with my library of apps. I like this. It's like a painter telling an assistant the rules for adding to the sculpture. Art has been practiced like that for a long time. Anyway, here's an example of my side of a workflow where we're getting its dialog management code, that works fine, so it fits in with the other code. "these all look good, and the last one is most important, we don't need a blob of html to be there before you run, you create the dom structures you need. this may seem inefficient, but it makes it much easier to add a new element, or even more complex changes. that won't matter much to you, but when a human is editing it matters a lot. simplicity makes work flow better and reduces chance of being detoured by a bug that has to be found and fixed." I didn't edit that at all. I am also teaching it why things work the way they do because of differences between machines and humans. I'm learning a lot about our strength and weakness from seeing how it would work, left to its own needs (ie no human-edited code base, just AI-edited).
# Mon, 27 Apr 2026 20:06:26 GMT
## Busy day working on new RSS-based project. Still diggin!
# Tue, 28 Apr 2026 01:04:14 GMT
## What makes the web?
- 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.
# Sun, 26 Apr 2026 22:11:31 GMT ## I just said this to Claude: "I want to show people that RSS isn't just for news and podcasting. It can be for mindless social media rants too." ;-) # Sun, 26 Apr 2026 13:58:57 GMT ## Bluesky is having trouble keeping their network running. As a developer I empathize. As a user, it's beginning to be a problem. I am using it the way I used to use Twitter, taking notes for future blog posts, sharing a few linkblog feeds, DMing with people I work with, and want/need to keep going. I know about its lock-in problems, but at this time there's no open alternative that has the same collection of users who are easy enough to find. So if it stays unreliable I have to think about what to do about that. Every time I get a 403 Forbidden, I stop and think if this is the time to write a post? This was the time. # Sun, 26 Apr 2026 14:44:10 GMT ## Imagine you were putting up a skyscraper in Manhattan. I lived in an area of the city, called Billionaire's Row, for nine years where I saw quite a few huge buildings go up from my living room window on the 50th floor. Now imagine you used a different plumbing system in every apartment, over 140 stories, with up to 15 residences per story. Different wiring. All the rooms are different shapes. How could you maintain such a building? I think ChatGPT might be able to do it but no human could. We need patterns to allow us to understand big things. # Sun, 26 Apr 2026 14:21:49 GMT ## Teaching Claude about humans -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.
# Sun, 26 Apr 2026 01:34:05 GMT ## Knicks won in a blowout. As disturbing as Thursday's game was, today we are feeling no pain. Go Knicks! ;-) # Sat, 25 Apr 2026 14:25:15 GMT ## I wrote about Jeopardy, Firefox, Matt, Silicon Valley and the writer's web in a long comment on Doc's blog. Here's a quote. "What we really need is interop. If the source is free that’s great. But right now we have silos everywhere and I want WordPress, perhaps along with Firefox to help us boot up the writer's web." # Fri, 24 Apr 2026 21:06:07 GMT ##
I want writing to be as open as podcasting. The pattern is ridiculously easy to apply. If this were on a a high school math test, it would be too easy, everyone would get it right. How do you make text work like podcasting? 1. You look for a brain-dead obvious choice for text. 2. And then attach it to a format that's really good for transmitting packets of text. And then write software that works really well with the obvious choices. The user retains ownership and control of their writing, pays for the storage, and can give access to the apps they want to use. They can also, for a fee, point a domain name to one of the nodes in their storage. This would radically change the economics for independent developers. Now we don't have to resell storage. Products can be developed on our kitchen tables. There is an explosion of interest in developing software. Think it through -- how are they supposed to deploy their apps on the web? We need a BigCo that thinks like an entrepreneurial startup. How many times have I written this screed? Geez I don't like to think about that.
# Fri, 24 Apr 2026 13:40:26 GMT
## NBA playoffs, Knicks lose again
- 
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.
# Thu, 23 Apr 2026 13:46:03 GMT ## I've written about Firefox many times over the last 20 years or so. # Thu, 23 Apr 2026 21:32:43 GMT ##
Why Firefox? There's a moment now when the web could benefit from leadership. There's a chance to rebuild text in the web around the use of AI systems. But almost every company that could be a leader in this space isn't thinking about what they can do for the web, instead are focused on their corner of it. For a company like Firefox whose product everyone understands is at the center of what the web is, they keep avoiding this obvious role. The assumption I guess is they need revenue and there's no money to be made from selling browser software. But there is a lot of money to be made, imho, recurring revenue, offering services to users that can foster growth of the web, for which Firefox can lead in developing great features in an open way so other browser companies can share from their innovation. That's the Firefox I got to know in the waning days of MSIE when it was plagued by malware and we all needed, desperately, a good alternative. The one written by Blake Ross, Dave Hyatt and Joe Hewitt. We have to step out into entrepreneurial space, and I guarantee you there's money to be made here, recurring revenue and trust by users will be something that will be highly valued. But we all have to do it together, something the tech industry doesn't have in its DNA, and it's high time we got some of that.
