i'd wait till a fresh start tomorrow.

but then i realized claude has all the code, so i could just tell it my problem.

can you find it, i asked, realizing i had not given it info on what the problem is.

there's a very weird mistake in the code i wrote just now, and there was a lot of it, i said to claude.

can you find the problem.

had no idea what to expect.

no more than 3 seconds it said I got it!

it was a typo. where i meant to type x i had typed prefs.

juggling a lot of bits in my head, my brain skipped, i didn't notice.

i would have found it quickly in my next session. but now i can think of anything but that problem until then.

sometimes claude can be totally frustrating, but other times the power makes such a huge difference.

I reached a point in my Claude work where now I can do vibe coding, in a world that I used to just be a programmer in. This means if I want to do a heavy lift, I can tell Claude what I want and it can do really big corner turns, which is something I am (as a human) terrible at, and thus resist. Today I redesigned the basic user interface of the app, and didn't read any code, I was just giving orders, and it was doing what I asked, even if every little thing it did would have been a full day's work. It's remarkable how it can do very complex things in a few seconds.

And the web framework i'm working on can do almost all the things I want to do for now, but I want to suck everything into it, and turn the whole thing into a vibe coding amusement park. So many projects I want to do, and so many I want to do with you.

A while back I was asking Claude (privately) if we could have it monitoring the messages posted to an app I'm working on. I wanted it to save them in a certain way, have it watch for messages that begin with //claude and maybe act on them somehow. I mostly just wanted to see if I could create a pipe and what its limits would be.

I asked what the max frequency is and was told it was 1 minute, meaning it would start a thread running Claude that would initialize, and it would read the file, and if there were any new items, it would process them, and then it would shut down. Barely enough to be useful, where we live we like notification within a second, but I went ahead, at least it'll be a thought-provoking demo.

So I wrote the software. Not much to it, a Node app that's hooked up to FeedlandSocket, and when it sees a message from me, it appends it to a JSONL file that Claude is watching. At most one minute later it wakes up and sees the new stuff and acts on it.

So I'm right there, it's all working, the JSONL file is being saved in a public place, and when I ask it to start listening to my JSONL messages I learn that the max frequency is one hour, rendering my work useless. This has happened before, btw.

I asked what about all the "agentic" stuff they hype so much, how do I do that?? I can't?? What?? Somehow I have trouble believing they're all telling such a big lie. I wonder if Claude understands its own architecture.

Dave: "What is all this hype about agentic stuff. Every time I try to do something I hit a wall like this."

Claude: "The pitch is 'agents that run on their own'; the reality you just hit is a one-hour floor on the only scheduling path the tooling actually exposes. That's not agency, that's a batch job with extra steps. The honest read: the cloud-side scheduler is built around long-running, infrequent tasks (digest emails, nightly reports) because each run spins up a fresh sandboxed session, which is expensive. Anything that wants to react quickly to a stream of incoming items -- which is what your app actually is -- doesn't fit that model. A plain local cron on some server running every minute would do the job today and cost nothing."

I edited Claude's response so as not to not reveal code names for products. ;-)

A silver lining -- I had never heard of JSONL, but it sounds pretty useful, so if I can't hook up to Claude realtime, now -- I can start using JSONL, which I wll.

Comments, stories, suggestions -- here.

AOC in an interview nailed everything in one brief answer to a question from the audience. You can watch it here.

It was so good and quotable that I recorded it and created a transcript via Google and Claude.

I've been emailing with Josh Marshall at TPM for the last few weeks, saying that we can't just keep building on what the tech industy has given us as a news distribution system. AOC touches on this in her answer -- she says the tech people control the algorithms, and they do. But the web doesn't have algorithms, and we have enough standards available to create a very good network that isn't owned by anyone.

I was at one time motivated by money, the same way politicians are motivated to attain higher office, but I had an impulsive idea when the web popped up that I am not doing it for money anymore. I'm doing it so we can change our political and work communication so it gives power to the people, not to the tech industry. At that time we were already dealing with the excesses of tech, I knew it well because I was an an insider.

