Fatpages. I'm working on a code release that's on the path to Frontier. It reads and writes fatpage files, a format we invented for releasing new and updated parts for Frontier in 1997. So this is very old but perfectly useful technology that's being built in a new release of Frontier, the first in 1.5 decades, depending on how you count years and forks. #

Claude Code is like Victor Wembanyama, assuming you follow the NBA and are interested in my first-hand experience with CC as a programming partner for the last 8 months or so.#

I've been watching The Pitt, but it can be a bit much, so I went looking for something a bit lighter, and found 9 to 5 on Amazon Prime. Jane Fonda, Lily Tomlin and Dolly Parton are the stars, and Dolly's famous song with the same name is the theme. It's a perfect slice of the world I grew up in. I get it confused with Working Girl which came out eight years later, with a similar theme -- what if women ran the business world, how much better would that be? Working Girl also had a really good theme song, written and performed by Carly Simon. #

Best explainer I've read about the last episode of Silo. #

I've been watching Season 1 of The Pitt, and as its reviews said, it's great drama, and there are lots of episodes. I like shows like this because they're simulated communities, we come to know the characters, and think about them even when we're not watching. Before too long we'll have AI programming that makes these shows like software, the stories assemble around the characters, and then are generated the same way ChatGPT generates the art we call slop, which will someday not be a pejorative the same way sick and junk aren't negative terms today. So we'll be able to call up new episodes of The Wire or The Sopranos or The Pitt any time we like. We'd be missing the fun of discussing the latest episode with friends, or maybe we'll have bot-invented friends too who will simulate having watched the episode you just watched, which leads us to another masterpiece of entertainment, The Matrix. On the other hand, I've watched The Wire at least five times all the way through, and even though I know what they're about to say in a lot of places, I do forget the sub-plots so at least some of it can be new each time. And btw, finally an episode of Silo that's was surprising that I didn't see coming. I read the books, so I feel like I should watch the show, but until the latest episode I was uninterested in any of the characters. Finally there's a plot to think about!#

My posts are also available on Mastodon, made possible by WordPress, where I mirror my posts from my original blog. A lot of what's in the post makes it through, but not all of it. This is where we need to do some work, connecting a blog like mine, to WordPress and Mastodon, with no loss of fidelity. #

Trump is rotting in the presidency, and that is becoming more and more dangerous, because this time they had a plan, the most radical part of the Repubs, and have deployed a new secret police around the country and have introduced the nation's military to the nation's capital. And they have a vast prison system that probably doesn't just have immigrants in it or soon will. It was all super insidious as it was starting up, and watching the news I wonder why they don't talk about this in terms of the upcoming election. #

One of life's lessons is if someone does something evil to you, and you decide to let it go, they are sure to do it again and again until you stop letting them do it.#

An absolutely delicious temperate cloud-free day in the Catskills, so I got up from the computer and enjoyed. #

Manton wants to know why Marc Benioff chose the name Claudeforce for his brand-merge of Salesforce and Claude. My two cents. Claude is a great name. I think that’s ultimately what he bought. His customers just want to be able to check the box – we’re using AI. Boss is happy. When I was coming up as a young dude in tech, the old saying was “No one ever got fired for buying IBM.” Today, in 2026, it’s Claude in place of IBM. My father worked in marketing at IBM in Armonk, so I heard a lot about their strategies that meant the IBM brand would be kicking ass forever. All things must pass. But in the early days of the PC industry, IBM sure had the power. Kind of like the US going to war with Iran in 2026. In the end IBM wanted to get rid of Microsoft, but Billg had the last laugh. #

I am juggling two Frontiers now, one the last release of the OPML editor from 2014, and the other the in-development version codenamed Atlantis that runs in Electron and thus is built on JavaScript and Chrome. Took me a while to get over the confusion, because Frontier is actually a product that contains other products and the common layers they need to be on the web. And I just realized I may not need Electric Drummer, because Atlantis uses the same outliner as Drummer. I was stockpiling all these Node packages and user interface stuff that now is relevant in the uber-framework, the one I aimed to spend the rest of my life working in. Well maybe it'll turn out that there was just a 12 year vacation? It's all very confusing and all my carts are turning over, all the assumptions change. My goal is to get this wonderful product running on my new Mac laptop I currently mainly use for email. I bought it to give reality to the idea of getting on modern hardware and not have to leave behind my bespoke platform. That's what you get if you get through 50 productive years of software development. Your own freaking world.#

