Reading Time: 8 minutes

A year and a half ago, I wrote candidly about my decision to go indie. At the time, I confessed that the post was an experiment, and I mentioned that I might never do another one like it.*

Here I am again, with another one like it.

I spent the last year and a half working for myself, with a bunch of different orgs. A few of the adventures:

1. I procrastinated sending my health insurance application because I didn’t have stamps. So I missed the special enrollment period, and I got stuck scrambling for an interim emergencies-only plan until the 2020 open enrollment period. After that, a good friend of mine bought me a sheet of stamps to stick on my fridge.
2. A doctor friend helped me sign up for health insurance on the exchange. I learned that signing up for health insurance is a little annoying and kinda expensive, but not nearly as annoying and expensive as employers want people to think it is.
3. I thought finding clients was going to be an issue, but fortunately, it wasn’t for me. instead, boundaries were the issue. I needed to learn how to be gentle but also firm…because if I fail to be firm until it’s driving me up the wall, it’s much harder for me to be gentle.
4. My first quarter in business, I expensed zero things, my CPA yelled at me. Now I have a spreadsheet for expenses and a file for photographs of my receipts.
5. I discovered that my limiting ingredient isn’t time—it’s energy. Sometimes I thought I ran out of time, when what actually happened was I ran out of energy and spent a perfectly good three hours on Twitter instead of doing the thing I thought I didn’t have time for. So, counterintuitively, I have to program my rest first—I have to make sure I leave time each day to journal, to exercise, to talk to my friends, to read, to draw, to watch TV.
6. Relatedly, I learned to put everything in my calendar—deadlines, meetings, and blocks of time to work. I even put to-dos in there. Having it all in one place, and getting timely notifications telling me what to do, helps me a lot.

In addition to the indie adventures, I also got to design a class for the CS master’s program where I teach. Throughout its design, I wrote to you, so I won’t rehash what I said.

Here’s what I didn’t say: designing that class got me through 2020. It was my distraction and my joy in the midst of a pandemic, and a race reckoning, and a watershed political moment. Tech, as an industry, has played a pretty damning role in how we got here, and teaching feels like an opportunity to prepare people to fix it. The class I built is effective enough that we’re already planning on running it in the spring of 2021, which I confess feels surreal.

I spent 35-40 hours per week in the spring designing that class, and less than that but still an appreciable amount in the summer and autumn. Had I had a full-time job outside the university, I could not have produced a class of this quality, or anywhere near it. I am grateful that I had the opportunity to focus on it the way I did. (And thank goodness Borja Sotomayor was right about something very important: “After the first few runs, it gets a lot easier.” Whew!)

Blah blah blah, Chelsea. What’s the announcement?

The announcement is that next week I start a lead role on building machine learning patterns on the Pocket team at Mozilla.

Why am I doing this:

1. Massive tech companies’ irresponsible technical practices have far-reaching impacts, and nowhere is this more true than in machine learning. I spend a lot of time researching how to avoid and fix that. Maybe this is masochistic, but I’m excited about the pressure of doing that work at a place that calls out some of those practices and impacts. If we get it right, maybe someone will notice.
2. I met the team, and it’s full of people who make me think “I cannot imagine knowing you exist and not finding a way to work with you.”

I started my career as a code slinger: I don’t chase a specific stack. For me, it’s about the goal and the team. Working at Mozilla gives me a chance to drive positive change across the tech industry. So, I’m gonna try it.

Aren’t you nervous?

Of course I’m freaking nervous. Pressure is exciting, but it also makes me nervous. Also, working for someone else has risks. Working for a company people have heard of has even more risks.

Last week, Google fired one of its ethical AI research leads, and then one of the VPs tried to frame it like she quit. The reason: they told her to take her name off a paper she co-wrote, and she said okay under some conditions (half of which were just knowing who wanted her to do it). So the VP fired her without telling her manager, framed the termination as a resignation, and then, when he realized a lot of people were upset, sent out an email that, I guess, was supposed to be placative, but had the opposite effect when several other Google researchers pointed out that the “standard process” described in Jeff’s explanation had never applied to them or their research.

This employee was a marginalized person who spoke up about an ethical issue and then set boundaries around the company’s ability to muzzle her. Look at who I am. Look at at anything I have ever written.

I have made my career, in large part, by stepping into the public eye and saying things in ways that others haven’t, or won’t, or can’t. How long could I reasonably expect to work under a VP like Jeff Dean? Six months before I’m fired? Twelve? How could I justify putting so many eggs as a full time job in such a volatile basket?

And it’s not just whether that exact thing would happen to me personally. This week, I watched Googlers take public shamings on every social media outlet as if Jeff Dean personally asked their permission before he fired Dr. Timnit Gebru. I assure you, if he didn’t ask her manager’s permission, he did not ask the permission of some site reliability engineer. There’s more to be said on the relationship between employeeship and loyalty, and on the relationship between employees and outside influences working in cahoots to drive change. This isn’t that post. But it does make me think “Gosh, suppose some Mozilla exec pulls a totally bonehead move. Does that make me evil in the eyes of readers who trusted me?”

I’m not comparing these two companies. They make different products, operate under different directives, and one of them has 213 times as many employees as the other one. But I’m anticipating a question, and here’s the answer: yes, I’ve thought about it. I’m committed to believing marginalized people’s experiences with marginalization. I’m committed to understanding the inequality that tech orgs and tech products perpetuate.

I’ve left a company over an unethical contract before. So, like, I’m scared to do it, but that wouldn’t stop me. Mozilla’s year didn’t start off great and then got worse, and I know that. The particular issues that Mozilla has are not issues that I find wholly unconscionable at this time. I have faith, but I also have skepticism. That’s always true—not just for this.

What’s changing around here?

Probably nothing. You can continue to expect live coding sessions on the NASA Landsat Data Processing Pipeline and the Scottish Gaelic Tattoo Handbook. You can expect more posts about Python (I’m teaching a python programming class this winter) and more reflections on the influence of white supremacy culture in tech.

I’m hoping to write to you more about designing useful machine learning models with commitments to ethics and data privacy. Expect me to wax poetic about the internet as a library a few times.

_{* I did finish a few of the books I mentioned in the last announcement, but not nearly as many of them as I initially thought. Typical 😉.}

If you liked this post, you might also like:

I dunno, maybe the machine learning category

This AMA…I’m going out on a limb here

Space? Do you like space? Look, doing the “you might also like” section on a blog post this personal just feels weird, so I’m really glad it’s behind me now