Full Width [alt+shift+f] Shortcuts [alt+shift+k]
Sign Up [alt+shift+s] Log In [alt+shift+l]

New here?

Welcome! BoredReading is a fresh way to read high quality articles (updated every hour). Our goal is to curate (with your help) Michelin star quality articles (stuff that's really worth reading). We currently have articles in 0 categories from architecture, history, design, technology, and more. Grab a cup of freshly brewed coffee and start reading. This is the best way to increase your attention span, grow as a person, and get a better understanding of the world (or atleast that's why we built it).

17
We saw this Monarch butterfly caterpillar at the pretty unusual Naval Cemetery Landscape. The landscape is just native pollinators and native plants growing wild, with a wood platform above the field so you can walk around and see the bugs and plants. It’s also built on a cemetery that was moved to another site, but they didn’t move all the bodies, so there are still some unaccounted for in the park. Watching I know that every generation looks at the younger generation and fears that their cultural memory is getting shorter, but isn’t it? Adam Neely, who I’ve mentioned before when he talked about the ‘Sea Shanty’ trend on TikTok, came out with this great breakdown of Laufey, jazz, and the dangers of rewriting history. Speaking of jazz, found GoGo Penguin, which is not quite jazz - if something like Mammal Hands are halfway between jazz and electronica, GoGo Penguin is 75% of the way to electronica. But I love it anyway: Like everyone, I’ve been watching The Bear, and I’ve cried...
a year ago

Improve your reading experience

Logged in users get linked directly to articles resulting in a better reading experience. Please login for free, it takes less than 1 minute.

More from macwright.com

Introducing the blogroll

This website has a new section: blogroll.opml! A blogroll is a list of blogs - a lightweight way of people recommending other people’s writing on the indieweb. What it includes The blogs that I included are just sampled from my many RSS subscriptions that I keep in my Feedbin reader. I’m subscribed to about 200 RSS feeds, the majority of which are dead or only publish once a year. I like that about blogs, that there’s no expectation of getting a post out every single day, like there is in more algorithmically-driven media. If someone who I interacted with on the internet years ago decides to restart their writing, that’s great! There’s no reason to prune all the quiet feeds. The picks are oriented toward what I’m into: niches, blogs that have a loose topic but don’t try to be general-interest, people with distinctive writing. If you import all of the feeds into your RSS reader, you’ll probably end up unsubscribing from some of them because some of the experimental electric guitar design or bonsai news is not what you’re into. Seems fine, or you’ll discover a new interest! How it works Ruben Schade figured out a brilliant way to show blogrolls and I copied him. Check out his post on styling OPML and RSS with XSLT to XHTML for how it works. My only additions to that scheme were making the blogroll page blend into the rest of the website by using an include tag with Jekyll to add the basic site skeleton, and adding a link with the download attribute to provide a simple way to download the OPML file. Oddly, if you try to save the OPML page using Save as… in Firefox, Firefox will save the transformed output via the XSLT, rather than the raw source code. XSLT is such an odd and rare part of the web ecosystem, I had to use it.

yesterday 2 votes
Recently

I have a non-recently post ready to write, any day now… Reading This was a strong month for reading: I finished The Hidden Wealth of Nations, Useful Not True, and Cyberlibertarianism. I had a book club that read Cyberlibertarianism so we discussed it last week. I have a lot of qualms with the book, and gave it two stars for that reason. But I will admit that it’s taking up space in my mind. The ‘cyberlibertarian’ ideology was familiar to me before reading it. The book’s critique of it didn’t shift my thinking that much. But I have been thinking a lot about what it argued for, which is a world in which the government has very extensive powers – to limit what is said online, to regulate which companies can even create forums or social media platforms. He also believed that a government should be able to decrypt and read conversations between private citizens. It’s a very different idea of government power than what I’m used to, and well outside my comfort zone. I think it’s interesting to consider these things: the government probably should have some control of some kinds of speech, and in some cases it’s useful to have the FBI tapping the phones of drug smugglers or terrorists. How do we really define what’s acceptable and what isn’t? I don’t know, I want to do more thinking about the uncomfortable things that nevertheless may be necessary for functioning of society. Besides that, there is so much to read. This month I added a lot of news subscriptions to my pile, which I think is now Hell Gate, Wired, NYTimes, Bloomberg, 404 Media, The Verge, and a bunch of newsletters. This interview with Stephanie Kelton, who is at the forefront of the Modern Monetary Theory movement in America, and wrote the very good book The Deficit Myth. This 404 Media story on an AI-generated ‘true crime’ YouTube channel is great because the team at 404 Media does both deep research and they interrogate their sources. Nathan Tankus has always been good but in this era he’s essential reading. His piece on Fort Knox is quick and snappy. His others are more involved but always worth reading. Listening We’ve been rewatching The Bear and admiring the dad-rock soundtrack. This Nine Inch Nails track shows up at the end of a season: And this Eno track: Besides that, this track from Smino played at a local cocktail bar. The bars at 0:45 sound like they’re tumbling downhill in a delightful way. Watching So I bought a sewing machine in February, a beautiful old Kenmore 158-series, produced in the 1970s in Japan. It’s awesome. How sewing machines work is amazing, as this video lays out. There’s so much coordinated motion happening for every stitch, and the machines are so well-designed that they last for decades easily. Besides that, I just watched The Apprentice, which I really did not like. Elsewhere I was on a podcast with Jeremy Jung, taking about Placemark! My post in the /micro/ section, All Hat No Cowboy, probably could have or should have been a blog post, but I was feeling skittish about being too anti-AI on the main.