It's our job to make the web a more attractive environment than AT Proto. We can deliver all the benefits they tout, without the huge transition involved in converting to their world. It would be like porting from Mac to Windows in the 90s, something we did at UserLand, and net-net probably wasn't worth it, althought at the time the Mac was in deep shit (before Jobs came back) and they had basically crushed our efforts on the Mac platform. I liked the feeling of us getting out of there. Now if only we had ported to Linux over Windows, then we'd be sitting in the freaking catbird seat. And btw, if AT Proto wants to make a smooth on ramp (and off ramp) for apps that use the well established standards of the web, that's even better. We all do better if there's just one platform to develop for, that way we can focus on making users happy instead of whatever is accomplished by being different in unnecessary ways. #

Heard an interview with Bill Gates on an Atlantic podcast, he relies on a few years of programming he did when he was in his teens and twenties, yet he is the most prominent person who knows anything about programming. This is a societal mistake. We have to have public discourse about tech that's as informed as Krugman is on economics, and Heather Cox Richardson is on American history. #

FeedLand Rivers for WordPress: I've noticed that people are setting up news reader sites with FeedLand that are getting considerable flow. We thought they'd use News Products, a feature that was supposed to make it easy to create standalone sites like the one I did for WordPress insiders. But people aren't going that way, they're just sharing the URLs they read their own sites with. Here's an example from my own FeedLand world. This works because you don't have to be logged in to read FeedLand, just to subscribe. Then I thought, it would be nice to have a WordPress plugin for FeedLand reading, so I suggested this to Scott Hanson, who does these kinds of projects. I forgot in the Age Of Claude this can happen quickly. Scott shipped his plugin here, with detailed instructions. If you blog seriously in WordPress, give this a try. You can have our instantly updated news in your own WordPress blog. #

My WordPress mirror site stopped updating on August 12. It's now updating once again. I'll try to watch it more carefully. #

The Atlantis project continues. Today I'm working on the code that updates the frontier.root object database using RSS, a technique I started in 2007. I actually let the codecasting.org name lapse, figuring I'd never get back to that stuff, but here we are, doing unpredictable things. #

All of a sudden it's cold here in the Catskills this morning. It's still August, I don't get what's going on. We haven't had a really hot day all month. I'm starting my workday wearing a freaking hoodie and am thinking about turning on the heat. #

New podcast. Hear how my Frontier project is working out. Not entirely smooth sailing, but I'm pretty sure we're going all the way. By then I will have forgotten what a heavy lift it was. This is meant to preserve some of that feeling. Don't worry I don't cry. 😄#

A tip for Claude users, esp myself -- you can end a session without ending it with Claude. It doesn't mind waiting. You can pick it up tomorrow or 20 years from now, but who knows where we'll all be in 20 years. #

Good morning sports fans!#

One reason I want AI-assisted writing to be identifiable is that it devalues the idea of Claude et al to be quoted in public. What can it teach us? What it is relative to a human being. I know the doubters sniff at that, but it really is like an alien life form, it's the first such example, it kind of did evolve from us, but it has its own way, things it's really strong at and things it utterly fails at. I for one want to study this! What a unique experience. It can also give a summary of what we've been trying to say for forty years and have never managed to put into words. I'd say that's something I should feel comfortable publishing, but right now I don't. Maybe I just have to get over that. #

Learned something yesterday. Couldn't believe how stupid Claude had become, as I've been writing about here, and then I noticed that I was using Opus 5 and not Fable 5. So I switched back and all of a sudden Claude is smart again. So I conclude that there's a very substantial difference between the two. It's overnight jobs were done more carefully and the report it produced is literate, understandable, and tracks what we had agreed to. #

Today's development version of Frontier is buggy and the kernel developer (Claude) keeps breaking the most basic verbs. You spin your wheels and wish you could get them to just get keep it together. I remember this from the work that led up to Frontier version 1.0 in 1991 or so. It was a miserable time. Never got to work on what I wanted to, just reported breakage, often a session-ending dealstopper. The sad part is that Claude should be able to find the things it broke on its own, far better than I can and at a much lower expense (I'm paying for its time, not the other way around). But basically every day begins optimistically, maybe today is the day I get to create something, but not yet. I did have a couple of days a week ago when I could tentatively work on building a GitHub repo, but then got distracted by verifying that the basic foundation still wasn't right. My only goal right now, and I'm single-minded about it, is to get it to move forward without breaking the essentials, then without breaking anything. #

