Another AI rant:

One of the “key features” of LLMs is that you can use “natural language”, because that is supposed to be easier than having to learn a programming language. So, when someone says to me, “I automated this process using AI!”, what they mean is: They have written a very, very large Markdown document. In this document, they list what the AI is supposed to do.

In prose.

This is a complete disaster.

Programming and programming languages have one crucial property: They follow a well-defined structure and every word has a well-defined meaning. That is absolutely brilliant, because I can read this and I can follow the program in my head. I can build a mental model. I can debug this, down to the precise instructions that the CPU executes. This all follows well-defined patterns that you can reason about.

But with these Markdown files, I am completely lost. We lose all these important properties! No debugging, no reasoning about program flow, nothing. It’s all gone. It’s a magic black box now, literally randomized, that may or may not do what you wanted, in some order.

People now throw these Markdown files at me … and … am I supposed to read this? Why? It’s completely random and fuzzy.

Sadly, these AI tools are good enough to be able to mostly grasp the authors intentions. Hence people don’t see the harm they cause, because “it works”.

We already have a ton of automations like this at work: Tickets get piped through an LLM and these Markdown files / prompts determine what will happen with the ticket, and maybe they trigger additional actions as well, like account creation or granting permissions. All based on fuzzy natural language – that no two humans will ever properly agree on.

Jesus Christ, we’re now INTENTIONALLY bringing the ambiguity of legal texts and lawyers into programming.

Using natural language is NOT easier than using a programming language. It is HARDER. Have you people never read a legal contract? And that stuff can STILL be debated in a court room.

I can’t begin to comprehend why we, tech folks, push this so hard. What is wrong with you? Or me?

(And, once again, we’re ignoring other factors here. LLMs use a ton of energy and ressources, that we don’t have to spare. It’s expensive as fuck. It doesn’t even run locally on our servers, meaning we give all these credentials and permissions to some US company. It’s insane.)

@lyse@lyse.isobeef.org Clearing legally? You must have an amazingly efficient legal team – there’s like 10 new tools every week. 🤣

@prologic@twtxt.net Welcome back 👋

@lyse@lyse.isobeef.org

even our hippest AI enthusiasts found it absolutely terrible

Does this refer to the training course or to the tools themselves? 🤔

@lyse@lyse.isobeef.org Yes, and that’s why I’m 100% convinced that we’ll see a massive brain drain in a couple of years. This will affect young people even more, because they don’t have all the “old” knowledge to fall back on.

It’s concerning, I’ve warned about it many times, nobody listens.

I think the best thing one can do is explicitly not use any AI tools but keep your actual skills intact. Might be out of a (good) job for a while, but once this bubble bursts, this is who is going to get hired again. (I think.)

And considering how insanely expensive all this is, I’m still (mostly) convinced that the bubble will actually burst. This stuff just isn’t sustainable.

… or I might be wrong. And if so, I see an even darker future that I don’t want to put into words right now.

@lyse@lyse.isobeef.org AI result ahead, feel free to ignore.

I “asked” the AI at work the same question out of morbid curiousity. It “said” that SQLite converts that integer to floating point internally on overflows and then, when converting back, the x86 instruction cvttsd2si will turn it into 0x8000000000000000, even if the actual floating point value is outside of that range. So, yes, it allegedly actually saturates, as a side effect of the type conversion.

I couldn’t find anything about that automatic conversion in SQLite’s manual, yet, but an experiment looks like it might be true:

sqlite> select typeof(1 << 63);
╭─────────────────╮
│ typeof(1 << 63) │
╞═════════════════╡
│ integer         │
╰─────────────────╯

sqlite> select typeof((1 << 63) - 1);
╭──────────────────────╮
│ typeof((1 << 63) ... │
╞══════════════════════╡
│ real                 │
╰──────────────────────╯

As for cvttsd2si, this source confirms the handling of 0x8000000000000000 on range errors: https://www.felixcloutier.com/x86/cvttsd2si

The following C program also confirms it (run through gdb to see cvttsd2si in action):

<a href="https://we.loveprivacy.club/search?q=%23include">#include</a> <stdint.h>
<a href="https://we.loveprivacy.club/search?q=%23include">#include</a> <stdio.h>

int
main()
{
    int64_t i;
    double d;

    /* -3000 instead of -1, because `double` can’t represent a
     * difference of -1 at this scale. */
    d = -9223372036854775808.0 - 3000;

    i = d;
    printf("%lf, 0x%lx, %ld\n", d, i, i);

    return 0;
}

(Remark about AI usage: Fine, I got an answer and maybe it’s even correct. But doing this completely ruined it for me. It would have been much more satisfying to figure this out myself. I actually suspected some floating point stuff going on here, but instead of verifying this myself I reached for the unethical tool and denied myself a little bit of fun at the weekend. Won’t do that again.)

