Now

A sporadically updated log of what I'm reading, exploring, and thinking about

December 2024

We’re back in that glorious post-Christmas, pre-New-Year’s liminal period when the days blur together and I’m allowed to spend inordinate amounts of time tinkering on side projects and laying on the couch eating handfuls of Twiglets.

For the last couple weeks I’ve been plodding through an overdue and much needed migration of this website from Next.js to Astro . No shade on Next, but it wasn’t the right framework for the job; I found myself constantly battling server-side rendering errors and trying to escape hatch out of React. Astro lets me return to writing plain HTML, CSS, and JavaScript, rather than being trapped in a single front-end framework. It also takes a lighter approach to JavaScript, only including it when needed, and is designed for content-heavy websites like this one. If you’re reading this, the migration is complete! It took a good month but I’m much happier working in the new codebase.

The past six months have been a bit of a rush. At the end of July I got married, promptly learned we were pregnant, and started a new job two weeks later, right when debilitating “morning sickness” began to kick in. Which needs to be renamed persistent-all-day-every-day sickness. While the timing wasn’t ideal, I’m obviously thrilled about all of it. Living life on warp speed over here.

Luckily the company I joined was very understanding! Normally is a small design and research agency focused on the practical and thoughtful application of AI. They’re based here in London and I’m now a Lead Design Engineer there. This primarily means designing and building early-stage prototypes for our clients – either completely novel products or developing new services and features for existing systems.

Currently reading a sobering but beautifully written history of the Sackler family, Purdue Pharma, and the OxyContin opiod crisis called Empire of Pain . I’m usually not very good with horror, thrillers, or anything involving too much human peril and suffering. But Radden Keefe is a brilliant writer and his detailed historical narratives make it worth my discomfort.

Empire of Pain: The Secret History of the Sackler Dynasty

July 2024

Despite the regular drizzle, this summer is looking splendid. The tories are out of power. London is out in force enjoying the warm weather. I’m temporarily funemployed and thoroughly enjoying doing almost nothing.

I decided to leave Elicit back in April, and worked my last day there in June. The team was lovely, but living a life halfway between San Francisco and London was not. I realised I needed to work somewhere with more in-person collaboration, more senior designers, and on the same time zone as me. Fortuitously, I met some people who run an AI prototyping agency here in London and found our interests aligned. I’ll be joining them at the end of July.

Until then I’m touching grass, prototyping new ideas for language model interfaces, writing things, shrinking my bedside book pile, improving this website a bit, and getting married somewhere in the middle. I now have a lot of thoughts on the trad femme world of weddings, but I’ll save it for a proper post.

Currently working through John McPhee’s Annals of the Former World, which is fat but satisfying. I’m not usually a rock-lover, but McPhee’s storytelling and prose is beautiful enough to make plate tectonics, basins, granite, sediments, and conodonts seem compelling.

Annals of the Former World

December 2023

The limbo-like lull in-between Christmas and News Years is my favourite period of the year. I never know what to do with myself. I arrive back from chaotic, overstimulating family visits and sit amidst my pile of gifts, half-unpacked bags, and leftover chocolate, wondering what in the world I used to do with my days. I’ve long forgotten what it’s like to have any semblance of life structure. Despite it only being a week since I was on a focused, tight schedule.

This weird floaty time is beautiful. I sit around writing long, trite reflections on the year. I look through my calendar week-by-week and count up arbitrary numbers of things.

1,140 kilometers run
89 coffee dates and park walks with friends
14 published pieces of writing (far less than I’d like)
6 podcast interviews
5 conference talks
5 Future of Coding events
5 driving lessons
1 driving test cancelled (I still haven’t booked another)
2 weddings

I’m not sure these capture anything, but they give some shape to the past 12 months. For the most part, it was a stressful and sad year. I split up with my wonderful, long-term partner at the start of it. I want children, and he didn’t. It’s the one thing you can’t compromise on. So I spent the following months being a pathetically heartbroken lump. I wallowed in work and distracted myself with travel and conferences and podcasts and excessively long park walks with friends.

But somehow, miraculously, just like the last 15 minutes of a cheaply made Netflix romcom, everything worked out in the final stretch. In our months apart my partner did a lot of reflection. He changed his mind and realised he does want a family. Knowing that you want to take on that level of stress, responsibility, and sleep deprivation isn’t a small question, and one everyone comes to in their own time. I wasn’t entirely sure I wanted children until my very late twenties. And it certainly would have been helpful to realise earlier…

Anyway, we got back together. And it’s been pretty happily ever after since. We even put a ring on it to secure the ever after part 😉

September 2023

London flipped from sweltering summer back to drizzling grey yesterday. Autumn was overdue. Summer was a little too manic for my taste anyway. I spent a bunch of time in Maine, New York, and San Fransisco. A mix of long back porch hangouts with family, collaborator coffees, and in-person problem-solving with the Ought team.