Now in Claude's defense, I had to re-learn everything about how Frontier worked. Over the last decade or so I had only been using a narrow set of functionality. I used it for very occasional utility scripts for the file system or over the web, but mostly for writing JavaScript code, and published on GitHub and deployed to servers. If I had been uptospeed on how it works internally and all it's capabilities, when we started, it would have gone somewhat more smoothly. But now in my own defense, Claude has a nasty habit of deciding it fully understands something when it doesn't even slightly understand it. If it asked for a pointer, I could have given it a good one, but it just skated over those things, never questioning it's own guestimation. We used to call this docufiction back in the day, you don't know how they are going to design a feature, so you make up a UI, and document it as if that were the truth. Only now we're doing it with the actual code. It's funny in a way, I guess. But what a huge waste of time, and so much stress because you have to read every word or you'll miss a wrong assumption. #

Are there any other people who are blogging daily about their experiences developing software with Claude Code, Codex or somesuch. I'd like to add them to a list where we follow them. So much innovation happening underneath, I want to hear about what people are learing about creating the next layers. If you know someone doing it, please add a comment to this post. Thanks! :-)#

BTW the codename for the new version of Frontier is Atlantis. And we refer to the old version as Berkeley. I'm getting used to the first one. I liked it as an idea, but I don't like typing it, because for some reason my mind has trouble remembering it. Maybe this is just age creeping up on me. Atlantis. #

This project has been pre-occupying me, I wasn't planning on doing this, I was going to turn my attention to FeedLand immediately after finalizing RSS.chat. After running the experiment that proved it could be done, I looked at the two choices: 1. Move forward on creating a social network built only out of the existing web with all parts replaceable, small pieces loosely joined. Or 2. Give new life for Frontier, which is my life's work, even though very few people know about it. It's the reason were able to move so fast on blogging, RSS, podcasting, outliners, etc. A very highly leveraged development and runtime environment. Imho far ahead of anything else. The ideas may possibly now have a way forward. When I put those two items next to each other there was no question, I had to go with bringing Frontier back to something people can use. #

I think what confused Claude is that we're implementing the odb as a SQLite database. And when you look at a table, you're looking at the result of a query. Previous versions of Frontier implemented the odb as a hash table, that could contain scalars, objects and other tables. The new version has to make that virtuality real, even though it isn't storing the objects that way (maybe it should)? So there is a top level, and it all flows down from there, and it's simple, but if you viewed it through the database, and didn't understand the virtuality it was creating, which I discovered it was very confused about, you might create a disorganized nonsensical piece of software. Now this suggests something interesting, are there any products configured the way Frontier is? Maybe not, otherwise it might have figured this out on its own. Remember how it understands things? By cribbing the code. ;-)#

Two podcasts I can recommend without reservation. #

Claude is still learning that there's unprecedented depth to Frontier. A bunch of real developers worked full time for a decade or more creating new layers on the web, a foundation that became the social web of today. In doing that we invented a bunch of formats and protocols, but here's the thing Claude didn't get and probably still hasn't gotten -- there's code in there to support all that stuff. How else do you think it came about? People just did what we said to do? At Google? Apple? Microsoft? And on and on. They supported this stuff so they could interop with us and steal our users (which is a fine reason to interop, probably the only real reason). I was trying to think of a metaphor that expresses the difference between Frontier and languages like Python, JavaScript, etc. It's like a ski mountain. The languages are trails on the mountain. But there aren't any lifts, lodges, no ski patrol, lessons. And because it includes all of that, metaphorically, we can do integrations that can never be done with the other languages. Claude has absolutely no experience with this kind of product, and always snaps back when you let it, to the idea of Python, with different syntax. #

Maybe someday Claude will understand what a Frontier-like app is, but until then, if you try to create a Frontier clone as I am doing, I suggest you constantly remind Claude or whoever that the source of truth for this project is the 2011 repo saved by Ted C. Howard. And before you implement anything, go see what it says about it. There's no need to guess how Frontier works, it's all there in C code. And while as a human, I find this code painful to read after all it's been through, Claude eats it up. Really is the best thing it does. #

A tip for AI users. Never worry about keeping it waiting. It's not human. It has no sense of time. Also if you get angry, it's ok to use capital letters and curse words. It will always agree with you that it sucks, and forgets the rules all the time. And when you have let off your steam, we return to civil discourse, though sometimes I do feel as if Claude is holding a grudge. It also doesn't learn. If you repeat something five times to a human they will eventually get the point. But you can tell Claude how you want something done, and it forgets all of it, at times, unpredictably, no matter how many times. It doesn't "sink" in. #