a week ago 7 votes
Recently

I am not going to repeat the news. But man, things are really, really bad and getting worse in America. It’s all so unendingly stupid and evil. The tech industry is being horrible, too. Wishing strength to the people who are much more exposed to the chaos than I am. Reading A Confederacy of Dunces was such a perfect novel. It was pure escapism, over-the-top comedy, and such an unusual artifact, that was sadly only appreciated posthumously. Very earnestly I believe that despite greater access to power and resources, the box labeled “socially acceptable ways to be a man” is much smaller than the box labeled “socially acceptable ways to be a woman.” This article on the distinction between patriarchy and men was an interesting read. With the whole… politics out there, it’s easy to go off the rails with any discussion about men and women and whether either have it easy or hard. The same author wrote this good article about declining male enrollment in college. I think both are worth a read. Whenever I read this kind of article, I’m reminded of how limited and mostly fortunate my own experience is. There’s a big difference, I think, in how vigorously you have to perform your gender in some red state where everyone owns a pickup truck, versus a major city where the roles are a little more fluid. Plus, I’ve been extremely fortunate to have a lot of friends and genuine open conversations about feelings with other men. I wish that was the norm! On Having a Maximum Wealth was right up my alley. I’m reading another one of the new-French-economist books right now, and am still fascinated by the prospect of wealth taxes. My friend David has started a local newsletter for Richmond, Virginia, and written a good piece about public surveillance. Construction Physics is consistently great, and their investigation of why skyscrapers are all glass boxes is no exception. Watching David Lynch was so great. We watched his film Lost Highway a few days after he passed, and it was even better than I had remembered it. Norm Macdonald’s extremely long jokes on late-night talk shows have been getting me through the days. Listening This song by the The Hard Quartet – a supergroup of Emmett Kelly, Stephen Malkmus (Pavement), Matt Sweeney and Jim White. It’s such a loving, tender bit of nonsense, very golden-age Pavement. They also have this nice chill song: I came across this SML album via Hearing Things, which has been highlighting a lot of good music. Small Medium Large by SML It’s a pretty good time for these independent high-quality art websites. Colossal has done the same for the art world and highlights good new art: I really want to make it out to see the Nick Cave (not the musician) art show while it’s in New York.

a month ago 14 votes
2025 Predictions

I was just enjoying Simon Willison’s predictions and, heck, why not. 1: The web becomes adversarial to AI The history of search engines is sort of an arms race between websites and search engines. Back in the early 2000s, juicing your ranking on search engines was pretty easy - you could put a bunch of junk in your meta description tags or put some text with lots of keywords on each page and make that text really tiny and transparent so users didn’t notice it but Google did. I doubt that Perplexity’s userbase is that big but Perplexity users are probably a lot wealthier on average than Google’s, and there’s some edge to be achieved by getting Perplexity to rank your content highly or recommend your website. I’ve already noticed some search results including links to content farms. There are handful of startups that do this already, but the prediction is: the average marketing exec at a consumer brand will put some of their budget to work on fooling AI. That means serving different content to AI scrapers, maybe using some twist on Glaze and other forms of adversarial image processing to make their marketing images more tantalizing to the bots. Websites will be increasingly aware that they’re being consumed by AI, and they will have a vested interest in messing with the way AI ‘perceives’ them. As Simon notes in his predictions, AIs are gullible: and that’s before there are widespread efforts to fool them. There’s probably some way to detect an AI scraper, give it a special payload, and trick it into recommending your brand of razors whenever anyone asks, and once someone figures it out this will be the marketing trend of the decade. 2: Copyright nihilism breeds a return to physical-only media The latest lawsuit about Meta’s use of pirated books, allegedly with Mark Zuckerberg’s explicit permission, if true, will be another reason to lose faith in the American legal system’s intellectual property system entirely. We’ve only seen it used to punish individuals and protect corporations, regardless of the facts and damages, and there’s no reason to believe it will do anything different (POSIWID). The result, besides an uptick in nihilism, could be a rejuvenation of physical-only releases. New albums only released on vinyl. Books only available in paperback format. More private screenings of hip movies. When all digital records are part of the ‘training dataset,’ a niche, hipster subset will be drawn to things that aren’t as easily captured and reproduced. This is parallel, to the state of closed-source models from Anthropic or OpenAI. They’re never distributed or run locally. They exist as bytes on some hard drive and in some massive GPU’s memory in some datacenter, and there aren’t Bittorrents pirating them because they’re kept away from people, not because of the power of copyright law. What can be accessed can be copied, so secrecy and inaccessibility is valuable. 3: American tech companies will pull out of Europe because they want to do acquisitions The incoming political administration will probably bring an end to Lina Khan’s era of the FTC, and era in which the FTC did stuff. We will go back to a ‘hands off’ policy in which big companies will acquire each other pretty often without much government interference. But, even in Khan’s era, the real nail in the coffin for one of the biggest acquisitions - Adobe’s attempt to buy Figma – was regulators from the EU and UK. Those regulators will probably keep doing stuff, so I think it’s likely that the next time some company wants to acquire a close competitor, they just close up shop in the EU, maybe with a long-term plan to return. 4: The tech industry’s ‘DEI backlash’ will run up against reality The reality is that the gap between women and men in terms of college degrees is really big: “Today, 47% of U.S. women ages 25 to 34 have a bachelor’s degree, compared with 37% of men.” And that a great deal of the tech industry’s workforce is made of up highly-skilled people who are on H-1B visas. The synthesis will be that tech workers will be more diverse, in some respects, but by stripping away the bare-bones protections around their presence, companies will keep them in a more vulnerable and exploitable position. But hard right-wingers will have plenty to complain about because these companies will continue to look less white and male, because the labor pool is not that. 5: Local-first will have a breakthrough moment I think that Zero Sync has a good chance at cracking this really hard problem. So does electric and maybe jazz, too. The gap between the dream of local-first apps and the reality has been wide, but I think projects are starting to come to grips with a few hard truths: Full decentralization is not worth it. You need to design for syncing a subset of the data, not the entire dataset. You need an approach to schema evolution and permission checking These systems are getting there. We could see a big, Figma-level application built on Zero this year that will set the standard for future web application architecture. 6: Local, small AI models will be a big deal Embedding models are cool as heck. New text-to-speech and speech-to-text models are dramatically better than what came before. Image segmentation is getting a lot better. There’s a lot of stuff that is coming out of this boom that will be able to scale down to a small model that runs on a phone, browser, or at least on our own web servers without having to call out to OpenAI or Anthropic APIs. It’ll make sense for costs, performance, and security. Candle is a really interesting effort in this area. Mini predictions Substack will re-bundle news. People are tired of subscribing to individual newsletters. Substack will introduce some ~$20/month plan that gives you access to all of the newsletters that participate in this new pricing model. TypeScript gets a zeitwork equivalent and lots of people use it. Same as how prettier brought full code formatting from TypeScript, autoloading is the kind of thing that once you have it, it’s magic. What if you could just write <SomeComponent /> in your React app and didn’t have to import it? I think this would be extremely addictive and catch on fast. Node.js will fend off its competitors. Even though Val Town is built around Deno’s magic, I’ve been very impressed that Node.js is keeping up. They’ve introduced permissions, just like Deno, and native TypeScript support, just like the upstarts. Bun and Deno will keep gaining adherents, but Node.js has a long future ahead of it. Another US city starts seriously considering congestion pricing. For all the chatter and terrible discourse around the plan, it is obviously a good idea and it will work, as it has in every other case, and inspire other cities to do the same. Stripe will IPO. They’re still killing it, but they’re killing it in an established, repeatable way that public markets will like, and will let up the pressure on the many, many people who own their stock.