But the Ought team are no longer the Ought team. We’ve changed names! And whole companies come to think of it. Ought as a legal entity was a non-profit research lab. But we found it difficult to get enough funding from the charitable space to hire people and grow Elicit , our primary product. Our founders decided to pull the same move as OpenAI and Anthropic and flipped to become a VC-backed startup. The new company is just called Elicit. We raised a seed round and rebuilt the product.

We’re slowly moving our 200,000 active users over to the new beta version. We shipped it while we’re still deeply embarrassed by it. It has a thousand big and small UX cuts that fill me with sadness. Every morning I get on and hack away at them, trusting at some point it will feel beautiful to use.

My work on Elicit and explorations of Language Model interfaces have compounded into some strong opinions. I gathered some of them into a new talk called Squish Meets Structure . I did one round of it for Smashing Meets AI , and then drastically improved it for Smashing Conference in Freiburg, Germany a few weeks ago.

I’m headed to Web Directions in Sydney, Australia and FFConf in Brighton later this year to present my slightly more macabre talk on the Expanding Dark Forest and Generative AI . Expect to be all talked out after that.

Currently taking in writing advice from men named Ste(v/ph)en. Working through Steven Pinker’s “ The Sense of Style ” and Stephen King’s “ On Writing ” – both excellent. The former for grounded advice on non-fiction communication, the latter for an extraordinary account of persistently writing through double-shift minimum wage jobs and a raging alcohol and drug problem.

Spent the summer discovering how much I love dresses and jumpsuits with deep pockets. Found some of my favourites from Boden , Ro&Zo , Whistles , and Thought .

I’ve realised I spent the last decade avoiding femme clothing, aesthetics, and mannerisms in an attempt to be taken more seriously in the all-male spaces I hang out in. Spaces where I am routinely treated like a small, insignificant girl, ignored, and spoken over. Feels pretty freeing to give fewer shits and permanently live in dresses. They’ll ignore me either way.

June 2023

The last six months have felt like a disorientating blur. While bouncing between overseas trips, writing conference talks, organising events, and keeping up with work, I ended a long-term relationship and moved across London. My writing took a hit – I haven’t published much this year. I feel like I’m still trying to figure out what to do when I get up every morning.

Having everything thrown out of whack is pretty useful. You figure out how resilient your systems are. You learn what you have committed to muscle memory and what you rely on cues from your environment to do.

I was also reminded how much I value my extended friend network in London (and the rest of Europe). My social groups here often feel like concentric overlapping circles – I can’t neatly distinguish them. Which feels right; like a small village where everyone knows everyone. My philosophy friends show up at the Future of Coding meetups I run. My Twitter friends appear at dinners with my more woo creative people. It’s difficult to beat the sheer density of curious, friendly, ambitious people in a big city like London. It is more than worth the price of rent IMHO.

It doesn’t help that everything in the world of language models and generative AI has been moving at warp speed this year. I’m reading notes I took back in October of 2022 on GPT-3 and truthfulness, and it feels like we’re in a completely different world now. The benchmarks are out of date. The prompting techniques are more sophisticated. The real-world use cases have proliferated.

I keep wanting to throw my hat into the discourse ring – I certainly have formed opinions at this point! But there’s so much noise it’s difficult to determine if you’re adding any signal. I’m being cautious about it. Being a careful thinker feels like a rare quality at this point; I still aspire to it.

Some good reads as of late:

Bought myself used copies of Christopher Alexander’s classics. Feels much better to hold these in your hands than scroll the PDFs.

November 2022

It’s a weird week. Twitter is (possibly) entering a slow death march and much of my community is reluctantly slouching over to Mastodon. I’ve started cross-posting there as well: indieweb.social/@maggie . The Elon-escapades only drive home how important it is to own and control the platform you’re publishing to. Which means writing on your own website first and cross-posting to external platforms. The IndieWeb community has a catchy acronym for this: POSSE . Publish (on your) Own Site, Syndicate Elsewhere.

I still don’t have a Twitter-like stream setup on this site, but I’m working on it. Tools like WebMentions and Bridgy are going to do a lot of the heavy lifting. I’m having trouble finding many tutorials or guides on how to do it though – perhaps another case of writing the piece you want to read.

