Links, August 30, 2024

Erika Hall Knows How to Fix Your Design Process (But You’re Probably Not Going to Like It)

I’ve looked at other industries that are more mature in terms of baking hazards or ethical considerations into the practice for a model. Architecture is a really great for that. So is medicine. Because architects and structural engineers and doctors understand that if they don’t pay attention to hazards, people will die.

In the early days of the web, you couldn’t kill somebody with a web page. But now, these systems are so complicated that there are real, life-or-death, consequences for what we build.

I found myself nodding my head a lot in this piece as well as disagreeing with chunks of it, but the themes of slowing down and getting back to treating design as a process collaborative inquiry resonate with my own experiences a lot.

The Secret Inside One Million Checkboxes

I was lucky to catch a presentation of this at xoxo, it’s so good, and like a lot of things at that event, reminded me why I got into technology work to begin with.

I panicked. There were URLs in my database! There were URLs pointing to catgirls.win in my database!! Something was very very wrong.

I assumed I’d been hacked. I poured over my logs, looking for evidence of an intrusion. I read and re-read my code, searching for how somebody could be stuffing strings into a database that should have just contained 0s and 1s.

Does this scale down? 📉

Why are people not more worried about complexity, the learning curve, and the cost of maintenance? Technology built for a large corporation may not be suitable for a small.

This nicely summarizes something I think has been one of the bigger problems in the culture of the technology industry over the previous two decades.

workerd, Cloudflare’s JavaScript/Wasm Runtime

Cloudflare as a company is problematic, yes. I’ve done a fair bit of work for clients with their Pages/Workers platform, and this website is as of this writing hosted on Pages — it’s been on my todo list to find an alternative run by better people, but nothing hits the sweet spot without costing at least $25/month, or involving a good chunk of work I haven’t made the time for to setup.

Unlike some of its competitors, I think their workers product is good. The “serverless”/nanoservices/function-as-a-service architecture pattern gets mis-used for a lot of things, but if you think about it as an evolution on CGI-bin, there are a lot of dyanmic-ish-shaped problems on the web that it’s well-suited for, where a full server process is overkill.

Unfortunately the nanoservice platforms are, like most cloud services, architected around platform lock-in. Apache has OpenWhisk, sure, but it strikes me as a piece of infrastructure meant to have a team surrounding its operation, not a lone operator playing many other roles. It’s meant to scale up, and because of that, it doesn’t scale down.

By contrast, workerd is simply a server-oriented runtime for javascript & wasm “functions” in the nanoservices pattern. It’s conceptually much simpler, and something you can run on a VPS without having to go onto some Discord server or forum somewhere and have people yell at you for not wanting to run an entire kubernetes cluster.

Why Do Electronic Components Have Such Odd Values?

Components built to the E6 standard have a 20% relative error tolerance, and if we look at the values again we’ll see a trend. Starting with 10 again and adding 20% error we end up with 12.

If you at all work with aggregating things on a logarithmic scale, you will be well-served by familiarity with the concept of preferred numbers. They effectively provide a framework for thinking about “rounding” for logarithmic values, which I’ve found many uses for, from data storage for monitoring engines to quantization points for many types of data visualizations.