Emacs Canz Spell Goodly Know

Like many nerds, I feel so embarrassed whenever I see that I’ve misspelled something. But some years back when I started producing text in huge amounts for things like the Eclipse Comics read-through I realised that my options were: 1) Remain neurotic about this stuff and produce one blog post per week, or 2) Start caring less and produce a lot of blog posts.

I think I made the correct decision — I just don’t have time to proff-read all this stuff, so typos will abound. It’s just not feasible to put as much work into that stuff as you’d do for a properly published article. The reader gets what they pay for.

But now there’s a way to have your delicious cake and eat it too — just ask an LLM for help.

NOOOO! *audience runs in slow motion towards camera*

I know, I know. Whenever I read somebody in deep AI psychosis who thinks they’re being AI sceptical saying “well, of course you shouldn’t let the LLM write your articles. Instead I let it do the research, and then I make it provide an outline, and then I write some text based on that, and then it does formats and restates what I wrote. IT”S A WHOLE DIFFERENT THING! I WROTE IT!!!!”

No. You should never let an LLM do those things: Not the research, not the outline, not the formatting, not anything. Reading the output of LLMs is abhorrent to people. You should never subject anybody to an LLM generated text, even if you’ve done “quality control”.

LLM-generated text gives people the ick. It’s better to have an awkwardly phrased article you’ve written yourself than subjecting somebody to a Claude text. It’s not just the ick — it’s the horror.

However, you can use LLMs for spell- and grammer check:

I’ve whipped up a li’l minor mode for Emacs (and put it on Microsoft Github). The default prompt is very clear that I want the LLM to do nothing but copy edits — no structural changes, and no other helpful suggestions. Just fix grammer and tippos.

And then you can interactively go through the fixes and toggle back to the original ones if you disagree.

Yes, yes, sure — Grammarly, etc — stuff like this already exists. But I just don’t like using them. I use the built-in Emacs spellcheck, but it doesn’t catch everything. Having an LLM give the text a quick once-over for obvious mistakes seems nice.

While testing this minor mode now, I see that LLMs haven’t changed as much as I thought — they’ve gotten so much better on programming the last half year, it’s incredible. For stuff like this, though, it’s still pretty unreliable. Any two runs won’t necessarily fix up the same thing, or the same thing the same way. And getting it (I’m using Claude Sonnet 5) to output it in the requested form is not reliable at all — I’ve been reformulating the prompt, trying to get it to output it in the format I the mode needs, and… I think I’m at 95% now?

So, like all with all non-programming tasks, LLMs are still pretty much toys. But useful ones!

Text Pages Redone

About a month ago I discovered that the scripts that classify pages as either “text” or “comics” on kwakk.info‘s section for “text pages from comics” (*phew*) didn’t like pages like the above. Which meant that all letters pages that had illustrations and stuff were categorised as comics and left out of the search index.

So, for instance, none of the Johnny DC pages were included. Which is a serious mishap as you can imagine!

Even worse — Usagi Yojimbo letters pages like the above were also excluded.

Well, a new classification script was put into production, and it chewed through nine million comics pages and spat out a new list. Those pages were then put through the OCR and indexing processes. The computer worked at this for three weeks.

The results seem promising — there’s now more pages included, of course, but it might well miss some other pages now. I mean, classifying pages into “text” and “not text” is a probabilistic enterprise, so it’s going to make mistakes: Some previously correctly identified text pages may now be misclassified and missing from the new search index.

But there you go.

Have at it.

And with that, we now seem to have over 1,1M pages (in total) in the search index.

Imaging The Past

I’m continuing to root through the hoard of Commodore 64 diskette images from the 80s I rediscovered the other day, and I found that the diskettes contained a large number of files called things like DDLARS3. That’s my name! So what could they be?

Well, some Googling shows that the DD files are from Doodle Draw, which I have no recollection of what so ever. But looking at the images (after converting with recoil2png), I’m starting to recover some (possibly false) memories…

I had a digitiser! Or was it called a “frame grabber”? It plugged into the C64 and digitised a one-bit image! In real time! It was like magic! But… er… how did it work? Was it a camera? No, that doesn’t sound likely. Did I borrow a video camera for the images below? Did it plug into the VCR?