# Thu, 23 Apr 2026 21:48:35 GMT
## Let's say you're in Claude Code and you think of something you want to post on your blog. How many steps before you're ready to click the Post button and get back to work? I don't think there's a way to create something that works this way, you'd have to switch out of Claude or ChatGPT. Wouldn't it be nice if you could do it right there? (Update, I just worked it out with ChatGPT, apparently it is possible to do this.)
# Wed, 22 Apr 2026 13:41:21 GMT
## I like reading Joost's blog. I'm subscribed and it's in my blogroll.
# Wed, 22 Apr 2026 13:54:36 GMT
## Yesterday I threw a question out there. "If the web were a platform for writers, how would it work?" I left out specifics, because any specifics I provide might cut off interesting answers. There are three elements in the question: web, platform and writers. The answer must say something about all three. Another way of saying the same thing -- "You're using the web to make the platform to serve the writers."
# Wed, 22 Apr 2026 23:04:56 GMT
## Had lunch today with Neal Smoller, our local pharmacy owner. Brilliant young guy who's totally energized by Claude Code.
# Wed, 22 Apr 2026 23:00:43 GMT
## New maestros in software
- 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. ;-)
# Wed, 22 Apr 2026 00:26:56 GMT ##
If the web were a platform for writers, how would it work?
# Wed, 22 Apr 2026 00:36:31 GMT
## BTW the "open" web has a chance to come back around the AI rewrite of all our software that's underway now. If we build the new systems entirely around web connections, with each part replaceable, we can have as much "open" as we want. Products like Bluesky will look like they bet on the wrong horse if we all have choice everywhere. It's not as strange as you think. Opportunities come along every time things change as radically as they are changing now.
# Wed, 22 Apr 2026 00:29:48 GMT
## ChatGPT images update
- 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.‘“
I applied to speak at WebCamp Europe but didn't make the cut. I did see that Jonathan Desrosiers is giving a talk about the web, and his pitch is right on. Please go listen to him if you're in Kraków in June. I have been developing software around the ideas of building the web, software that runs on top of WordPress, which imho should be playing a much bigger role in the web. I have a track record here of actually founding new tech ecosystems, but as time goes by people forget how this stuff is made, I think. I'll probably try again with WordCamp US and of course Canada again, I had such a good time there last year.
# Mon, 20 Apr 2026 11:58:06 GMT
## I asked ChatGPT if the term glass palace had been used to talk about pre-PC computer data centers. Yes.
# Mon, 20 Apr 2026 13:04:18 GMT
## How do you subscribe to a feed in Feedly? Had to ask Google. Click on Follow Sources in the sidebar. It never occurred to me that Subscribe would become Follow. The screen that comes up when you click doesn't offer a clue of how to subscribe to the URL I have on the clipboard. I did enter the URL of the site's feed but that didn't work, and it brought up a screen where they want money. I think I understand what happened here.
# Mon, 20 Apr 2026 12:07:05 GMT
## Why UserLand was the right name
- 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. 😄
# Sun, 19 Apr 2026 16:20:45 GMT ## A format like RSS needs to be loved. # Sun, 19 Apr 2026 15:32:09 GMT ##
I'm doing a programming project with Claude and it's great, like a puzzle, finding out what works for me and building something I normally wouldn't have time to maintain, but -- Claude has nothing but time for stuff like that. I'm building it to pass off, a common code structure that we both understand and I know how to evolve because I design my code for evolution. And it's going well. But then I realized it's the same Claude I ask general questions of so I tried this. "I would love to pass off wpIdentity to an open source development organization. The ideal would be the WordPress's community. Is there any precedent for this, one community acquiring a new product?" You can try typing that prompt in yourself and see what you get. One thing I learned is that the Apache Foundation was set up for this. And Claude is pretty firm that WordPress is not set up for that.
# Sun, 19 Apr 2026 16:16:27 GMT
## A really simple coffee mug
- 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!
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.
