Idle thoughts for October
I decided to take a bit of a writing break after last week’s mammoth essay. Not because writing these things takes so much of my time. I’m a relatively quick writer and most of the time even a two-thousand word post only takes up a couple of hours of my time each week.
No, it’s more a question of distance. One of the more effective tools you have for assessing a piece of your own writing is cognitive distance. Wait until it’s a bit less familiar to you and then, when you read it again, its strengths and flaws will stand out as obvious.
But it’s hard to create that distance if you go straight into preparing another text on the same topic.
So, a few days of not thinking about writing brings – not quite clarity, but more of an empty space where idle thoughts and ideas creep in.
- The “Liskov’s Gun” essay was successful in that it helped clarify my own thoughts on Web Components and frameworks. This left me much more positive about web components than before and led me to have another look at frameworks.
- That look, which I might write up at some point (or I might not), led me to the conclusion that I disagree with what most client-side frameworks like Vue.js, SolidJS, and Svelte are trying to do, in that I don’t think they’re interested in the same kind of problems I am.
- The Risk-Reward calculation for writing keeps changing. Chatbots, LLM training data theft, the popularity of video, the declining popularity of reading, roving gangs of social media assholes, and the fact that most publishers are utter fucking ghouls who will take a bite out of your face at a moment’s notice, mean that writing with the intent of publishing it somewhere comes with considerable risk.
- The rewards are also slim. The Intelligence Illusion did extremely well by any metric for a niche non-fiction book, and it still only sold just under seven hundred copies. Writing for the newsletter informs my own decision-making and dev practice – writing is a form of reasoning – so for the newsletter the Risk-Reward calculation is largely positive, but I need to be doing something different for that calculation to become positive for writing another book or book-like object. Still not entirely sure how that would work, but I have ideas.
- Having a book like The Intelligence Illusion out there also worries me. I’ve heard too many reports of people who have read it using the ideas from the book to sell “AI” crap – some of these reports coming directly from those readers themselves. Turns out that an analysis of the business risks of generative models is useful if you’re a sales type putting together a convincing cost-benefit analysis for a pitch, or if you’re pitching yourself as a “moderate” authority on how to best adopt “AI”. Not sure what to do about this, but it’s not a particularly comfortable thought.
- I’m working on updating this site’s design and move over to Eleventy at the same time. Making sure the CSS works for the chaotic mess that is fourteen years of blog posts takes a bit of time, but getting off Hugo is well worth it. Eleventy also happens to be one of my favourite web development tools. Working with it is much more fun than Hugo.
- Iceland’s highly dysfunctional right-wing government fell apart over the weekend – the biggest party in the coalition felt that not being allowed to deport a chronically ill child was untenable so they broke up the coalition – so now I have two separate November elections where I can be anxious about how many voters will vote for out-and-out cruelty and whether they will outnumber those who vote for solidarity and cooperation.
Lots to think about. Lots to worry about. But nothing conclusive. That’s just what some weeks are like.