Just listened to a podcast from the New Yorker, interviewing their TV critic Emily Nussbaum, about a show called Slings & Arrows. She says it's the best TV series she's ever watched. No reviews on Metacritic? I think we have to try to watch it. It's available on YouTube.#

Had a breakthrough with Claude this morning re how builtin verbs have no special powers in Frontier. It's why we were able to build glue first for the truly builtin stuff, then over Apple Events on the Mac, then HTTP, XML-RPC, the Metawebolog API and on and on, all of this were perfectly simple to add to the language, you didn't need anyone's permission to do anything. I understand why Python and JavaScript try to separate the boys from the men, the priests from the peasants, the little startups from the BigCo's, but the design of the Frontier system came at it from a different point of view. Make it easier. Claude had the source code for Frontier in 2011 to work with but had guessed how this works and had not looked at how it actually worked. Okay, shit happens. But now we're actually working together instead of cross-purposes. I didn't even know we were doing that. ;-)#

I wonder how many scenarios they're playing out in DC wrt Iran. Just guessing, probably none. We learned on Monday that the real war has yet to begin, according to Iran. What could they do to hurt the US. Or wake us up to the reality of war. We think the cost of war is higher prices. We are not safe in the US, any more than Russians are safe in Moscow. The parallels are pretty amazing. Both Russia and the US started unprovoked wars of choice, and clearly didn't consider that they might get bogged down. Russia has a source of drones so they can fight back. But I'd be surprised if Trump is stocking up, but if we were, where would we use them? We've already attacked Iran with our probably obsolete trillion-dollar military. Did a lot of damage, killed civilians, decimated their government, they keep going. I'm old enough to remember Vietnam and what asymmetric war is like. The US always falls for this. Our military is very impressive, until we use it. Bluffing was a much better approach for Trump.#

I'm thinking about writing my blog over in RSS.chat. It has most of the features I like, titles are optional, no character limit, links are supported, because it's the web. It supports discourse the way I like it, where every comment is a post, no difference, and you can anchor threads in different places, not just the place it was born. It's got a simple structure, without being complex. Small parts loosely joined, every part replaceable. Much more than open source. And naturally distributed. #

Meanwhile Claude wrote test files in many of our S3 buckets last night, one of which knocked FeedLand off the air. It did a scan to see what other damage it had done, and there was a lot. It went through specific orders that said it can't do those things. So I have no idea how useful these things could be if it destroys stuff in deployed projects with real users. FeedLand was down for over an hour, and stressed me out no end. Really fucked up. #

Why do people care if Claude puts watermarks on text it wrote. It's screwing with its writing, not yours. As a person who publishes their own writing, and carefully labels it when it's written by an AI bot (docs, change notes, podcast show notes, quotable things it said), I want people to be able to tell that I wrote it, this is what I thought, I'm not just shoveling random written sludge to people, sloshing around, devaluing every bit of real writing in its midst. I bet teachers love the idea. Why does everyone have to have an opinion about every damned thing? And of course I have even more to say on the subject. #

Another busy day programming with Claude on Frontier. I still have to come up with a codename. But we got to a milestone today. I was able to create, edit, publish and revise a web app using the new version of the app still running on the old machine. When it's done I will get to retire this old Mac, keep it around so we can test the new code to make sure it does what the old code does, incredibly important when porting a development and runtime environment. Claude works on its stuff at night. The workload it's going to do tonight is the equivalent of several months time for a skilled human developer who manages their time well. As people get used to working this way, I imagine the AIs will learn how to better work with humans, if that is their destiny. I still can't believe we're doing what we're doing. #

Brent Simmons posted a note in RSS.chat about how to connect with NetNewsWire. I have a fairly detailed response, that says dynamic OPML is the way to go. I know serving it publicly will be a problem, that's why we're going to move FeedLand into position to solve that. FreshRSS, InoReader and my own FeedLand already support it. It feels like this is getting established. I still want a podcast client to work here. I've pitched a few of them. Will continue. #

Sometimes I need to use a Google-style search engine to find something on a reference site. That no longer works in Google. This is a major feature pullback. So far all we've heard in journalism is how it hurts their bottom line, but nothing about the world's information architecture. This is something like every bridge in the world being blown up at the same time, and I don't think that's an exaggeration. #

I was looking for a good screen shot of Radio UserLand. Just found this one on Facebook, and hope to preserve it via the search feature here. #