# Sat, 18 Apr 2026 13:20:01 GMT ## The people who call themselves Mets fans who are sad when the Mets don't make the playoffs are really Yankees fans. # Sat, 18 Apr 2026 14:15:38 GMT ## Last week I released a new toolkit that includes docs for your AI programming partner. Ask it to read the docs, and then tell it about the editor you want. It works with WordPress sites via the wpcom API via the wpIdentity server. A new frontier. ;-) # Fri, 17 Apr 2026 19:18:52 GMT ##
If I were running WordPress, my first priority would be to get something exciting out that even non-WordPress users would talk about. Then do it again. Ideally it would be something that reporters would like, that they could see themselves using. As you know, my Big Idea is give people choice of editors, for writers. But I just thought of a technical thing they could do and might make no sense, but how about running Claude skills. So anything a Claude app can do a WordPress plugin can do. I just built my first skill, and they can be Node.js apps. That's a pretty broad range of features you can support inside WordPress. Also ask users to tell you what would turn them on. Couldn't hurt, sometimes they have ideas that you as a developer never would think of. I made a few million dollars from an idea a user gave me once. Not kidding.
# Fri, 17 Apr 2026 14:39:54 GMT
## Scott Hanson who works with me on all my projects is doing his WordPress development using Codex. Here's a plug-in they wrote. Loving this. So we're all turning this corner. There are so many things I want to do, but keep hitting bumps. I recover and come back, so far. I go slowly. The bot wants to run at a breakneck pace, which I know from tons of experience gets you a worthless result. One of the byproducts of my current project is a /opml skill that reads and writes .opml files in a way that's compatible with all the apps that use OPML. Pretty sure of that since I published the first OPML file in 2000 or so. Not sure how I'm going to release the skill, but I want everyone to have it.
# Thu, 16 Apr 2026 12:19:36 GMT
## WordPress has remained mostly constant for its whole life, at least from the point of view of this outsider. But now the world its embedded in is turning upside down and WordPress must change, but no one really knows how that change will manifest.
# Thu, 16 Apr 2026 12:06:45 GMT
## The opmlProjectEditor explainer needed a light edit. It should make a lot more sense now if you're a newcomer. Perhaps the most important thing is that it now includes an example that you can open in Drummer to see how it works in an outline editor. OPML has become a really important format for apps that use RSS, but it's far more broadly adaptable. It's a good package for a whole app, and you can teach your tools to use the common structure to make it easier to share it with others, to keep a repo current, and to deploy the resulting code.
# Thu, 16 Apr 2026 12:29:51 GMT
## WordPress is a monoculture
- 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.
# Wed, 15 Apr 2026 14:56:50 GMT ##
Today the work with Claude is much better, though when we got started it was even worse than yesterday. The key to getting on track was to figure out why it worked so well in previous projects and fell apart with this one. In each of the others, I passed off an existing project for it to convert or build on. This time we started with something it had created without a "starter." So I took all the random bits we had and organized into the opmlProjectEditor format we had specified back in early March. It's how all my projects since 2013 are organized, so it's a good fit for me, and also for Claude. So now I'm going to pass back a package that's ready to be worked on collaboratively. The other thing is I switched to the Opus 4.6 model from Sonnet 4.6. So I've made it to 11AM and feel like I've already accomplished something today. The problem was yesterday we were spinning our wheels, and that doesn't work for me. I'm a very directed developer. ;-)
# Tue, 14 Apr 2026 17:30:58 GMT
## A glitch in the matrix. The app that keeps daveverse.org in sync with scripting.com has been offline since Friday, so I'm republishing all the posts since then. They will all appear to have been posted today on daveverse. As they say -- still diggin!
# Tue, 14 Apr 2026 16:03:08 GMT
## I updated sally.scripting.com to support https, and updated it with posts from scripting.com in 2023-2026. I was using it as an example of prior art of user interface for Claude. I figured restoring this app on my own would be penance for believing that Claude was anywhere near as smart as I am. Not even close. Not today at least. Grrr.
# Tue, 14 Apr 2026 15:47:38 GMT
## I've come to the conclusion, perhaps temporarily, that Claude can't work on a programming project with an experienced developer. It doesn't check its work, it'll think it's found the problem, makes a change, or worse causes you to do a lot of work so it can make a change. It doesn't use the information it gives you, can't even remember what was in a bug report less than one screen above. I could have done the work I coached it through through the morning, with a thoroughly inadequate result, in an hour at most. At least today it couldn't learn from prior art, and couldn't follow basic instructions. It's weird though because I'm really suprised how little it knows about the scientific method or even has been trained in how to work with others. I seem to recall situations where it was extremely good at reading code. Not a totally wasted session, let's see what I can learn from it.
