Weeknote: 18 September 2023
I’m sending this weeknote to the newsletter as well, just so that the email subscribers have some context for why it has been relatively quiet over the past couple of weeks. The short version is that I’m working on print versions of my books.
Links at the end.
Work #
My focus over the past couple of weeks has been twofold:
1. The new web dev setup #
Working on figuring out more straightforward and less complex ways to implement test-driven web development, with bundling as necessary, but without a framework. I’ve been documenting this work as a project setup template on GitHub, which I’m also using for a couple of my own projects.
I’ve figured out a slightly simpler and more capable ways of handling coverage reports for the Continuous Integration side of the setup, which I haven’t implemented in the template yet, but other than that it’s reached the point where it’s covering a lot of bases.
This little project has helped clarify quite a few of my thoughts and ideas on web development, which I’m hoping to write a lot more about in the coming weeks.
2. The print edition of my books. #
I’ve been working on the typesetting and design for hardcover editions of both Out of the Software Crisis and The Intelligence Illusion.
The test printing came out better than I expected, although there are a number of fixes and adjustments I have to make.
My plan is to get Out of the Software Crisis out first (doing one book at a time minimises errors).
Some day soon I’ll put up a pre-order page for it—doesn’t look like Payhip is going to work, so I might have to resort to a single-use Shopify site—but the long term plan is to get it out into regular print book distribution, so you could order them through your preferred book store. I still don’t know if I can pull that off, but it’s starting to look likely.
Very excited about this.
Reading #
Most of my reading lately has been documentation related to the web dev project above or the print editions, but I also had the pleasure of reading through this Master’s thesis: Framsækin bókaútgáfa: Útgáfustefna smærri forlaga og áhrif á bókamarkaðinn.
I guess that would translate as Progressive book publishing: the publishing policies of smaller publishers and their effect on the book market.
It’s in Icelandic so won’t be accessible to most of you, but if you are curious about the overall shape of the Icelandic book market, and you do know some Icelandic or are willing to risk auto-translation nonsense, it’s worth a look.
Movies #
Still continuing my quest to re-watch or watch all of John Carpenter’s movies. Still have ways to go, but I will say this: boy is his body of work uneven. It’s an odd mix of great movies, interesting experiments, forgotten gems, fun trash, not-so-fun trash, and a bit of genuine weirdness. Still digesting and I’ve got several movies to go, but it’s a lot of fun, overall.
Also did a bit of a Cronenberg re-watch. The problem with David Cronenberg is that he began as a genuinely poor to mediocre director who made up for it with vision. His movies pre-Videodrome are either genuinely bad or mediocre with sparks of brilliance.
I’ve always had a soft spot for Scanners, for example—who doesn’t love a well-executed exploding heads scene?—but when I was a teen I had it on VHS tape recorded after John Woo’s The Killer. (You could get four hour VHS tapes and if you were the type to record and share movies you could usually squeeze two on one of those without them looking worse than the usual VHS awful.)
This meant that most of the time I watched Scanners, which I did several times as a teen—remember, only weirdos don’t like exploding head scenes—I also watched The Killer. Whatever you can say about the narrative handwaving and plot shortcuts John Woo takes in it, the overall movie is just a perfect execution of the formula with impressive acting by the two male leads.
And Scanners? Not so great.
Videodrome is better, much better—it’s a fever dream of a movie—but it’s The Fly where he really comes into his own as a director and after that the level of quality in his work is much more consistent and always interesting.
Links #
Dev #
- "The hardest part of building software is not coding, it’s requirements". “The problems have become bigger, harder to fix, and more costly, but the source of the problem is usually the same: The requirements were unclear, inconsistent, or wrong.”
- "Pixels of the Week – September 10, 2023 by Stéphanie Walter - UX Researcher & Designer."
- "Design Under Pressure – Simply Secure"
- "Storing UTC is not a silver bullet | Jon Skeet’s coding blog"
- "On productivity metrics and management consultants – Surfing Complexity"
- "Remembering Molly :: Aaron Gustafson"
- "Silicon Valley’s Slaughterhouse". “This, in my mind, began tech’s age of empty innovation, where software poisoned every aspect of our lives without making them appreciably better.”
- "The Rules of Margin Collapse". Honestly, margin collapse in CSS was probably a bad idea and because I’m a lazy person I’ve mostly switched to flex/grid + gap to create this sort of spacing
- "Scope vs Shadow DOM". “I have never personally needed shadow DOM. Every time I dig in to assess it, I come away frustrated.”
- "Fixing Search". “However, browsers selling users to search engines to keep the web going is a bit like selling weapons in a war zone to bankroll cancer research. As mentioned, it’s a betrayal of users.”
- "Bricolage | Some notes on Local-First Development". I think Local-First is going to be an essential option for web development, but you can tell that devs have already given wholesale into their worst instincts towards hyper-complexity.
- "The death of unity"
- "I think I kind of hate lazy loading – Terence Eden’s Blog". I ran into this quite a bit back when I lived in the UK. Lazy loading can get quite counterproductive if your internet is sporadic.
- "Team size isn’t a measure of success - Jacob Kaplan-Moss"
- "Your feedback means more to small teams | daverupert.com"
“AI” #
- "Advanced NLP with SpaCy | Hacker News". “The world cannot migrate all its current NLP models for text classification and NER to ICL. There are nowhere near enough GPUs in the world for that to happen.”
- "OpenAI confirms that AI writing detectors don’t work | Ars Technica"
- "Microsoft announces new Copilot Copyright Commitment for customers". I see that some of Microsoft’s big enterprise customers have finally forwarded the “WTF!” email they got from their legal department to Microsoft’s sales.- "We’re still not innovating with AI-generated UI.". “I generally feel like companies with poor leadership who make bad business decisions will view these tools as a cost-cutting opportunity.”
- "Ableist interactions | hidde.blog". I agree with everything in this post but “production-ready” in this context does not mean “up to industry standards”. It means “replaces programmers.”
- ’Microsoft Publishes Garbled AI Article Calling Tragically Deceased NBA Player “Useless”. “AI should not be writing obituaries.”'. “Hard to believe that there’s anybody left that still believes in this tech, let alone an entire industry that seems to be falling for it.”
- "The Myth of Artificial Intelligence - The American Prospect". An older article but still relevant.
- “This book is required reading for everyone involved in looking at Generative AI for their business or in their work.” Andrew Doran on “The Intelligence Illusion” www.linkedin.com/feed/upda….
- "The A.V. Club’s AI-Generated Articles Are Copying Directly From IMDb"
- "On exponential growth of LLMs". LLMs aren’t growing exponentially. Even if they were their capabilities don’t grow with their size. Often they even regress or stagnage.
- "Can Large Language Models Reason?". See Betteridge’s law of headlines. “On a dataset of programming challenges, GPT-4 solved 10 out of 10 problems that had been published before 2021 (GPT-4’s pre-training cutoff date) and zero out of 10 problems that had been published after 2021.”