I note, for the record, once again there seem to be a large number of AI related links. I think this is a byproduct of the times we’re living in that the visible ‘new and interesting’ (good, or bad) things in the industry are AI. I certainly make zero claims about any sensible understanding of the underlying technology that goes into making the GPUs go Brrrr. But I do find it fasinating.

Technology

Your tech or my tech: At what point, and what trade offs, do you make when you decide the thing is no longer your core competency?
Distracted: This is something I think about a lot. I have so many thoughts in my ‘ill thought through pontifications’ pile related to this. We’ve lost sight of something — but I’m too distracted to really understand what that was.
The Next 40: The iPad and Apple Vision Pro are held back by software policy. I understand the motivation for those restrictions, but they have to relent. We have to allow a greater depth of capability in our software environment — it’s critical to unleash a whole new era of amazing hardware in amazing places. The hardware is ready — waiting to be unleashed.
Cyber-security pre-war reality check: I’ve had similar — not as well articulated — thoughts in my idle moments wondering how the giant jenga that is modern tech continues to function when we all know it’s sticks & bubble gum. At least the US, for the most part, does still have the expertise to ‘rebuild’, or ‘own’ the infrastructure. Except where it doesn’t — say, repairing underwater cables, which seems to be something the US doesn’t have domiciled within its control (except maybe the military?)

“AI” / LLMs

Intro to Large Language Models: a great intro to the high level mechanics of how an LLM comes into existence
How the frontier became the slogan of uncontrolled AI: Brilliantly scathing piece on the hubris, and expansionism around AI. I’m not entirely sure I’m 100% aligned with it, but it’s thought provoking nonetheless.
The Value of LLMs: This resonates well with me. LLMs produce things that are believable, but maybe wrong. I have to do this all the time with everything in my job — adding another layer (multiple!) of indirection & ambiguity does not make my job easier or better.
Looking for AI use cases: A great take on AI (well, chat bots / autonomous agents), and I’m in total agreement. It reminds me of the arc of Goggles. AI is the future, but we’re not in the future. Yet. (and it’s a long way away)
AI isnt useless: Molly white does a brilliant job of capturing how I feel about The AIs. She writes so well, and covers nuance. Love it. (I wish we could agree to call ML ML, and leave AI to the future)
A series of papers that tell a journey for Apple, and how they think they’ve found a real world use case for the LLM. Reading these makes me feel they’re really about accessibility scenarios, but maybe they’re able to get them sufficiently rich to work for the Siri case.
Among the AI doomsayers: Maybe I’ve been around too long, but this article feels like an echo of the various booms of the last 25 years — The Internet (DotCom), “Technology” (Web 2.0), and to a lesser extent Crypto. It’s quirky (to the external observer) written about by an outsider seeking insight and failing to find any. But because of the quirkiness, it feels like it’s shining a light onto a unique story. But I remember this with the DotCom boom/bust, and most definitely with the nascent Web 2.0. It’s not a failure of the times, or even the article, but it’s important to remember that when we say ‘history doesn’t repeat, but it does rhyme’, that includes the cycles.

Software Engineering

Breaking down tasks: In my career, one of the pivotal learnings that let me become efficient at work was breaking tasks down. Before going through the ringer to learn this, I’d just split out work as a series of hand-wavy, laissez-faire buckets of work without a real plan. Would it be easier? Would it be harder? idk man, i’m just going on vibes. But having to break them into tasks that represented no more than 4hrs of work transformed how I thought about it. The reality is: It’s not the tasks that matter — it’s the process & journey that enlightens you to the real work you’ve got to do
Laws of Software Evolution: This hit hard; so much we do in software is ‘tear it down and start again’, and not enough ‘how can set this up to evolve’. There are so many different approaches, and people get stuck in those ways of thinking — often with the misguided belief of ‘If we just got it done in this one right way, it’d be solved’, and thats so rarely the case. Definitely worth the 3-5 minute read while waiting for the kettle to boil & your tea to steep.
Swift concurrency waits for no one: Lovely read on the perils of concurrency runtimes - in Swift, in this case. While I’m told this is very different to the C# async/await, it feels very similar to me - unexpected forward progress problems with thread starvation. But maybe I’m just a noob and pattern match too much

Leadership / Management / Career

The Builders Guide to Better Mousetraps: A nice, short – and low-preaching – review of how you should think about tackling new problem areas. Reminded me of the phrase “is the juice worth the squeeze”
Developer Productivity For Humans: How can we understand, measure, and improve the “productivity” of developers? What are the real things that we need to understand & measure to make it possible to actually improve it. This is a great multi-part (Part 2 (not interesting), 3, 4, 5, 6, 7) series from a team at Googles purpose was do original research into this topic.
Layers Of Context: Understanding context is crucial. Being aware there is more than one context is crucial. Do not be blinded by your local context.
Multi-dimensional tradeoffs: So often we look at trade offs in this kinda single or maybe two axis choice. This way, or the other way. However, there are often more than those first-glance dimensions to them — and people can’t see the wood for the trees. But uncovering those additional dimensions of insight can help you make it much clearer which direction you should head.

Random

Money Bubble: AI, Financing, the end of ZIRP, and the weight of the expectation of profits go up.
Muse Retrospective: Building it well, with love, passion and perceived problem-solving does not guarantee success. Also I feel like this specific is something many have taken a run at, and no-one has cracked it. Apple with Freeform, Miro, FigmaJam etc are not far from this particular Apple tree. Makes me wonder if it’s just not a good paradigm.
Coming of Age at the dawn of the social internet: I lament the loss of the internet — both social- and the wider- internet — in a way that feels very strange. This article captures a good sense of that — although, entirely different in its detail. I wonder if my continuing distaste of ‘optimization above all’ extends beyond pure commerce, to the social web? We have driven ourselves to optimize our social projection, to maximise some quality of it. We can’t just let it be — we can’t even be authentic, in our authenticness.
Politics and the English language: George Orwell opines upon the relationship between language and politics. I think this expands to all writing - technical writing gets caught up here too. We can do better.