Becoming a Resident Toolbuilder

I want to build tools for important problems. Some of the most important problems are being tackled in academic biology labs. They seem like a suitable serious context of use to invent powerful new representations and tools and help improve public health.

https://twitter.com/sashachapin/status/1453423907898232836?s=21

I'm considering reaching out to bio labs across the country and volunteering (for a small stipend) to be a "resident toolbuilder". Not just to solve immediate digital problems for the lab but to examine and improve their processes and accessibility for everyone. Perhaps down the road I can generalize one lab's tooling into a more universal platform (Figma for Science).

Two caveats to look out for:

The "faster horse" problem: will labs be biased towards conventional tools and just want me to do plumbing? Assuming plumbing is actually the biggest bottleneck, maybe I can do something to improve productivity there. Do they even know what they need in general? It would be best if I could observe (a la Laboratory Life) what they're doing and act from there, using my own imagination.
Accruing technical debt: I don't want to leave the lab in more of a mess than it started, with rapidly-rotting bespoke frontends that will end up in the trash quickly. Ideally I build an oeuvre that is either very easy for the lab to maintain, or is so cohesive and compelling that I can maintain it myself as a "platform".

In general I'm becoming more biased in favor of building for "hard tech" for urgent problems in the world (I resonated with Kevin Lynagh's post on important work). Effective altruism is another realm I'm adjacent to and very much a supporter of.

I incidentally discovered Bret Victor considered going this route early in his career:

I saw climate change as this crisis and one of the most important things to work on. And I’d — I’d tried to work directly on that in certain ways and realized that I just didn’t really have the kind of temperment for going really deep on a scientific problem. I was more of a toolmaker. I wanted to make tools that enabled other people to rally other people in the problem. Even before Apple I had come up with this plan of, “Well, I wanna make this really powerful scientific tool. I don’t really know enough of how science is practiced. And so I’m going to like travel around the country, spending a month at a different scie — a different lab um every month. Just kind of offering myself as like, “I’m a free toolmaker, use me how you will.” And then thereby kind of get — get the experience of um after a year of like, “Oh here’s the sorts of needs that scientists have and here is the super tool that I can make.” That plan didn’t really get off the ground because um it turns out I’m not really social enough to reach out to people actively like that.

I went through this period of traveling around and kind of coming up with these themes. But in the process of doing that, I uh — a team of scientists at um at Georgia Tech and um did some prototyping with them. And the structures that I came up with and the enthusiasm that they had for them made me think, kinda go back to that, “Hey, I really wanna make a tool for scientists. I wanna make a new MatLab or a cross between MatLab and Garageband.” And so that was um I’ve never admitted this publicly before but that was kind of my — my major project for like the next year or two after that was I was making this um this new scientific tool.

It’s hard to have the level of motivation to pull off something really huge like that, if you don’t have the right support structures in place.

https://jsomers.net/i-should-have-loved-biology/

https://newscience.org/how-software-in-the-life-sciences-actually-works-and-doesnt-work/

Fun random example: Chaim Gingold & Shawn Douglas's Gelbox

Related: Figma for Science

Kevin suggests commercial labs "get more done" - just a matter of being good about IP transfer.

https://twitter.com/patio11/status/1069422308526239744

https://twitter.com/ArtirKel/status/1505284608971083779

Consultancy with lab clients https://latticeautomation.com/

Backlinks