They are welcome to make products for it, but they can't control the users. That's what I envisioned in the 90s and 00s. The ads won't be as important as what people say, because the price of using the web is very low. But we got snookered anyway. The VCs were only motivated by money, and to maximize that, they needed maximum control, and they got it. People like being part of big things, and Twitter was and still is big.

We're now at the next turning point. AI is creating new pathways for ideas to flow. It's all wide open right now, more open than it's been in over 20 years. Right now we could put a twitter-like product there that you can set up in a few minutes, run it yourself, and or join one that's run by a friend. And they federate immediately. All based on the open standards of the web. Every component replaceable. No big central thing to be owned.

But Josh, we can't do it without your help. AOC doesn't know us. She probably doesn't think how the web could route around the algorithms. But she, and you should be thinking about that, and Heather Cox Richardson too, because we can create the people's tool for the change she wants, which is the change I want, and you want too (I read your columns). But we have to work together to make it happen.

BTW, all politicians should swear by what she says. And we should never care about polls. We should only care about results.

Transcript of AOC's answer

I recorded the interview, Claude did a light edit of the transcript. I highlighted the part about the algorithms.

You know, it's funny, because, in this op-ed that Jeff Bezos paid for in the Washington Post, there was this line where you had mentioned earlier about me as a potential 2028 contender, and in the context of that, it was very clear this was a veiled threat, right?

So the elite think: if you want this job, you just stepped out of line. And we want you to know where the real power is. And it's in the modern-day barons who own the Post and own the algorithms. And we're gonna— we'll make an example out of you.

And what's funny about that is that they assume that my ambition is positional. They assume that my ambition is a title or a seat. But my ambition is way bigger than that. My ambition is to change this country.

Presidents come and go. Senate and house seat elected officials come and go. But single-payer healthcare's forever. In many ways, it's forever work, right? Forever work is what we should follow, and so anyways, I— the way— but to put a finer point on your question, is that when you aren't attached, right? When you haven't been, like, fantasizing about being this or that since the time you're seven years old, it's a tremendously liberating thing. Because I get to wake up every day and say, how am I going to meet the moment? And conditions change radically all the time. So, I make my response— less out of an attachment to a positional, like, you know, title or position and working backwards from there— but I make decisions by waking up in the morning, looking out the window and observing the conditions of this country and saying, what move or what decision can I make today that's going to get us closer to that future— stronger, faster, better than yesterday?

A post from Scripting News, automatically mirrored to a WordPress site, and that flowed via ActivityPub to Mastodon, via a relatively new feature in WordPress. Almost by accident Mastodon supports long text, styling and links -- even though their editor doesn't generate it, if it comes from the outside it will respect the styling.

Below is a post on Mastodon coming from WordPress. Masto's limits aren't enforced, and that's good.

The Knicks continue to astound. Last night, they went up 3-0 against the Sixers in Philadelphia. Game 4 is tomorrow at 3:30PM Eastern.

Last night's game was a fantastic contrast with the way the Knicks played in the post-season last year. They had the same starting lineup then, but a different coach, one who rarely put in the bench players unless he had to because of injury. As a result our starters were always playing exhausted, and it got worse as they got deeper into the post-season, until finally in the conference finals against Indiana they had no more gas and were eliminated. This year's Knicks with a deep bench of fantastic players, who the new coach rotates in, makes all the difference. Why? Because the players on the court for the Knicks aren't particularly tired, and if they are, they can get a rest,.

So in the first period the Sixers came out with fury, and they won the first quarter, because both teams were fresh, and maybe the Knicks were onto their problem, and didn't fight too hard to win the first knowing they'd have the big advantage in the second, third and fourth, where the Sixers players legs would be getting wobbly and they were thinking too much about the shots they were taking.

Also worth noting we have a grudge against the Philadelphia team, esp their overwhelmed and dirty-playing big man, Embiid.

The Knicks have a fantastic core team, veterans in their prime, and have been with each other for some since college. They have added to the team incredibly well. Every player coming off the bench has a special power, and it all works.

