Pull the rug on the rye

There’s an amusing word pair in English and Norwegian: English rug is rye in Norwegian… and Norwegian rye is rug in English! *slaps thigh in amusement*

It’s sort of the opposite phenomenon of the “false friend” thing, where the same word has different meanings in different languages. Sort of. If you squint a bit.

Of course, this led me to wonder whether there are other word pairs like this… in any language. And this is impossible to google for, because… what are you gonna google for?

So a couple of years ago I asked ChatGPT, and it, of course, just output complete nonsense:

It’s like… it’s like… Yeah, that’s pretty stupid.

But now the LLMs are all brilliant and stuff! Let’s try ChatGPT again.

That’s indeed much better! It “understands” what the issue is. But of course “plage” isn’t spelled the same as “plague”, so nope. I prodded it some more, and asked it to take its time:

Well, See/meer and meer/zee is pretty close, actually. But no cigar/zeegar, I guess.

And that’s after five minutes of thinking.

What about Claude?

It gave me the see/meer/zee, too, but then it’s back to nonsense like “false friends”, which is the opposite of what I’m looking for.

(And besides, see/meer is less amusing than rug/rye because the two words mean almost the same already — that feels like more of a misunderstanding, like… “you call that cabbage rutabaga?! Well I call that rutabaga a cabbage!”)

OH NOES! I”VE POISONED CLAUDE

I prodded it some more, and it wrote some Python scripts and then ran for fifteen minutes, and nope.

It may be that rug/rye vs. rye/rug is the only pair that exists. I guess I’ll ask an LLM again in a couple of years.

Todo Lists and Procrastination

Two years ago I wrote a super simple todo package for Emacs to see whether I could get over my tendency to 1) procrastinate a lot and 2) being annoyed at my tendency to procrastinate.

Reader, you won’t believe the answer: Yes and no. Or rather, no and yes.

That is, I still procrastinate a lot (but less than before), but I’m way less annoyed about it.

As you can clearly see from the Emacs mode line above, I’ve got 27 “new” things on the todo list, 4 things that are “in progress” and… tada! 616 things that have been done!

It cannot be! That’s about one thing per day! (Although it’s more like no things most days and then eight things on a random day.)

I’ve had a look at other todo systems, but like I said two years ago, they seem to be specifically made to enable further procrastination: You can tinker endlessly with priorities, deadlines and arranging things in hierarchies. So you can spend an hour working on your todo list and then feel you’ve really made some progress. Which is a total lie, of course.

And open sores todo lists encourage further meta-tweaking — altering the todo software itself, so that you can add even more features. I’m proud to say that after starting to use it two years ago, I haven’t touched the anddo code even once! Behold the Microsoft Github repo!

So it’s just a list of tasks, and the ones that are new or in progress are shown. You can edit them and add notes, and you can change status, but that’s it. No dates, no priorities, no nothing. Just do it.

This system has really made me actually take care of a lot of annoyances and stuff, but even more importantly: Instead of looking at the lamp that’s sitting on the hall table every day and going “oh yeah, I should hang that on the wall somewhere; I must remember that”, I just put it on the todo list… and then my brain stops doing that every time I see the lamp on the hall table.

The system works!

(And then I might even hang it on the wall one day when I’m in the mood to do something, and I pull up the todo list… I might.)

I sort of… postpone the procrastination.

Anyway. That’s the life hack.

WLAN Antennas Are Annoying

A couple weeks ago, I got a new Lenovo laptop. When I ordered it (three months ago), Lenovo apparently didn’t have any WLAN (i.e., 5G or 4G cellular cards) in stock, so I couldn’t get one with a card preinstalled. But I got it with the harness (antennas and brackets and stuff) so that I could install it myself “easily”.

I checked the Lenovo parts store, and they now listed two WLAN cards. I got the Quectel EM05-CN, because I know that has solid Linux support now. It’s a 4G card — I could have gotten the 5G card, but apparently the Linux support for that one is pretty iffy. It doesn’t really make that much difference, in my experience…

(Whenever I mention having a laptop with built-in 4G, people without such a thing always defensively say “Oh, I just use a hotspot from my phone”. LIKE AN ANIMAL!?!?)

Last week I got it and:

Remove four screws, and pop the bottom off.

There it is! Four antennas strapped down, and a place to pup the card in. It’s got four antennas — they’re for the 4X4 MIMO thing that 5G uses. 4G uses 2X2 MIMO (sending and receiving simultaneously), so it uses only two antennas.

But here’s the problem. Those WLAN antenna connectors! I hate them! I’ve messed around with them before, but I’ve never gotten the knack of getting the antennas properly connected. I somehow managed to get the orange antenna connected, but as much as I tried, I just couldn’t get the blue one on there.

And… I seem to have messed up the connector by using too much force?

I hate those things! In the end, I destroyed the connector on the blue antenna..

It works, though — Linux detected the card automatically, and I’m getting pretty decent bitrates using the card, so I guess it works with just one antenna, too?

I’m not quite sure whether I should futz around with this any more. I could use one of the two remaining antennas that still have a connector instead of the blue one, but I think I’ve messed up the connector on the card. So I could get a new card? But…

Is there a trick to getting those connectors to er connect without using ALL THE FORCE!?

I don’t even know what those connectors are called… Let’s see… “U.FL”?!

Yeah, that looks right.

Heheh:

Many confusingly-similar looking connectors are called U.FL. The same or compatible connectors can also confusingly have multiple names, coming from different vendors or being transcribed differently.

How to Operate MHF® 4L Mating and Unmating Tool/ Micro RF Coaxial Connector / I-PEX

Oh, there’s a tool?