@lyse@lyse.isobeef.org

Disclaimer: Can’t guarantee that I’m fully awake and I’m being trained at work not to use my brain anymore, so maybe this is complete bullshit. 😪🧟‍♀️

It says here that SQLite uses signed integers:

https://sqlite.org/datatype3.html

In pure bits, 1 << 63 would be 0x8000000000000000, but as a signed value, it gets interpreted as -9223372036854775808. Subtracting 1 yields -9223372036854775809 – but that doesn’t fit in 64 bits anymore. It’s possible that SQLite doesn’t want to wrap around but instead saturates? Haven’t checked. 🤔

With 62 bits, there is enough room.

With 1 << 64, I have no idea how SQLite wants to handle this, because this should immediately trigger a warning, because it doesn’t fit right away. Maybe it gets truncated to 0?

sqlite> select printf('0x%x', 2 * (1 << 64));
╭──────────────────────╮
│ printf('0x%x', 2 ... │
╞══════════════════════╡
│ 0x0                  │
╰──────────────────────╯
sqlite> select printf('0x%x', 0 - 1);
╭──────────────────────╮
│ printf('0x%x', 0 ... │
╞══════════════════════╡
│ 0xffffffffffffffff   │
╰──────────────────────╯
sqlite> select printf('0x%x', 0 - 2);
╭──────────────────────╮
│ printf('0x%x', 0 ... │
╞══════════════════════╡
│ 0xfffffffffffffffe   │
╰──────────────────────╯

@prologic@twtxt.net Oh, so that’s where you are! 😅 Great scenery. Enjoy!

@bender@twtxt.net Ah, great, thanks!

@lyse@lyse.isobeef.org Yeah, I really don’t know anymore. 😅

By the way, why do so many of them wear glasses? As a kid, I’ve been told that people with glasses can’t become astronauts. So I gave up my dreams. Now it looks like that was a lie? ☹️

Christina Koch looking at Earth is my new wallpaper:

(Sorry, forgot where I originally found the image. Some NASA photo collection.)

@lyse@lyse.isobeef.org It’s impossible to avoid. Gotta wait a few years and then we’ll see. 🍵

This about sums it up.

Source: https://mastodon.online/@astro_jcm/116358684548644607

The problem is, they jump hosts all the time.

Maybe it’s time to add automated blocking after all … God, I’m too lazy for that. 😞

@bender@twtxt.net 🤔

I’m so sorry. 🙈

@lyse@lyse.isobeef.org Indeed. Very unpopular, though. I’ve long given up that fight at work.

In reality, there are too few real incidents. It doesn’t hurt enough. It’s always: “Something could happen!” But we’ve never been hit big time by an attack like this … so I just look like a paranoid idiot.

This whole thing was pretty weird, btw. I had no idea it was happening until basically yesterday. No news coverage, nobody mentioned it. 🤔 And suddenly, boom, we’re going to the moon. What? 😅

@bender@twtxt.net You saw it in person, I suppose? I watched the stream last night. 😅

In case you’re wondering where they are: https://artemistracker.com/

@quark@ferengi.one

talk next to nothing

I could rant about AI a bit and how it ruins every day at work, if that helps? 🤣

@quark@ferengi.one Ta-tah 🥳

@prologic@twtxt.net @lyse@lyse.isobeef.org Yay! Time for a new jenny release, then. 😊

@lyse@lyse.isobeef.org This is good! 💪 Let’s merge this.

(This one actually has the potential to live longer than 3 days.)

And another fork: https://drewdevault.com/2026/03/25/2026-03-25-Forking-vim.html

@lyse@lyse.isobeef.org 22 thru 25 are wallpaper-worthy. 👌👍

@prologic@twtxt.net Nice. 😊 That’s the beauty of a small instrument like that: You can just pick it up, play a little bit, put it back. 👌 (Can’t do that with my stuff. 🤣)

There you go, user-defined color schemes:

@iolfree@tilde.club Will do. 🫡

@lyse@lyse.isobeef.org A-ha! That means you haven’t spent enough time with these tools! Go on, try it! (If you don’t, we’ll fire you.) I’m sure you’ll like it!