I’ve googled a bit now, and I can’t seem to find anything that jogs a memory — and they talk about the digitiser being slow? No, it was real time as far as I can recall? I mean, it showed a moving 1-bit 320×200 image in real time — when you triggered a snap, it took a while to spit out the data so you had to pause it… So I guess it was plugged into the VCR?

Does this ring a bell with anybody?

Oh, I probably have the software to the device on one of the diskette images. Yes!

This must be it? This seems to belong with the “Print Technik Video Digitizer”, which doesn’t look familiar at all, but it was a cartridge that plugged into the C64 and took composite in? So I guess that means that I borrowed a VHS camera from someone…

Anyway, here’s a field of images that must have been captured by this device, whatever it was:

I think I remember asking him to look particularly evil in those last shots… I was going to use him as the villain in a adventure game I was writing.

Speaking of which! I also found some images that were meant as assets to that game, so… enjoy?

Some of those are clearly copied from comics and stuff — that’s a Barks ghost at least… And a house from Stig’s Inferno… Oh, and Maggie from Love & Rockets, but I mentioned that before.

Anyway.

The Bubbles Aesthetic

I’ve been vaguely reading the Bubbles RSS feed for the past month or so. Bubbles aggregates a bunch of “small web”/”indie web” blogs, which sounds like it should give you a variety of diverse, interesting things to read: Original, different approaches to writing and design and etc etc etc, right?

I don’t want to be mean, but it’s not working.

The above grid is pretty representative, and it’s striking how uniform the blogs are. OK, I’m being unfair. While going through the newest entries in the feed, I removed a couple that didn’t use this template. But this really is what you’re going to see almost all the time if you’re reading these blogs.

How did all of these people arrive at the same design? Well, many of these “indie” blogs are hosted on Bear Blog.

And that’s the colour scheme they use… so everybody’s who’s hosting their texts there are using something similar? That is, a kinda desaturated bluish-gray background, and then text that’s also kinda bluish grayish, leaving you with a pretty low contrast page. I guess it’s supposed to be soothing? “Less distracting”? Demure. Making you concentrate on the text?

No menus… no images… no header… Just one (wide) column of text.

But apparently you can choose your own colours… and almost everybody has chosen colours that are within three degrees of the default ones? And while a lot of these are Bear blogs, many of them aren’t, and they’re still using the same design.

There’s a striking lack of images on these blogs, but that’s more easily explained: Most of them are written in Markdown, which makes it really easy and streamlined to write the text. But images are a totally different thing that you have to arrange separately, so it’s just too much work to casually drop something in there. So people don’t?

(Here’s where the Markdown blogger enthusiasts are going to say “but you just snap an image on your phone, and then drop it in Dropbox, and you just sign up for an AWS account, and then you put your image in an s3 bucket, and then you use the AWS command line tool to determine the URL, and then you paste that into your Markdown file, trying to remember the syntax, and of course, add a descriptive alt text. It literally couldn’t be easier”.)

What about the texts themselves, though?

Well, half of them are about how awesome it is to blog, and that everybody should blog, and rah rah rah INDIE WEB whoho! And also that you shouldn’t worry about whether anybody’s reading anything, because that doesn’t matter. JUST WRITE DAMMIT

I’m all for enthusiasm, but it’s starting to sound more like people are trying to will Something Interesting into happening, and that’s not really how things work.

But perhaps things will be different this time?

While typing away at this blog post, a Twitter user linked to this article about Camille Paglia on the Madonna blog (because of course it is). I just mention it because I was just struck by how diametrically opposite it is to the Bubbles aesthetic: There’s lots of images; several columns; stuff that moves; and black text on a white background.

Some of the Bubblers talk about how they absolutely can’t read anything if it’s like that… which I just find strange. But perhaps this is just the way people are these days? Can’t read a text if there’s anything else on the screen?

Anyway, here’s a picture of the neighbourhood cat as a reward for reading all that. Or if you just skipped down here without reading; that’s fine, too. In fact, that might well have been a better choice.