At least half of my enjoyment of Gullet of the Rust Demon, a one-page, system-agnostic tabletop RPG adventure by Dan D., comes from the “establishing shot” of the dungeon exterior at the top of the sheet. The tufts of grass and single, stray flower, cartoonish as they are, do so much to suggest where some adventurers might come across this minimalistic dungeon’s entrance, and how quietly, temptingly anomalous it would appear.

Discovered this one after DJ of Planets & Monsters reviewed it, which was brought to my attention in turn by Lusunati on the His Majesty the Worm Discord.

My friend and fellow New Yorker Francisco González launched a Buttondown-based newsletter last year to announce the release of Rosewater, the latest in his long-and-growing oeuvre of independently produced point-and-click adventure games. Having played and enjoyed that one over the intervening winter holidays, I was delighted to see Francisco start to use the newsletter as a development blog for “Project Lion”, the public code name for his current work-in-progress.

He’s updated the newsletter with spoiler-free notes, screenshots, and other tidbits about the upcoming game once a month since then; I just read the June issue. As someone who has let the BumpySkies newsletter sit idle since March, I find this inspirational and motivating.

Andrew Nesbitt is a software engineer and package-management blogger who also writes excellent satire in the form of standardized industry documents, such as change logs and incident reports. I have discovered his fiction by way of Incident Report: CVE-2026-LGTM, a hilariously brilliant story of countless autonomous AI agents responding to a malefactor in the stupidest way possible, eventually “succeeding” by accident, and all written up an an after-action report.

I consider this work a legitimate sci-fi comedy, and I reflect on how I seem to have a soft spot for sci-fi written in the form of archival documents adhering to some real-world standard, rather than traditional prose narratives. The epitome for this presentational trope, for me, is qntm’s Lena, a startlingly good horror tale written as a Wikipedia-style entry.

The Software Freedom Conservancy has published Recommendations When Using LLM-backed Generative AI Systems for FOSS Contributions. This article represents that nonprofit’s official stance and guidance regarding the use of genAI coding assistants when contributing to open-source software projects—and, by extension, the use of these assistants in general.

Some of the article’s more generally applicable principles make a deep impression with me, such as this one:

Do not overuse LLM-gen-AI, or allow your skills to atrophy. In our discussions with the FOSS community about LLM-gen-AI, there seems to be one universal conclusion: the systems are most effective and help the most when a very experienced FOSS developer sits at the prompting helm. LLM-gen-AI systems should complement existing skills and tools, not replace them. Developers should remain curious about why software acts the way it does, and this curiosity should extend to the LLM-gen-AI outputs — and even the system itself.

I also find the article notable for coining and consistently using the term LLM-gen-AI to describe this technology. Since departing the land of corporate style guides I’m left without a standard shortname for this stuff, so I’m interested to see what others are settling on.

I very much enjoy A Most Amusing Bot, a Mastodon bot by Andy Lundell which regularly posts scanned newsprint ads of novelty gadgets from the 19th and 20th centuries. And I like the very eerie note that this one ends on.

During some recent tinkering I revisited this Fogknife post from 2021, which I had illustrated with the following XKCD comic:

The original post reminisced on my working with real biologists as a young know-it-all coder, decades ago, memories that the cartoon had bubbled up. Looking at it for the first time in five years, though, all I can think is: OK, well, in both cases the scientists would probably ignore the tech people and ask an AI, writing a short prompt and attaching their CSVs or Polaroids. And it would probably do a pretty good job. Or they’d ask the tech people anyway, who might write a better prompt. But, either way.

A couple of weeks ago, on an errand to fetch the bread and milk ahead of the winter storm expected to wallop New York, I sidestepped into the neighborhood Barnes & Noble to see if they had the latest issue of 2600, The Hacker Quarterly in print. They didn’t, but they did have the autumn 2025 issue, so I bought that instead. I love it.

I have been peripherally aware of 2600 for a long time. I think I bought one issue as a PDF a year ago, as part of a personal push to read more professionally edited, socially conscious technology magazines, but I didn’t actually read it. But I did end up on the magazine’s mailing list, so I got reminded four times a year that it existed, and I suppose it took four reminders to finally break through.

The full-color covers of this little magazine belie the content, which is page after page of nuthin-but-text, one article after another by a variety of writers, and each one torturously typeset so that the text is either crammed to fit two pages or spread to stretch across one, with little consistency in kerning or leading. In the middle is a letter-column that goes on and on longer than you’d think, where the editors seem happy respond to practically all the mail the magazine gets. The whole vibe reminds me of the years I spent working on the student paper at UMaine. There is such a scrappy joy to all of it, and it’s all real, and current.