Oh. Feed rotated. hfgl 🙃

@lyse@lyse.isobeef.org I bet that their AI agent handles that as well, so … 🤷‍♀️

@rrraksamam@twtxt.net They won’t.

@lyse@lyse.isobeef.org That would have been the noble thing to do. It’s more work for me, though. I blocked that idiot and moved on with my life. 🫤

Another vibe coded bot, I guess. 🤦‍♀️

@quark@ferengi.one Ignored. ✅

@lyse@lyse.isobeef.org Hm, I’m not sure I would want to do that:

ForwardAgent
    ...

    Agent forwarding should be enabled  with  caution.   Users
    with  the ability to bypass file permissions on the remote
    host (for the agent's Unix-domain socket) can  access  the
    local agent through the forwarded connection.  An attacker
    cannot  obtain  key  material from the agent, however they
    can perform operations on the keys that enable them to au‐
    thenticate using the identities loaded into the agent.

@lyse@lyse.isobeef.org Changing the user name helped, it now says Authenticated to git.mills.io ([199.247.16.95]:2222) using "publickey". ssh-add ... had no effect (even after ssh-add -D).

Here’s a debug log, @prologic@twtxt.net, perhaps you could take a look at this 🙏: https://movq.de/v/116c5f514b/clone2.txt

(Might be a silly mistake on my part. Wrong remote path or something?)

Anyone else having trouble pulling from git.mills.io? 🤔

$ g clone ssh://git@git.mills.io:2222/yarnsocial/twtxt.dev.git
Cloning into 'twtxt.dev'...
git@git.mills.io: Permission denied (publickey).
fatal: Could not read from remote repository.

The key verification function on https://git.mills.io/user/settings/keys says I’m using the correct key.

This also looks good:

$ GIT_SSH_COMMAND="ssh -v" g clone ssh://git@git.mills.io:2222/yarnsocial/twtxt.dev.git
...
debug1: Authentications that can continue: publickey
debug1: Offering public key: /home/user/.ssh/keys/key-millsio ED25519 SHA256:nVNT... explicit
debug1: Authentications that can continue: publickey
debug1: No more authentication methods to try.
git@git.mills.io: Permission denied (publickey).
fatal: Could not read from remote repository.

Does it work for you, @lyse@lyse.isobeef.org @prologic@twtxt.net?

@falsifian@www.falsifian.org My condolences. I mean, errr, congratulations! 😅

@lyse@lyse.isobeef.org We’ll see. Hasn’t really happened in the past. 🤔

@arne@uplegger.eu (Arrrrrgh, bitte lass’ mich nicht noch ein Spielzeug kaufen … 😂)

Now that Winter has come to an end, I’m realizing that the default amber color scheme of my widget toolkit might be problemaic.

Readability isn’t great when the sun is blasting through the windows. 🥴

I should probably make this full themeable by the user …

(Haven’t worked on this code in a month, sadly.)

@thecanine@twtxt.net

as things are now in fact even worse

You mean this, right?

Contrary to a vague mention of a possible “advanced flow” that may eventually allow “experienced users to accept the risks of installing software that isn’t verified”, Google’s description of the program continues to state plainly that:

Starting in September 2026, Android will require all apps to be registered by verified developers in order to be installed on certified Android devices

Until such time that they have shown evidence that it will be possible to bypass the verification process without undue friction, we must believe what is stated on their official page: that all apps from non-registered developers will be blocked once their lock-down goes into effect.

Putting obstacles in their way: To make scraping my code slightly less trivial, the repos on my website now only show the newest 15 commits, e.g.: https://movq.de/git/katriawm/log.html

@rdlmda@rdlmda.me Well, I’m reading. 😅 twtxt/Yarn is rather slow and low-traffic, though. And let’s be honest, there aren’t that many people around.

Well, someone just did this:

https://codeberg.org/NerdNextDoor/evi

We’ll see how long it’ll last …

@lyse@lyse.isobeef.org Yah, I don’t really know what to do about it. I’ll probably keep using Vim as it is – until there are bugs. And then we’ll see. Build an older version from source maybe?

I’ve been using Vim for such a long time now, I have very little motivation to switch. 🫤

Ah, found it: Lauren Hart did a laryngoscopy a while ago:

https://youtu.be/zy1El8U2kaA?t=175

Here around the 3:00 minute mark, the doctor explains what’s going on and you can (kind of) see which tissue is vibrating. Pretty interesting.