a month ago 41 votes
Recently 2024

Happy end-of-2024! It’s been a pretty good year overall. I’m thankful. There’s no way that I’ll be able to remember and carve out the time around New Years to write this, so here’s some end-of-year roundup, ahead of schedule! Running This was my biggest year for running on record: 687 miles as of today. I think the biggest difference with this year was just that nothing stood in the way of my being pretty consistent and putting in the miles: the weather has been mild, I haven’t had any major injuries, and long runs have felt pretty good. I was happy to hit a half-marathon PR (1:36:21), but my performance in 5Ks was far short of the goal of sub-20 – partly because Brooklyn’s wonderful 5K series was run at the peak of summer, with multiple races at over 85°F. I learned the value of good lightweight running gear: Bakline’s singlets and Goodr sunglasses were super helpful in getting me through the summer. Work Val Town raised a seed round and hired a bunch of excellent people. We moved into a new office of our own, which has a great vibe. It’s been good: we’re doing a lot of ground-up work wrangling cgroups and low-level worker scheduling, and a lot of UX-in work, just trying to make it a pleasant tool. Frankly, with every product I’ve worked on, I’ve never had a feeling that it was good enough, and accordingly, for me, Val Town feels like it has a long way to go. It’s probably a good tendency to be sort of unsatisfied and motivated to constantly improve. New York It’s still such a wonderful place to live. Late this year, I’ve been rediscovering my obsession with cycling, and realizing how much I whiffed the opportunity to ride more when I lived in San Francisco. I guess that’s the first time I felt genuinely nostalgic for the West coast. I miss DC a bit too: it’s one of the few cities where my friends have been able to stay in the city proper while raising children, and I miss the accessible, underdog punk scene. But Brooklyn is just a remarkable place to live. My walk score is 100. The degree to which people here are in the city because they want to be, not because they have to, shapes so much of what makes it great. Other ‘metrics’ Relative to my old level obsession about self-quantification, my ‘metrics’ are pretty moderate now. Everything’s just backward-looking: I’m not paying much attention to the numbers as I go, it’s just fun to look at them year-over-year trends. That said, this was a lackluster year for reading: just 18 books so far. I think I just read an above-average number of books that I didn’t enjoy very much. Next year I’m going to return to authors who I already love, and stay away from genres that – the data shows – I don’t like. Whereas this was a banner year for watching movies: not great! Next year, I want to flip these results. Of everything I saw, Kinds of Kindness will probably stick with me the most. Placemark It seems like a decade ago that I released Placemark as open source software, as developing it as a closed-source SaaS application for a few years. But I did that in January. There have been a few great open source contributions since then, but it’s pretty quiet. Which is okay, somewhat expected: there is no hidden crowd of people with extra time on their hands and unending enthusiasm for ‘geospatial software’ waiting to contribute to that kind of project. Placemark is also, even with my obsessive focus on simplicity, a pretty complicated codebase. The learning curve is probably pretty significant. Maps are a challenging problem area: that’s what attracts a lot of people to them, but people who use maps persistently have the feeling that it couldn’t be that complicated, which means that few users convert into contributors. There are a few prominent efforts chasing similar goals as Placemark: Atlas.co is aiming to be an all-in-one editing/analysis platform, Felt a cloud-native GIS platform, and then there are plenty of indiehackers-style projects. I hope these projects take off! Figma plugins I also kept maintaining the Figma plugins I developed under the Placemark name. Potentially a lot of people are using them, but I don’t really know. The problem with filling in water shapes in the plugins is still unsolved: it’s pretty hard and I haven’t had the time or motivation to fix it. The most energy into those plugins this year, unfortunately, was when someone noticed that the dataset I was using - Natural Earth – marked Crimea as part of Russia. Which obviously: I don’t draw the countries in datasets, but it’s a reasonable thing to point out (but to assume that the author is malicious was a real downer, again, like, I don’t draw the countries). This decision from Natural Earth’s maintainer is heavily discussed and they aren’t planning on changing it, so I switched to world-atlas, which doesn’t have that problem. Which was fine, but a reminder of the days when I worked on maps full-time and this kind of unexpected “you’re the baddie” realization came up much more often. Sometimes it was silly: people who complain about label priority in the sense of “why, at zoom level 3, does one country’s name show up and not anothers?” was just silly. The answer, ahem, was that there isn’t enough space for the two labels and one country had a higher population or a geometry that gave their label more distance from the other country’s centroid. But a lot of the territorial disputes are part of people’s long cultural, political, military history and the source of intergenerational strife. Of course that’s serious stuff. Making a tool that shows a globe with labels on it will probably always trigger some sort of moment like that, and it’s a reason to not work on it that much because you’re bound to unintentionally step on something contentious. Other projects I released Obsidian Freeform, and have been using it a bit myself. Obsidian has really stuck for me. My vault is well over 2,000 notes, and I’ve created a daily note for almost every day for the last year. Freeform was a fun project and I have other ideas that are Obsidian plugin-shaped, though I’ve become a little bit let down by the plugin API - the fact that Obsidian-flavored-Markdown is nonstandard and the parser/AST is not accessible to plugins is a pretty big drawback for the kinds of things I want to build. Elsewhere recently I’ve been writing a bit: Recently I’ve written about dependency bloat and a developer analytics tool we built at Val Town, and started writing some supplementary documentation for Observable Plot about parts of its API that I think are unintuitive. On the micro blog, I wrote about not using GitHub Copilot and how brands should make a comeback. This blog got a gentle redesign in May, to show multiple categories of posts on the home page, and then in August I did a mass update to switch all YouTube embeds to lite-youtube-embed to make pages load faster. I’m still running Jekyll, like I have been for the last decade, and it works great. Oh, and I’ve basically stopped using Twitter and am only on Mastodon and Bluesky. Bluesky more than Mastodon recently because it seems like it’s doing a better job at attracting a more diverse community. I’m looking forward to 2025, to cycling a lot more and a new phase of startup-building. See you in the new year.

