Things are changing a lot. Huge flow of ideas, and some catching up to do. Mind bombs in every direction.
Last night while watching sports I learned via ChatGPT about MCP.
Here's what it can do and people *are* using it for this
You could turn ChatGPT into an easy editor for WordPress posts.
Just as I have developed the habit of getting it to create a handoff.md file when I'm done with a session, I could write something with ChatGPT helping, I don't ever do that myself but i might, if it were easy. and when I'm ready to publish, I'd say "Please publish this on my daveverse site now." I might specify a category or two, or set defaults, it's good at that stuff. I've taught Claude to write code in my style, so I can maintain it (to answer Aral Balkan's question on Mastodon).
# Fri, 03 Apr 2026 13:31:20 GMT ## Little hierarchies everywhere -We create little hierarchies everywhere we go.
So many places. I have no room for new ones, yet I have to make room because there are people there I want to work with. Now I have to manage it.
If an alien came to Earth and asked why we don't just create a way for a little hierarchy in one place to appear where ever you want it.
It's not out of reach, it would take two or three developers with enough imaginative users to get the ball rolling.
Write down the features you'd have to support, concisely and simply, and provide conventions for making those hierarchies accessible through a very simple format, in JSON or XML or anything isomorphic, and then we start building.
And start releasing apps that work together. That's what I want to do.
WordLand is supposed to be the first such app. But maybe I need to go even simpler for example code. Thinking about it.
This is something we can and should research.
Let's give one of the ai apps a fairly good idea for an app we want to use, and help it -- not by coding, just by answering questions about how it will work, and Iterating over the product until it works like we want it. Sometihng simple, like perhaps a text editor for Mastodon. Something that isn't squished in a tiny little text box, and has icons for bold, underline, links, etc. It could be useful.
Then let's look at the code with an open mind. I think i've given it enough examples of good maintainable code that I could get it to produce maintainable code.
This was in reply to a Mastodon post by Aral Balkan.
# Thu, 02 Apr 2026 14:46:36 GMT ## Please follow me at my new Twitter address: bullmancuso. Whatever anyone thinks of the company the product is still unique, there are people and communities there that I need to communicate with, and I just don't have that kind of network anywhere else. # Thu, 02 Apr 2026 14:57:25 GMT ## My first real post in the New Dave On Twitter, or N-DOT. # Thu, 02 Apr 2026 13:16:36 GMT ## Yesterday I wrote about AI introducing doubt with something as fundamental as how software is created now with the advent of AI software that can be used effectively to write software. Behind that I wondered if the open source developers of WordPress had changed their methodology? Is their codebase managed by ChatGPT now or Claude.ai? Not only did I get the answer to that question overnight (yes, they have made the change), but there was an announcement of a new WordPress competitor, something that hasn't come along in decades, actually. It's called EmDash from CloudFlare. I read their announcement, and then asked ChatGPT to walk through an analysis of it with me. Here's a link to the conversation, hope you can read it. It understood my concerns. Is this something that can work with my product WordLand. Short answer: No, not as-is. It apparently doesn't support the wpcom api what we use to connect to WordPress. By design, you can import WordPress sites into EmDash, but they don't interop with each other. It's for moments like this that I have my WordPress news FeedLand flow. Already there has been some analysis. No doubt anything written today is going to see sketchy in the days to come, first impressions don't usually end up meaning much, even so I'm anxious to read what other people think. Meanwhile I'm thinking that maybe I should shift gears back to working on FeedLand, thinking that the WordPress world is too shaky now to try to introduce something new there. Likelihood of success is decreasing every day it seems. # Thu, 02 Apr 2026 13:31:53 GMT ## Continuing, isn't it a shame that CloudFlare didn't take a different approach? What if they had created a fantastic WordPress runtime, which seems to be where most of their effort went, and that's where their expertise lies, not in crafting new user experiences. A service you could buy from CloudFlare, along with all the other services, that does a fantastic job of running WordPress sites. The customer wouldn't need to