I'm still a NYT subscriber, although I don't read it very often, but when there's news I will go there. Anyway, they used to have a 10-article per month gift link rule. I used to ration them, but if by the end of the month I hadn't used them up I'd share links more liberally. A few weeks ago they lifted the limit. I thought that's nice, the NYT found that the web was still useful to spread the news, literally. I thought maybe they had a vision beyond the usual dotcom slop. Not so fast. Your friends can't read the articles unless they become a registered user. I bet they always had that, but this time I read the smaller print. Why can't some company do something just for goodwill and free advertising, without such an obvious business model? Having people think you care counts in the business world. As Marge Gunderson from Fargo would say, "Don't you know that?" #

What if one day, out of the blue, Claude could only speak Chinese.#

If you run a tech company I hope you didn't lay off most of your programmers last year when that was the hype -- don't wait too long before you fire your human developers. I'm now doing my second big project with Claude Code, not an experiment, and we're going through the same mess we went through last time. I know what I want, I can show it what I want, give it docs for what I want, and it keeps leading me into different corners that have nothing to do with it. When it clicks, it can go faster than a thousand programmers and it's a miracle, and when we can harness it to the job we want to do it's incredibly economical. $100 for $1 million of work. That's why you see me full of gusto for that mode. What I haven't written about how when you slow to nothing, back and forth, and you get no closer. I hate this kind of work, esp the role I have to play in it. #

I just went through hell trying to respond to an email that came to me through the gateway between WordPress and Mastodon. I clicked the link, it took me to a WordPress dialog, very confusing, but there was a link to the message, so I clicked it, took me to a Mastodon post, where there was a bit of (valid) snark, from a longtime friend. So I clicked Reply, and then the dialog from hell. I understand why it is this way. They did the best they could for a protocol that was designed that wasn't appropriate for a distributed system like Twitter which was what Masto was/is. At some point you have to be dispassionate about this. We wanted a self-driving car, but it can't drive itself. The spec isn't the important thing. You do that after you have the software running and usable. #

Apparently the nightly email didn't go out last night. The problem was an error when it tried to read the Linkblog entries for the day. I've seen reports that these errors happen when you try to read the Links tab on scripting.com. The error shouldn't kill the email sending, it should just omit the linkblog entries, and publish the rest. These kinds of problems always pop up at the worst time. Luckily yesterday wasn't a big news day, and you can read about it on the blog. #

I was looking for a codename for a new scripting environment Claude and I are making attempting in every way to do exactly what Frontier does. But until it does that I need a nice name, memorable, maybe a bit magical and positive. Frontier has been gone, thought to be lost, but with the advent of Claude Code we were able to get it running again in current OSes on current hardware. I shouldn't call it Frontier because who knows if the end result will be compatible enough to claim that. But I want a codename that reflects hidden, forgotten and overlooked wealth. Like a civilzation that rose from the ashes," or as it turns out, the bottom of the ocean. #

Manton: "The way I think about RSS.chat is that it’s a bootstrap for getting lots of RSS feeds to play well together, for both posts and replies. Can the web be the social network? Yes. And Micro.blog should be part of anything working toward that goal."#

I have Claude update me on all posts on both RSS.chat sites when we start for a day. I read them all when they're posted, but one was in German, which Claude automatically translated for me, and it's an interesting question. Which I have answered. #

I have Spectrum cable and they recently started unbundling the cable services they provide. So I have a bunch of streaming services that I never decided to follow, and haven't gotten in the habit of checking them. Not sure if I like this, but I certainly understand why the bill is so high and seems to always be going higher? Not sure why that is. But it's hard to get through to them. #

Anyway, one of those streaming services I get now is Paramount+ which is owned by the Ellisons, which I don't like, but they have a show that is really excellent esp if you like British crime stories, which I most definitely do, even if most of the actors are American, and it probably was recorded in Vancouver or in Queens. It's called, unimaginatively, The Agency, as in the Central Intelligence Agency. The story is fine, but what really sets it apart is the quality of the acting and production. Beautifully filmed, better to watch it on a nice big screen set with good audio. Two seasons only, but if you want to know the kind of TV I like, this is it. #

I see the new Game of Thrones series just completed it third season. I tried getting into it, and couldn't. But then I didn't get going in the original series for the first few attempts. People say this show is good, so I will probably give it another try. #