I’ve been working at Ought since late August. The London ↔ SF time difference is a bit crap, but I can’t say enough good things about my fellow Oughters 🦦. It’s rare and special to work alongside such thoughtful and ambitious people. We’re in the middle of designing and prototyping some wildly different interfaces for Elicit , our AI research assistant and Ought’s main product. Users say they love the current version, but we know it could be so much better.

In other news, I bought a cheap, used laser printer and it’s drastically improved my reading experience for long articles and PDFs. Can’t recommend it enough. There are plenty going for £30-40 on eBay and you’ll finally read all those 5000-word blog posts you have bookmarked.

I’ve finally gotten back into a good writing cadence. I recently published Programming Portals and Command K Bars . Now I’m slowly chipping away at Tools for Thought as Cultural Practices, Not Computational Objects . I’ve been writing it forever, but that’s kind of the point of gardening in public. You post up loose skeletons of ideas and flesh them out over time. Nothing forces you to write like knowing people are reading your half-baked version.

July 2022

I’ve decided to join Ought , a non-profit research lab. I wrote about the move here , but TLDR: Ought is exploring how machine learning techniques can support open-ended reasoning and research. At the moment their main focus is building Elicit an AI research assistant that helps automate literature reviews. The long-term vision is to develop tools and techniques that mitigate AI alignment risks.

They’re based in the Bay Area of California, so I’ll be on flights over the Atlantic more often. London is still my home base for the long-term future. I’ll spare you my gauche love letter to this city, but leaving only makes me realise how deeply I appreciate it. Stepping back onto TFL platforms, Sainsbury’s aisles, and hodge-podge pavements feel like breathing again. Perhaps it’s just a matter of contrasts though; San Fransisco’s grimey vibes are not for me.

I’ve started hosting a monthly event for the Future of Code community here. I’m keeping it low-key for the moment. Meaning it’s just thirty people in a pub talking shop about CRDT’s, spatial canvases, and embodied cognition. My kind of Friday night.

We’ve got an exceptional class of people coming at the moment. I’m thinking hard about how to preserve this early group energy; keeping it small, not promoting it too much (lest the recruiters and VC’s get wind of the event), and maintaining the “don’t take this too seriously” vibes. Most other tech meetups in London turn out to be overstuffed, impersonal networking evenings. They’re held in some fluorescently lit office where a handful of people give lightweight talks over lukewarm pizza. Everyone wants to escape home as soon as they’ve arrived. Avoiding this vibe is my first priority.

October 2021

After five years of working with egghead , I decided it was time to move on. I started as an illustrator in early 2016. By 2018 I had moved into an art director and leadership role. By 2020 I’d shifted to focusing on UX and product design.

I grew up in this company, but over the last year it’s become clear I need/want to engage more in the rising “tools for thought” ecosystem. I’ve now joined HASH as the design lead. We’re working on building a knowledge management platform (aka. “note-taking app”) backed by schemas and structured data.

August 2021

With lockdown easing in London I’ve spent most of summer outside throwing tennis balls around in the park, rather than sitting in front of screens writing words. It’s been a good break, but I’ve missed researching. The British winter will descend again soon and I’ll be back inside tending to the digital garden.

Dan Abramov and I finally launched Just JavaScript after 2 years of work. It’s an interactive book that teaches JavaScript with a specific focus on mental models. I created the diagrams and animations that help communicate the core concepts. We created it with the support of the egghead team who took care of logistics like email promotion, web infrastructure, and customer support.

I’ve become enamoured with Visual Programming and End-user Programming over the last 6 months. The Future of Coding community and Gordon Brandon ’s newsletter have been particularly valuable hubs of insight.

Chatted with a few podcast folks;

  1. Adam Wiggins and Mark McGranaghan on Metamuse about visual programming.
  2. Scott Francis and Evan Payne on Frontend Heroes about a little bit of everything

Currently preparing a talk for two conferences in October; React Live Conference in Amsterdam and React Advanced here in London. On various ways to visualise React, why visual representations work so well, the pros / cons / wicked historical problems of adding graphical and spatial affordances to programming interfaces, and why we should keep trying despite the challenges.

September 2020

Digging into the work of Barbara Tversky and reading the literature around embodied cognition . I’ve been a long time fan of George Lakoff so much of it us familiar. Tverksy’s new book Mind in Motion builds off many of the same themes.

Delaying a part-time Masters degree in Digital Anthropology at UCL until next year when we’ll hopefully be less Covid-y

Reading Working in Public by Nadia Eghbal

Exploring the connections between technology and magic through the writings of Alfred Gell . Specifically enjoyed the paper The Technology of Enchantment and the Enchantment of Technology

Playing Thing From the Future – A game that challenges players to collaboratively and competitively describe objects from a range of alternative futures.