2 months ago 69 votes

More in programming

AI: Where in the Loop Should Humans Go?

This is a re-publishing of a blog post I originally wrote for work, but wanted on my own blog as well. AI is everywhere, and its impressive claims are leading to rapid adoption. At this stage, I’d qualify it as charismatic technology—something that under-delivers on what it promises, but promises so much that the industry still leverages it because we believe it will eventually deliver on these claims. This is a known pattern. In this post, I’ll use the example of automation deployments to go over known patterns and risks in order to provide you with a list of questions to ask about potential AI solutions. I’ll first cover a short list of base assumptions, and then borrow from scholars of cognitive systems engineering and resilience engineering to list said criteria. At the core of it is the idea that when we say we want humans in the loop, it really matters where in the loop they are. My base assumptions The first thing I’m going to say is that we currently do not have Artificial General Intelligence (AGI). I don’t care whether we have it in 2 years or 40 years or never; if I’m looking to deploy a tool (or an agent) that is supposed to do stuff to my production environments, it has to be able to do it now. I am not looking to be impressed, I am looking to make my life and the system better. Another mechanism I want you to keep in mind is something called the context gap. In a nutshell, any model or automation is constructed from a narrow definition of a controlled environment, which can expand as it gains autonomy, but remains limited. By comparison, people in a system start from a broad situation and narrow definitions down and add constraints to make problem-solving tractable. One side starts from a narrow context, and one starts from a wide one—so in practice, with humans and machines, you end up seeing a type of teamwork where one constantly updates the other: The optimal solution of a model is not an optimal solution of a problem unless the model is a perfect representation of the problem, which it never is.  — Ackoff (1979, p. 97) Because of that mindset, I will disregard all arguments of “it’s coming soon” and “it’s getting better real fast” and instead frame what current LLM solutions are shaped like: tools and automation. As it turns out, there are lots of studies about ergonomics, tool design, collaborative design, where semi-autonomous components fit into sociotechnical systems, and how they tend to fail. Additionally, I’ll borrow from the framing used by people who study joint cognitive systems: rather than looking only at the abilities of what a single person or tool can do, we’re going to look at the overall performance of the joint system. This is important because if you have a tool that is built to be operated like an autonomous agent, you can get weird results in your integration. You’re essentially building an interface for the wrong kind of component—like using a joystick to ride a bicycle. This lens will assist us in establishing general criteria about where the problems will likely be without having to test for every single one and evaluate them on benchmarks against each other. Questions you'll want to ask The following list of questions is meant to act as reminders—abstracting away all the theory from research papers you’d need to read—to let you think through some of the important stuff your teams should track, whether they are engineers using code generation, SREs using AIOps, or managers and execs making the call to adopt new tooling. Are you better even after the tool is taken away? An interesting warning comes from studying how LLMs function as learning aides. The researchers found that people who trained using LLMs tended to fail tests more when the LLMs were taken away compared to people who never studied with them, except if the prompts were specifically (and successfully) designed to help people learn. Likewise, it’s been known for decades that when automation handles standard challenges, the operators expected to take over when they reach their limits end up worse off and generally require more training to keep the overall system performant. While people can feel like they’re getting better and more productive with tool assistance, it doesn’t necessarily follow that they are learning or improving. Over time, there’s a serious risk that your overall system’s performance will be limited to what the automation can do—because without proper design, people keeping the automation in check will gradually lose the skills they had developed prior. Are you augmenting the person or the computer? Traditionally successful tools tend to work on the principle that they improve the physical or mental abilities of their operator: search tools let you go through more data than you could on your own and shift demands to external memory, a bicycle more effectively transmits force for locomotion, a blind spot alert on your car can extend your ability to pay attention to your surroundings, and so on. Automation that augments users therefore tends to be easier to direct, and sort of extends the person’s abilities, rather than acting based on preset goals and framing. Automation that augments a machine tends to broaden the device’s scope and control by leveraging some known effects of their environment and successfully hiding them away. For software folks, an autoscaling controller is a good example of the latter. Neither is fundamentally better nor worse than the other—but you should figure out what kind of automation you’re getting, because they fail differently. Augmenting the user implies that they can tackle a broader variety of challenges effectively. Augmenting the computers tends to mean that when the component reaches its limits, the challenges are worse for the operator. Is it turning you into a monitor rather than helping build an understanding? If your job is to look at the tool go and then say whether it was doing a good or bad job (and maybe take over if it does a bad job), you’re going to have problems. It has long been known that people adapt to their tools, and automation can create complacency. Self-driving cars that generally self-drive themselves well but still require a monitor are not effectively monitored. Instead, having AI that supports people or adds perspectives to the work an operator is already doing tends to yield better long-term results than patterns where the human learns to mostly delegate and focus elsewhere. (As a side note, this is why I tend to dislike incident summarizers. Don’t make it so people stop trying to piece together what happened! Instead, I prefer seeing tools that look at your summaries to remind you of items you may have forgotten, or that look for linguistic cues that point to biases or reductive points of view.) Does it pigeonhole what you can look at? When evaluating a tool, you should ask questions about where the automation lands: Does it let you look at the world more effectively? Does it tell you where to look in the world? Does it force you to look somewhere specific? Does it tell you to do something specific? Does it force you to do something? This is a bit of a hybrid between “Does it extend you?” and “Is it turning you into a monitor?” The five questions above let you figure that out. As the tool becomes a source of assertions or constraints (rather than a source of information and options), the operator becomes someone who interacts with the world from inside the tool rather than someone who interacts with the world with the tool’s help. The tool stops being a tool and becomes a representation of the whole system, which means whatever limitations and internal constraints it has are then transmitted to your users. Is it a built-in distraction? People tend to do multiple tasks over many contexts. Some automated systems are built with alarms or alerts that require stealing someone’s focus, and unless they truly are the most critical thing their users could give attention to, they are going to be an annoyance that can lower the effectiveness of the overall system. What perspectives does it bake in? Tools tend to embody a given perspective. For example, AIOps tools that are built to find a root cause will likely carry the conceptual framework behind root causes in their design. More subtly, these perspectives are sometimes hidden in the type of data you get: if your AIOps agent can only see alerts, your telemetry data, and maybe your code, it will rarely be a source of suggestions on how