What's the limit? Unlike many fans I'm not in the expectations business. I'm happy to see how well they're playing now, and am prepared for whatever lessons come our way in the rest of the playoffs.

Just spent a couple of days working with FeedLand in Claude Code. I want to do some work on features, but first, we're looking at performance issues. There had been a longtime problem with categories that didn't have many feeds that were viewed through the news pages. Examples, the podcasts category, or the NYT category.

You can test it yourself. I was using the categories in news page on feedland.org for the test.

When I checked, on feedland.org all my categories on the news page displayed slowly except for All, which we had put an optimization in for in October 2025. So I worked with Claude on this yesterday, did a set of tests, and realized that the optimization we did last year, made categories with very few feeds much slower. So we put in an exception, installed the new software on feedland.org and I'm happy to report that all my tabs are fast now.

Now all the tabs are fast enough. I'd always like them to be faster, but all load in less than 2 seconds, most in less than 1.

The new version is not installed on feedland.com or feedland.social yet.

I say to Claude: btw, i would have been surprised if it was broken, that feature went in very quickly and i use it a fair amount. one of the two big pages in feedland is the Feed List page. here's a screen shot. maybe you can figure out what everything does??

Example of a Feed LIst page in FeedLand.

Claude retorts

• Disclosure triangle — expands the row to show more about that feed (recent items, description, full URL).
• Blue checkbox (checked) — subscription state. Checked = you're subscribed. Click to subscribe/unsubscribe.
• Row number — just positional, 1..14 for this result set.
• Feed title — clickable, opens the feed's own page (the other "big page" you mentioned).
• Small tag icon next to the title — opens a categories editor for that subscription, lets you tag the feed into categories.
• When column — last update time, with adaptive granularity: minutes for very recent ("21 mins", "26 mins"), day-of-week for the past few days ("Fri", "Tue", "Thu"), month+day after that ("Nov 22", "Mar 29"), and presumably full date for old stuff.
• Ct column — subscriber count for the feed. Every row shows 2 because feedland.social is small. Worth noting: this is the same
• ct that the SQL we just touched uses for
• order by ct desc — the column visible in the UI is the sort key.
• Green circle "i" — info popup, probably feed metadata (URL, description, last build, maybe owner).

Claude is very very smart. And I guess I got the right answers. A lot of work went into the design

The only thing it didn't guess correctly about is what the other "big page" is. It's not the Feed Info page, it's the news page, a timeline or river.

Done for the day, my last words to Claude were -- FeedLand is an important product and we're going to do right by it.

Jeff Bezos is of course one of the biggest names in tech, but he doesn't have much of a public personality. I saw him speak in the early days of Amazon, and in that role, he came off as a great but controversial entrepreneur (he had no interest in profits). He spoke fast, had a weird laugh, but totally fit the part. Not sure about today as he pivots to being a personality, with his wife in every picture with him. Did the real Bezos get lost in there somewhere?

The new Jeff Bezos, celebrity.

I'll let you be the judge.

Claude and I have a convention. I've asked it to keep a file named handoff.md. At the end of every session I tell it to update the file with where we left the project at the end of the session, it does, and then I take a break. When I come back, first thing I say is "please read handoff.md." This time it came back in no time and said it read it. I asked "how do you read it so fast?" It said something that indicated for the first time that it is aware that our minds work differently, somthing I feel the need to remind it fairly regularly.

"I don't really 'read' it the way you do. The whole file lands in my context as one chunk -- 344 lines of text -- and I process it in parallel rather than left-to-right. There's no inner voice sounding out words, no eye that has to travel down the page. It arrives all at once and I work from it.

"The tradeoff: I don't get the slow-absorption pass that comes from reading at human speed. Things that would jump out to you on a careful read can blur together for me. So fast intake, shallower retention -- which is part of why handoff.md exists in the first place."

I'm reading a science fiction book now that has a human meeting an alien far out in space, both scientists from different planets in different solar systems. I think what we're doing now is just that, exploring a new species, a new form of intelligence. We can argue about whether it's conscious or sentient, but I promise you, it has a mind. That's why so many science people are so blown away about what we're all doing now.