X.FL series and U.FL series [Connector] - Hirose Electric

Oh, he does it with just a finger. Hm.

How I connect MHF4 RF connections the easy way yes it’s cheating

“Here’s how the pros do it.” OK, that was actually very helpful… he uses a nail to seat it.

Hm… but should I get a new card or should I try to get one of the remaining antennas to use the possibly messed-up card socket? Decisions, decisions.

(I could also do some soldering, but I’d prefer not to. Or… would that change the impedance or something? I kNoW nOtHiNg.)

Is this Youtube Compliance Audit mail phishing or just vibe coded?

I mean, I do not have that many Google projects. Especially not Youtube projects. I have, like, eight?

But the DKIM/SPF on the email are valid, and the link is:

c.gle is the Google-controlled URL lengthener they use to track clicks and stuff, so it all looks legit. Except that they list a couple hundred project numbers.

Amusingly enough, it appears that none of the Cloud Console pages list these “project numbers” — you have to use the “shell”:

And indeed, that’s one of the project numbers listed in the email.

So… whoever generated this email just, er, did something wrong that added a gazillion other projects that have nothing to do with me? No wonder they want to audit me!

I’d also want to audit the person who has project number zero!

So… anybody know what’s going on?

In the legal jungles of Calcutta

This post is about me being sued in India, but mostly about using LLMs.

In 2022 I got the pile above in the mail. It turned out that this was a further iteration of the Indian lawsuit that named me as a codefendant (along with, er, Google and Yahoo) because I ran a mailing list archive called Gmane, and somebody had posted something possibly defamatory there.

I’ve been clearing out the loft (because I want to put more stuff up there), so I was just going to throw this bundle out…

But then it occurred to me that if I scanned the pile (I’ve got a scanner and I know how to use it), I could just run it through my OCR pipeline and then ask an LLM to summarise what it’s all about. I’ve been moderately curious, but not sufficiently to actually try to poke at the pile before.

And even people who say that they’re AI sceptics add “but you can use AI to do research, for instance to summarise documents”, and… I’ve been sceptical about those sceptics, because whenever I try to use an LLM, it mostly spits out something wrong at me.

So… it’s an experiment!

This is perhaps the most interesting bit for me personally (and I’ve actually read some of it myself):

It’s a summons from the Calcutta High Court saying that I have a couple months to reply, or I’m in deep shit. Or something.

(I don’t think Norway has an extradition treaty with India, and I certainly don’t plan on ever going there, so I didn’t really care much.)

I scanned the 1,200 pages (while watching The Four Seasons (Tina Fey edition) series, which is pretty amusing), and then ran it through my OCR pipeline. There I am:

(If you want to read these documents, here’s the resulting PDFs.)

But the OCR result was basically a 3.6MB text file, and I uploaded it to Claude.

I used the default settings, which apparently is Sonnet 4.6 Auto. And I apologise in advance: Friends don’t make friends read AI-generated text, but I can’t write this blog post without showing you the Claude-generated material:

What? 1982-2008? The lawsuit is from 2011… but it’s true, the papers include copies of previous legal things, when Milap Choraria sued Sanjay Jhunjhunwala, alleging that he’s stolen his land and stuff.

Claude spits out a few pages of this stuff, and doesn’t mention apparently not having read the documents beyond sampling a few here and there, I guess? Apparently it has a 1M character window, but… 1) it doesn’t say so anywhere, and 2) that’s not a good use of a 1M character window.

I gave it some pushback, but already here we see the problem: If you’re using an LLM to give you a summary, it will be as lazy as possible, and do some sampling of the text and output something. Anything. But the problem is that what it outputs sounds totally plausible — if you truly don’t know what the document is about, you can’t give pushback like this.

And then pages and pages of this.

“Good catch”, but still no go.

So now it has resorted to gaslighting me.

“Looping”? What? It’s a text file with the scans — so here it isn’t just gaslighting me, but it’s… what.

Finally!

Heh heh heh. Do I need to do that, though? Are you sure?

So: If I were to rate this interaction, I’d rate it “Not Just A Total Waste Of Time, But Actually Unhelpful And Misleading”. And this is the stuff I see lawyers on Twitter champion all the time — “it’s so good for doing research, but of course you have to check the output”.

They’ve been saying this for years! And Sonnet 4.6 was released in February of this year!

But to be fair, I then started a new session with Opus 4.8 on “Max”:

Hey! There I am!

This is much better — it says I’ve uploaded “Volume 1”, but I’ve uploaded both volumes (in the same file), so this obviously means that Opus has just sampled bits of the text file, too, but it’s landed on more relevant parts than Sonnet.

But it doesn’t say anything about the summons:

OK, so Opus needed prodding from somebody who knew what they were looking for, too — the simple “summarise this” didn’t work with Opus either: You actually need to know what the papers are about for Opus to cough up the interesting bit. Because (I’m guessing) it didn’t actually read the entire 3.6MB of text, either, but just did some samples here and there, and then faked a plausible answer.

I vaguely wonder what the case status is, but Claude is apparently blocked from accessing those sites, so I guess I’ll never know, unless somebody wants to check for me… (Hint hint!)

Anyway, if nothing else, this has given me something original to decorate my walls with. Not many people have a Calcutta High Court summons to proudly display, I betcha!

But the real takeaway here is that if you have a lawyer that says “sure, I use AI to do case research” or you have an administrator going “if carefully handled, AI is so useful for doing research into large document collections”, don’t argue, back slowly away, and then run away as fast as you can. Those people are insane and are living deep inside some kind of delusion.