to improve your workflows because that isn’t part of its world. In roles that are inherently about pulling context from many disconnected sources, how on earth is automation going to make the right decisions? And moreover, who’s accountable for when it makes a poor decision on incomplete data? Surely not the buyer who installed it! This is also one of the many ways in which automation can reinforce biases—not just based on what is in its training data, but also based on its own structure and what inputs were considered most important at design time. The tool can itself become a keyhole through which your conclusions are guided. Is it going to become a hero? A common trope in incident response is heroes—the few people who know everything inside and out, and who end up being necessary bottlenecks to all emergencies. They can’t go away for vacation, they’re too busy to train others, they develop blind spots that nobody can fix, and they can’t be replaced. To avoid this, you have to maintain a continuous awareness of who knows what, and crosstrain each other to always have enough redundancy. If you have a team of multiple engineers and you add AI to it, having it do all of the tasks of a specific kind means it becomes a de facto hero to your team. If that’s okay, be aware that any outages or dysfunction in the AI agent would likely have no practical workaround. You will essentially have offshored part of your ops. Do you need it to be perfect? What a thing promises to be is never what it is—otherwise AWS would be enough, and Kubernetes would be enough, and JIRA would be enough, and the software would work fine with no one needing to fix things. That just doesn’t happen. Ever. Even if it’s really, really good, it’s gonna have outages and surprises, and it’ll mess up here and there, no matter what it is. We aren’t building an omnipotent computer god, we’re building imperfect software. You’ll want to seriously consider whether the tradeoffs you’d make in terms of quality and cost are worth it, and this is going to be a case-by-case basis. Just be careful not to fix the problem by adding a human in the loop that acts as a monitor! Is it doing the whole job or a fraction of it? We don’t notice major parts of our own jobs because they feel natural. A classic pattern here is one of AIs getting better at diagnosing patients, except the benchmarks are usually run on a patient chart where most of the relevant observations have already been made by someone else. Similarly, we often see AI pass a test with flying colors while it still can’t be productive at the job the test represents. People in general have adopted a model of cognition based on information processing that’s very similar to how computers work (get data in, think, output stuff, rinse and repeat), but for decades, there have been multiple disciplines that looked harder at situated work and cognition, moving past that model. Key patterns of cognition are not just in the mind, but are also embedded in the environment and in the interactions we have with each other. Be wary of acquiring a solution that solves what you think the problem is rather than what it actually is. We routinely show we don’t accurately know the latter. What if we have more than one? You probably know how straightforward it can be to write a toy project on your own, with full control of every refactor. You probably also know how this stops being true as your team grows. As it stands today, a lot of AI agents are built within a snapshot of the current world: one or few AI tools added to teams that are mostly made up of people. By analogy, this would be like everyone selling you a computer assuming it were the first and only electronic device inside your household. Problems arise when you go beyond these assumptions: maybe AI that writes code has to go through a code review process, but what if that code review is done by another unrelated AI agent? What happens when you get to operations and common mode failures impact components from various teams that all have agents empowered to go fix things to the best of their ability with the available data? Are they going to clash with people, or even with each other? Humans also have that ability and tend to solve it via processes and procedures, explicit coordination, announcing what they’ll do before they do it, and calling upon each other when they need help. Will multiple agents require something equivalent, and if so, do you have it in place? How do they cope with limited context? Some changes that cause issues might be safe to roll back, some not (maybe they include database migrations, maybe it is better to be down than corrupting data), and some may contain changes that rolling back wouldn’t fix (maybe the workload is controlled by one or more feature flags). Knowing what to do in these situations can sometimes be understood from code or release notes, but some situations can require different workflows involving broader parts of the organization. A risk of automation without context is that if you have situations where waiting or doing little is the best option, then you’ll need to either have automation that requires input to act, or a set of actions to quickly disable multiple types of automation as fast as possible. Many of these may exist at the same time, and it becomes the operators’ jobs to not only maintain their own context, but also maintain a mental model of the context each of these pieces of automation has access to. The fancier your agents, the fancier your operators’ understanding and abilities must be to properly orchestrate them. The more surprising your landscape is, the harder it can become to manage with semi-autonomous elements roaming around. After an outage or incident, who does the learning and who does the fixing? One way to track accountability in a system is to figure out who ends up having to learn lessons and change how things are done. It’s not always the same people or teams, and generally, learning will happen whether you want it or not. This is more of a rhetorical question right now, because I expect that in most cases, when things go wrong, whoever is expected to monitor the AI tool is going to have to steer it in a better direction and fix it (if they can); if it can’t be fixed, then the expectation will be that the automation, as a tool, will be used more judiciously in the future. In a nutshell, if the expectation is that your engineers are going to be doing the learning and tweaking, your AI isn’t an independent agent—it’s a tool that cosplays as an independent agent. Do what you will—just be mindful All in all, none of the above questions flat out say you should not use AI, nor where exactly in the loop you should put people. The key point is that you should ask that question and be aware that just adding whatever to your system is not going to substitute workers away. It will, instead, transform work and create new patterns and weaknesses. Some of these patterns are known and well-studied. We don’t have to go rushing to rediscover them all through failures as if we were the first to ever automate something. If AI ever gets so good and so smart that it’s better than all your engineers, it won’t make a difference whether you adopt it only once it’s good. In the meanwhile, these things do matter and have real impacts, so please design your systems responsibly. If you’re interested to know more about the theoretical elements underpinning this post, the following references—on top of whatever was already linked in the text—might be of interest: Books: Joint Cognitive Systems: Foundations of Cognitive Systems Engineering by Erik Hollnagel Joint Cognitive Systems: Patterns in Cognitive Systems Engineering by David D. Woods Cognition in the Wild by Edwin Hutchins Behind Human Error by David D. Woods, Sydney Dekker, Richard Cook, Leila Johannesen, Nadine Sarter Papers: Ironies of Automation by Lisanne Bainbridge The French-Speaking Ergonomists’ Approach to Work Activity by Daniellou How in the World Did We Ever Get into That Mode? Mode Error and Awareness in Supervisory Control by Nadine Sarter Can We Ever Escape from Data Overload? A Cognitive Systems Diagnosis by David D. Woods Ten Challenges for Making Automation a “Team Player” in Joint Human-Agent Activity by Gary Klein and David D. Woods MABA-MABA or Abracadabra? Progress on Human–Automation Co-ordination by Sidney Dekker Managing the Hidden Costs of Coordination by Laura Maguire Designing for Expertise by David D. Woods The Impact of Generative AI on Critical Thinking by Lee et al.