My favorite article of this issue is a two-pager by Micah Silverman titled “How I Became a Repo Man for a Day”, where the author, a seasoned security professional, is hired to help someone legally stop paying the insurance on a car co-owned by their ex-boyfriend, who was ghosting them. Creative solutions ensue. And there’s articles on AI, and articles on the current state of digital privacy rights, and retrospectives on phone phreaking, and so on. It’s all pretty great.

The magazine’s website is a wonderful mess and I can’t quickly figure out how to subscribe to a print edition. Maybe it’s not possible, but I hope it is. I will bring my own hacker spirit to this one and figure it out.

This week finds me gearing up to run a session or two of Mörk Borg, a doom-metal role-playing game and tiny cultural phenomenon that I discovered in November. As part of my research and preparation over this past weekend, I came across To Hell and Galgenbeck, the official Mörk Borg webcomic. It’s pretty great.

Łukasz Kowalczuk started the comic in 2022, and has averaged about one gruesomely colorful page per month since. It does a fine job expressing the Kirby-esque larger-than-life grand-guignol attitude of Mörk Borg, shot through with Kowalczuk’s own sense of the ridiculous, and I’ve submitted it as recommended reading to the friends and family who’ve volunteered as players for my experiment at game-mastering.

I was personally delighted to rediscover this comic because, as soon as I glimpsed it, I remembered that I’d seen it before: the artist was one of many who fled from Twitter to Mastodon as soon as Musk started wrecking the place, some three years ago. For a brief time, Mastodon sparkled with amazing cartoonists of every kind. I gleefully mail-ordered crazy comix-zines from many of them, Kowalczuk included. I remember seeing Galgenbeck at the time and following links to Mörk Borg, but I didn’t have much mental space to spare for table-top RPGs, and so merely took note of it. Then I forgot all of it again until this past fall.

Mastodon still has artists and cartoonists on it, certainly, but most who visited post-Musk have moved along to corporate-owned platforms like Instagram and BlueSky which offer much larger audiences and simpler authority structures, and I can hardly blame them for it.

I have discovered Nothing Doing, a webcomic by the pseudonymous HORSE ON VHS. I like it very much. High-quality sitcom-style funny-animal office follies, and the attached strip resonated especially well with me, given my very recent experiences in running through tech-job interview gauntlets.

I tend to do great at everything except the “Why do you want to work here?” questions, and cannier employers can tell when I struggle to come up with reasons besides “I already have senior-level expertise at what you all you do here, and I need a job.” I try anyway, and it never goes well.

But, that was last year. We turn the page to 2026, and the following Mastodon post by the mononymous Corey, one of the last ones I saw in December, sets the tone for me:

So 2025 goes down as the year I finally gave up on the tech startup industry, where I have worked for nearly thirty years. Such skills and experience as I possess are no longer valued. I take possession of my new venture, a tiny cocktail bar (The Bad Angel), tomorrow and am about to embark on the riskiest experiment I’ve ever attempted. Whatever 2026 brings, it feels good to not be begging some chowderhead to give me a job.

I’m not opening a bar, but I am choosing to gracefully accept the world’s messages about professional approaches that don’t work for me, and embrace my own self-knowledge about ones that do. Here’s to a weird and successful 2026.

Here is the specific reason I find Google so actively unpleasant to use today: we have all been trained for literally decades on how to “speak Google” by typing space-separated lists of keywords and phrases into the famous search bar, and then using the resulting page of “hits” to either find what we need, or to iteratively refine our search query. And you can still use Google like that, more or less. But, in many-perhaps-most cases, the top search results get preempted by the “AI Overview” answer, which reads your query completely differently.

The attached screenshot shows a detail of my trying to locate the particular Tom the Dancing Bug comic strip that I wanted to hyperlink to from my previous ’Twas This post. The first actual hit took me straight there, in remarkably classic Google fashion. But before that, I had to have the AI respond to me as if I had rolled up to a chatbot and said “Duhhh, Tom the Dancing Bug… twelve years old?!” out loud, barely able to string words together. Because I wasn’t trying to string words together! I was typing in a search query! Into a search engine! Which is not a chatbot!

But that’s why I once again got a bizarrely insulting answer from Google’s AI Overview treating me like a grunting simpleton. Today I apparently wanted to know about a 12-year-old named ‘Tom the Dancing Bug’, and so it assured me no such person exists—ho ho, that’s the name of a comic strip, not a real child, silly human!—and suggested other 12-year-olds I might have been looking for instead.

Maybe Google wants us to stop using Google like Google and just type in complete sentences now, ending in question marks. Is this how everyone on earth already uses it except for me, my narrow band of friends, and William Riker? Maybe! I don’t know! But if not, I feel like Google needs to, I dunno, put some re-training information out there or something.