know how it worked behind the scenes. Yes, that would still be competiting with existing WordPress vendors, they make money off runtimes, but for the users it would mean they could keep using WordPress the way they always have, and the result would run better. That they didn't do it this way, that's it's all-or-nothing, might turn out to be the reason the product doesn't take off. It's a serious consideration. On the other hand there probably are a few WordPress users that would like to try something new out, esp if the cost of conversion is near zero (which they kind of claim it is). # Thu, 02 Apr 2026 13:38:11 GMT ## When I think of "Slack" my brain immediately translates it to "AOL." I'm not kidding. # Wed, 01 Apr 2026 13:49:26 GMT ## BTW, suggestion to web-based companies that send out notices via email. It's good to do that, but make sure somewhere there's a link to exactly the same material on the web. It can only build traffic for your ideas, earned media. # Wed, 01 Apr 2026 13:18:55 GMT ## Archived Scripting News OPML source for March 2026. # Wed, 01 Apr 2026 13:58:29 GMT ## The fog of tech -Got an email from Automattic about MCP support in WordPress, which is now available on their servers. With this new interface you can write prompts in Claude etc that do things in your WordPress workspace. Kind of like a scripting language, but English, like this -- "In WordPress, please set the category for the current post to Project 32."
I guess it's very much like the wpcom api we're using for WordLand. It's going to be harder to get people to look at wpcom with this kind of functionality out there. It was always going to be hard, but I liked the challenge of telling a story about a great bit of technology that could save the web but wasn't known to almost all developers. WordPress never attracted the kind of devs that care about APIs like that one, ones which would let you build on WordPress as opposed to in WordPress.
Tech is always foggy and full of hype, but rarely is it as intense as it is in 2026. AI is the major thing people are talking and thinking about, trying to figure out if there's a way to be part of the fun with our software and ideas. And there are so many quick ways to get hooked up to the hype, that seem pretty desperate, the kind of ideas that emerge from management offsites in orgs that have little sense of direction -- "let's add AI" everyone agrees, without any idea of what that means, and not much comes of it. Firefox, the perennial hype-harvester very predictably did this late last year. No we don't need another browser with AI. You have to think harder and more creatively. My advice was to be better for the web, and eventually if there is a link to AI it will reveal itself. But you have to pay attention for that.
As revolutionary as AI is, some things aren't going to be done with prompts, pretty sure of that. It 's a lot easier to pick categories from a dialog than typing an instruction in ChatGPT. Think about how you drive a car, you don't slowly tell the car to "turn the wheel left and tap the brake, now right, and hit the gas." Maybe this will turn out to be like the difference between using a mouse or a keyboard. Some people thought keyboards were obsolete when the Mac came out in 1984. I'm using a keyboard right now.
I'm going to finish the new WordLand and ask some people I want to connect with to try it out. The goal is to create a new kind of structure for the web, made out of posts that both stand alone and are part of a graph that you can walk around in. Far more spontaneous than web rings of the early web, like my blogroll does so much more than the static blogrolls of the 90s and 00s. But it is going to be hard to get attention for it, in the midst of all that's going on with AI.
On the other hand, I haven't seen the AI tools get into social structures, I feel very much alone with my AI collaborator. I know there are ways to set up collaboration, but that hasn't reached me yet, and at this time I'm not actually receptive to the idea. I haven't yet seen how we can plug away together human to human.
Like everyone else we're feeling my way around this, looking for ways to add value, and at the same time help to revive the web, which definitely needs help.
I'd like the web to make the transition to AI, not to become even more forgotten. I feel like this is the last chance, I want to get the web hooked into AI, but I have to work with other people, going it alone won't work.
Just some random thoughts on a Wednesday morning, having absolutely nothing to do with the fact that it's freaking April 1.
# Wed, 01 Apr 2026 14:07:15 GMT ## What April 1 means here -There's so much bullshit, why deliberately add more -- in hope of being either funny or memorable -- and only succede at annoying.