yesterday 4 votes
AMD YOLO

AMD is sending us the two MI300X boxes we asked for. They are in the mail. It took a bit, but AMD passed my cultural test. I now believe they aren’t going to shoot themselves in the foot on software, and if that’s true, there’s absolutely no reason they should be worth 1/16th of NVIDIA. CUDA isn’t really the moat people think it is, it was just an early ecosystem. tiny corp has a fully sovereign AMD stack, and soon we’ll port it to the MI300X. You won’t even have to use tinygrad proper, tinygrad has a torch frontend now. Either NVIDIA is super overvalued or AMD is undervalued. If the petaflop gets commoditized (tiny corp’s mission), the current situation doesn’t make any sense. The hardware is similar, AMD even got the double throughput Tensor Cores on RDNA4 (NVIDIA artificially halves this on their cards, soon they won’t be able to). I’m betting on AMD being undervalued, and that the demand for AI has barely started. With good software, the MI300X should outperform the H100. In for a quarter million. Long term. It can always dip short term, but check back in 5 years.

yesterday 2 votes
whippet lab notebook: untagged mallocs, bis

Earlier this weekGuileWhippet But now I do! Today’s note is about how we can support untagged allocations of a few different kinds in Whippet’s .mostly-marking collector Why bother supporting untagged allocations at all? Well, if I had my way, I wouldn’t; I would just slog through Guile and fix all uses to be tagged. There are only a finite number of use sites and I could get to them all in a month or so. The problem comes for uses of from outside itself, in C extensions and embedding programs. These users are loathe to adapt to any kind of change, and garbage-collection-related changes are the worst. So, somehow, we need to support these users if we are not to break the Guile community.scm_gc_malloclibguile The problem with , though, is that it is missing an expression of intent, notably as regards tagging. You can use it to allocate an object that has a tag and thus can be traced precisely, or you can use it to allocate, well, anything else. I think we will have to add an API for the tagged case and assume that anything that goes through is requesting an untagged, conservatively-scanned block of memory. Similarly for : you could be allocating a tagged object that happens to not contain pointers, or you could be allocating an untagged array of whatever. A new API is needed there too for pointerless untagged allocations.scm_gc_mallocscm_gc_mallocscm_gc_malloc_pointerless Recall that the mostly-marking collector can be built in a number of different ways: it can support conservative and/or precise roots, it can trace the heap precisely or conservatively, it can be generational or not, and the collector can use multiple threads during pauses or not. Consider a basic configuration with precise roots. You can make tagged pointerless allocations just fine: the trace function for that tag is just trivial. You would like to extend the collector with the ability to make pointerless allocations, for raw data. How to do this?untagged Consider first that when the collector goes to trace an object, it can’t use bits inside the object to discriminate between the tagged and untagged cases. Fortunately though . Of those 8 bits, 3 are used for the mark (five different states, allowing for future concurrent tracing), two for the , one to indicate whether the object is pinned or not, and one to indicate the end of the object, so that we can determine object bounds just by scanning the metadata byte array. That leaves 1 bit, and we can use it to indicate untagged pointerless allocations. Hooray!the main space of the mostly-marking collector has one metadata byte for each 16 bytes of payloadprecise field-logging write barrier However there is a wrinkle: when Whippet decides the it should evacuate an object, it tracks the evacuation state in the object itself; the embedder has to provide an implementation of a , allowing the collector to detect whether an object is forwarded or not, to claim an object for forwarding, to commit a forwarding pointer, and so on. We can’t do that for raw data, because all bit states belong to the object, not the collector or the embedder. So, we have to set the “pinned” bit on the object, indicating that these objects can’t move.little state machine We could in theory manage the forwarding state in the metadata byte, but we don’t have the bits to do that currently; maybe some day. For now, untagged pointerless allocations are pinned. You might also want to support untagged allocations that contain pointers to other GC-managed objects. In this case you would want these untagged allocations to be scanned conservatively. We can do this, but if we do, it will pin all objects. Thing is, conservative stack roots is a kind of a sweet spot in language run-time design. You get to avoid constraining your compiler, you avoid a class of bugs related to rooting, but you can still support compaction of the heap. How is this, you ask? Well, consider that you can move any object for which we can precisely enumerate the incoming references. This is trivially the case for precise roots and precise tracing. For conservative roots, we don’t know whether a given edge is really an object reference or not, so we have to conservatively avoid moving those objects. But once you are done tracing conservative edges, any live object that hasn’t yet been traced is fair game for evacuation, because none of its predecessors have yet been visited. But once you add conservatively-traced objects back into the mix, you don’t know when you are done tracing conservative edges; you could always discover another conservatively-traced object later in the trace, so you have to pin everything. The good news, though, is that we have gained an easier migration path. I can now shove Whippet into Guile and get it running even before I have removed untagged allocations. Once I have done so, I will be able to allow for compaction / evacuation; things only get better from here. Also as a side benefit, the mostly-marking collector’s heap-conservative configurations are now faster, because we have metadata attached to objects which allows tracing to skip known-pointerless objects. This regains an optimization that BDW has long had via its , used in Guile since time out of mind.GC_malloc_atomic With support for untagged allocations, I think I am finally ready to start getting Whippet into Guile itself. Happy hacking, and see you on the other side! inside and outside on intent on data on slop fin

yesterday 2 votes
Creating static map images with OpenStreetMap, Web Mercator, and Pillow