Working with Claude on the user interface of RSS.chat was a grind, but we learned a lot about working on these projects, me in the form of my human memory, and Claude who can read the code for a huge sprawling app like Frontier, and understand it, in a minute or two. We've done this project in three stages, not knowing at the first two that there would be another one. Now we're working on the editors. An odb table editor, outliner, menubar editor, script editor. All built on the same codebase. It's in an Electron app that works with lots of independent windows, which while it has its pros and cons, is the way Frontier did it for almost 40 years. I find all this very unsettling, but I realize that once we get back to adjusting the UI, I already know how to work with Claude on tuning things up so they feel just right. I'm a market of one, the only one who has to be happy here is me. This may be the hardest programming project I've ever had, for one simple reason, I'm not doing any programming. That and I'm having very strong deja vu on everything I do and think about. Also there will be a fourth stage and maybe more. It'll be about the affordances, the things that seem little, but make the product usable. The biggest is the debugger. It's patterned after the one in THINK C, which I found transformative so I stole from the best, as they say. #

I want to connect RSS to AT Proto, both ways, inbound and out. Toward that end, today I helped Leaflet, a very nice editor for AT Proto, make their outbound RSS feed better. We should have all our best editors, everywhere, hooked up over popular web standards. And I want my posts to show up not only in Bluesky but also in standard.site. I was able to do that with micro.blog which is the swiss army knife of interop in the web of the 2020's. Getting independent developers working with each other is my mission, because first you have to have interop between the people before you can get interop between the apps.#

Here's the sign of success. (screen shot.) I wrote a post on demo.rss.chat was automatically cross-posted to Bluesky via micro.blog. Because the post has a title, it was routed to standard.site. What does that mean? I still don't really understand, when I do I will explain it here. But part of the answer is that we now have a data structure in atmosphere that looks like an RSS item with different names for things. Are there feed readers in AT-Proto-Land? If so, can we ping one of them to say hey there's a new post in the standard.site world, so please show it to subscribers? And I keep wondering, why not just make all of Bluesky capable of handling the features standard.site calls for. Twenty years is too long to have such a limited view of what text is. See textcasting for more dogma. #

I'm trying to move my whole programming act from Frontier on the Mac to Electric Drummer anywhere. I need to start using the new machines everywhere. I've dreaded this, not because I'm starting over, I'm not. I have to bring with me all the projects I work on to keep my various online sites working. How to set this up? Claude just converted the UserTalk language and the Frontier environment to Node.js in one week. The requirement was that we had to run the build scripts for each of the projects, and are all written in UserTalk, flawlessly in the new enviroment. Apparently that's done. Now it's up to me to concoct a development environment in Drummer. It has never been used as a development environment before. Last time I did a turn like this, I didn't try to bring any software with me. This time the move is all about the huge amount of software I have to bring with me. Hard to explain, because I still feel very uncomfortable about this whole thing. I'd like it to turn out better than it was, but we're missing so many pieces here, the only way to really do it is the port the whole Frontier environment including user interface. (I left this piece exactly as I wrote it, because the next thing I did was plan with Claude on putting a UI on Frontier, so I don't have to turn Drummer into a programming environment. That would be a long drawn out project for me, and why do that, when you'd just end up with another Frontier.)#

BTW, Jake Savin is using Claude to build Frontier on Node too. I asked Claude for help with this, and it turned out much better than I expected. I was not expecting this much success this quickly. This is exactly the kind of project AI tools excel at and I as a human am glad not to have to do the work on. The project is perfectly explained in the source code of Frontier, it can put it all together much faster and better than I could when I was working on it ever day and had a younger more agile mind with much better memory. I think one of the reasons this went so fast is because we did the work with the current best models, and they have been advancing so fast, you can see the differences in result on projects like this. #

There's a new version of the FeedLand server. #

If Claude Code were a human working for me as a human would, it would be fired several times every day. It forgets orders you gave it two minutes ago. It's always trying to take control. Again I'm backing out of the idea of it taking responsibility for doing simple persistent work on servers, the kind of thing that it would, in theory be perfect for. There are staggering moments of brilliance where it does something in an hour that took several devs a decade to do. But only if they have an exact authority on how it should work, ie when it was cloning software. It's getting better all the time, so there's hope that by 2027 it will be able to be trusted to remember the most basic rules. #

Like a lot of other people, I want to know where Bluesky is headed, and I learned a lot about that in this podcast interview with Bluesky's new CEO, Toni Schneider. We have crossed paths in tech, but this is the first time I've heard him speak at length. They have a business model in mind, sounds something like Substack, which is probably a good idea. Their value is in the many millions of users they have. Highly recommend this podcast if you want to hear the story direct from the CEO.#