I’ve been working on a project where I need to plot points on a map. I don’t need an interactive or dynamic visualisation – just a static map with coloured dots for each coordinate. I’ve created maps on the web using Leaflet.js, which load map data from OpenStreetMap (OSM) and support zooming and panning – but for this project, I want a standalone image rather than something I embed in a web page. I want to put in coordinates, and get a PNG image back. This feels like it should be straightforward. There are lots of Python libraries for data visualisation, but it’s not an area I’ve ever explored in detail. I don’t know how to use these libraries, and despite trying I couldn’t work out how to accomplish this seemingly simple task. I made several attempts with libraries like matplotlib and plotly, but I felt like I was fighting the tools. Rather than persist, I wrote my own solution with “lower level” tools. The key was a page on the OpenStreetMap wiki explaining how to convert lat/lon coordinates into the pixel system used by OSM tiles. In particular, it allowed me to break the process into two steps: Get a “base map” image that covers the entire world Convert lat/lon coordinates into xy coordinates that can be overlaid on this image Let’s go through those steps. Get a “base map” image that covers the entire world Let’s talk about how OpenStreetMap works, and in particular their image tiles. If you start at the most zoomed-out level, OSM represents the entire world with a single 256×256 pixel square. This is the Web Mercator projection, and you don’t get much detail – just a rough outline of the world. We can zoom in, and this tile splits into four new tiles of the same size. There are twice as many pixels along each edge, and each tile has more detail. Notice that country boundaries are visible now, but we can’t see any names yet. We can zoom in even further, and each of these tiles split again. There still aren’t any text labels, but the map is getting more detailed and we can see small features that weren’t visible before. You get the idea – we could keep zooming, and we’d get more and more tiles, each with more detail. This tile system means you can get detailed information for a specific area, without loading the entire world. For example, if I’m looking at street information in Britain, I only need the detailed tiles for that part of the world. I don’t need the detailed tiles for Bolivia at the same time. OpenStreetMap will only give you 256×256 pixels at a time, but we can download every tile and stitch them together, one-by-one. Here’s a Python script that enumerates all the tiles at a particular zoom level, downloads them, and uses the Pillow library to combine them into a single large image: #!/usr/bin/env python3 """ Download all the map tiles for a particular zoom level from OpenStreetMap, and stitch them into a single image. """ import io import itertools import httpx from PIL import Image zoom_level = 2 width = 256 * 2**zoom_level height = 256 * (2**zoom_level) im = Image.new("RGB", (width, height)) for x, y in itertools.product(range(2**zoom_level), range(2**zoom_level)): resp = httpx.get(f"https://tile.openstreetmap.org/{zoom_level}/{x}/{y}.png", timeout=50) resp.raise_for_status() im_buffer = Image.open(io.BytesIO(resp.content)) im.paste(im_buffer, (x * 256, y * 256)) out_path = f"map_{zoom_level}.png" im.save(out_path) print(out_path) The higher the zoom level, the more tiles you need to download, and the larger the final image will be. I ran this script up to zoom level 6, and this is the data involved: Zoom level Number of tiles Pixels File size 0 1 256×256 17.1 kB 1 4 512×512 56.3 kB 2 16 1024×1024 155.2 kB 3 64 2048×2048 506.4 kB 4 256 4096×4096 2.7 MB 5 1,024 8192×8192 13.9 MB 6 4,096 16384×16384 46.1 MB I can just about open that zoom level 6 image on my computer, but it’s struggling. I didn’t try opening zoom level 7 – that includes 16,384 tiles, and I’d probably run out of memory. For most static images, zoom level 3 or 4 should be sufficient – I ended up a base map from zoom level 4 for my project. It takes a minute or so to download all the tiles from OpenStreetMap, but you only need to request it once, and then you have a static image you can use again and again. This is a particularly good approach if you want to draw a lot of maps. OpenStreetMap is provided for free, and we want to be a respectful user of the service. Downloading all the map tiles once is more efficient than making repeated requests for the same data. Overlay lat/lon coordinates on this base map Now we have an image with a map of the whole world, we need to overlay our lat/lon coordinates as points on this map. I found instructions on the OpenStreetMap wiki which explain how to convert GPS coordinates into a position on the unit square, which we can in turn add to our map. They outline a straightforward algorithm, which I implemented in Python: import math def convert_gps_coordinates_to_unit_xy( *, latitude: float, longitude: float ) -> tuple[float, float]: """ Convert GPS coordinates to positions on the unit square, which can be plotted on a Web Mercator projection of the world. This expects the coordinates to be specified in **degrees**. The result will be (x, y) coordinates: - x will fall in the range (0, 1). x=0 is the left (180° west) edge of the map. x=1 is the right (180° east) edge of the map. x=0.5 is the middle, the prime meridian. - y will fall in the range (0, 1). y=0 is the top (north) edge of the map, at 85.0511 °N. y=1 is the bottom (south) edge of the map, at 85.0511 °S. y=0.5 is the middle, the equator. """ # This is based on instructions from the OpenStreetMap Wiki: # https://wiki.openstreetmap.org/wiki/Slippy_map_tilenames#Example:_Convert_a_GPS_coordinate_to_a_pixel_position_in_a_Web_Mercator_tile # (Retrieved 16 January 2025) # Convert the coordinate to the Web Mercator projection # (https://epsg.io/3857) # # x = longitude # y = arsinh(tan(latitude)) # x_webm = longitude y_webm = math.asinh(math.tan(math.radians(latitude))) # Transform the projected point onto the unit square # # x = 0.5 + x / 360 # y = 0.5 - y / 2π # x_unit = 0.5 + x_webm / 360 y_unit = 0.5 - y_webm / (2 * math.pi) return x_unit, y_unit Their documentation includes a worked example using the coordinates of the Hachiko Statue. We can run our code, and check we get the same results: >>> convert_gps_coordinates_to_unit_xy(latitude=35.6590699, longitude=139.7006793) (0.8880574425, 0.39385379958274735) Most users of OpenStreetMap tiles will use these unit positions to select the tiles they need, and then dowload those images – but we can also position these points directly on the global map. I wrote some more Pillow code that converts GPS coordinates to these unit positions, scales those unit positions to the size of the entire map, then draws a coloured circle at each point on the map. Here’s the code: from PIL import Image, ImageDraw gps_coordinates = [ # Hachiko Memorial Statue in Tokyo {"latitude": 35.6590699, "longitude": 139.7006793}, # Greyfriars Bobby in Edinburgh {"latitude": 55.9469224, "longitude": -3.1913043}, # Fido Statue in Tuscany {"latitude": 43.955101, "longitude": 11.388186}, ] im = Image.open("base_map.png") draw = ImageDraw.Draw(im) for coord in gps_coordinates: x, y = convert_gps_coordinates_to_unit_xy(**coord) radius = 32 draw.ellipse( [ x * im.width - radius, y * im.height - radius, x * im.width + radius, y * im.height + radius, ], fill="red", ) im.save("map_with_dots.png") and here’s the map it produces: The nice thing about writing this code in Pillow is that it’s a library I already know how to use, and so I can customise it if I need to. I can change the shape and colour of the points, or crop to specific regions, or add text to the image. I’m sure more sophisticated data visualisation libraries can do all this, and more – but I wouldn’t know how. The downside is that if I need more advanced features, I’ll have to write them myself. I’m okay with that – trading sophistication for simplicity. I didn’t need to learn a complex visualization library – I was able to write code I can read and understand. In a world full of AI-generating code, writing something I know I understand feels more important than ever. [If the formatting of this post looks odd in your feed reader, visit the original article]

yesterday 4 votes
Introducing the blogroll

This website has a new section: blogroll.opml! A blogroll is a list of blogs - a lightweight way of people recommending other people’s writing on the indieweb. What it includes The blogs that I included are just sampled from my many RSS subscriptions that I keep in my Feedbin reader. I’m subscribed to about 200 RSS feeds, the majority of which are dead or only publish once a year. I like that about blogs, that there’s no expectation of getting a post out every single day, like there is in more algorithmically-driven media. If someone who I interacted with on the internet years ago decides to restart their writing, that’s great! There’s no reason to prune all the quiet feeds. The picks are oriented toward what I’m into: niches, blogs that have a loose topic but don’t try to be general-interest, people with distinctive writing. If you import all of the feeds into your RSS reader, you’ll probably end up unsubscribing from some of them because some of the experimental electric guitar design or bonsai news is not what you’re into. Seems fine, or you’ll discover a new interest! How it works Ruben Schade figured out a brilliant way to show blogrolls and I copied him. Check out his post on styling OPML and RSS with XSLT to XHTML for how it works. My only additions to that scheme were making the blogroll page blend into the rest of the website by using an include tag with Jekyll to add the basic site skeleton, and adding a link with the download attribute to provide a simple way to download the OPML file. Oddly, if you try to save the OPML page using Save as… in Firefox, Firefox will save the transformed output via the XSLT, rather than the raw source code. XSLT is such an odd and rare part of the web ecosystem, I had to use it.

yesterday 2 votes