Bethany Crystal
Over 600 subscribers
Last night I crashed my own app's server by attempting a database migration for a new feature I'm pushing out. This morning, I rebooted the server, and we're back online. Such is life.
This is the kind of thing that would have really stressed me out a few years ago. But in the AI age, I've grown a higher tolerance of room for error, and I've taken on a stronger bias toward action over perfection.
Why? Because I've realized that I learn more quickly by trying stuff (and breaking stuff) than just sitting around reading tutorials on how to not crash your server with a database migration. And since I'm not exactly serving around-the-clock users of people scanning stuff with MuseKat (yet...), I realized I could break the app overnight with limited long-term impact.
This relative risk assessment helped me decide what to do next, which is a mindset I’ve been adopting across a lot of my work lately.
When it comes to learning something new (especially with AI), I’ve found that what we really need are safe sandboxes to play, break, fix, and repeat. That might mean giving yourself (or your team) a longer leash than usual, and getting comfortable with the occasional crash. For today's post, I thought I'd share a bit about how I’ve been building up my tolerance for risk over the past few years and why I think that matters more than ever in the AI era.
Back in 2020, I decided to leave a job in VC to go back into the operating side of building at tech startups. Since then, I've worked in about 12 different places, largely in a fractional capacity. Needless to say, in this near-constant state of changing jobs, I spent a lot of free cycles assessing the risk profile of one career decision over another.
Here are some of the relative risk considerations that informed some of my job changes:
Some examples of considerations I've made while assessing relative risk of job shifts over the past few years:
If I cut away any equity-earning options by no longer working full-time, what's the risk to my long-term investment strategy?
If I switch industries entirely, what's the risk that I miss out on building upon my expertise for better jobs, had I stayed?
If I take two jobs at once, what's the risk that I overload myself?
If I don't take a job right away, what's the risk of running out of money before I find the next one?
If I take a job I've never done before (and fail), what's the industry risk to my reputation?
I was particularly worried about that last one last year when I took a job as the interim General Manager of a dev shop in the distributed systems engineering space. The org was in a tricky situation, and I didn't consider myself a subject matter expert in the domain area, nor had I been in a sole operating role in that way before. But I really wanted to see what would happen if I inserted myself into a highly uncomfortable situation outside of my comfort zone.
After looking closely at the big picture, I realized a few things were true:
This was one job, out of a long series of fractional jobs. One failure wouldn't implicate a broader track record trend.
The team was already in a tricky situation; anything I did would (hopefully) be better than nothing.
Since I work across so many different industries, even a total burnout in one industry wouldn't necessarily impact my work in another sector.
Given all that, I decided to take the job.
At the onset, I was still pretty worried that I wouldn't be able to get the team out the tough spot they were in, then I'd have to deal with a lot of downside reputational risk from being "the one who couldn't help." But after attending a few conferences on behalf of that org, I found the exact opposite to be true. People weren't dissuaded by the mess; they were inspired by the resilience.
Even while this didn't end up being the right long-term fit for me (to no one's surprise), I ended up leaving on better terms that I'd imagined, no hard feelings. In fact, the person who tapped in after me ended up being an even better fit for a lot of reasons. And I've been surprised to see how even that brief stint earned me a little more respect from my industry peers, many of whom would never have touched a situation like that with a ten-foot-pole.
In short, that job gave me a safe space to fail.
What I've learned is–getting comfortable with risk (and the risk of failure) is not a switch you can turn on one day; it's a muscle that you must actively build over time. For someone like me with a lot of intense perfectionistic tendencies, it's something I've had to work toward slowly, over many years.
Even if you're not ready to throw away a cushy job in exchange for a circuitous adventure, you can still practice building your risk tolerance wherever you are today. Here are some examples:
Working at a full-time job where you've had the same role for a long time? Raise your hand to volunteer to lead a new project or initiative that's slightly outside of your comfort zone. The act of learning something new and teaching it back to your colleagues in real time will teach you how you react in unknown circumstances.
Spending your free nights or weekends on Netflix or on social media? Pick up a new hobby instead, but cap the amount of time you dedicate toward practicing on your own. Pick a deadline to "push publish" or "show your work." If it's an athletic pursuit, sign up for a competition or match, if you're learning how to bake, try selling some goods at a local fare, or if you're writing, try submitting a draft of your work to a few publications.
Looking for a new job right now? Rather than seek out roles that match your prior experience, try reframing your search around a pursuit of future knowledge, rather than a demonstration of past expertise. Ask yourself, "What do I want to learn next? Where could I go to best acquire that skill?" and then go apply there.
Think you're already pretty good at some special skill? Prove it. Write a blog post, a research paper, or a deep analysis on what you've learned. Or start a YouTube or TikTok channel with tutorials and see if other people agree. It's scary to put yourself out there; you'll start to see how people respond to what you have to say.
Notice yourself falling into the same habits with friends or travel? Pick an entirely new place you've never been before. Show up with less preparation than usual, see what happens when you try to figure it out as you go. When things completely go off the rails (which they will), notice how you move through the tough moments.
Managing a team of individual contributors? Ask them to build something new with AI this week. Dedicate a half-day training toward "free play" time, without boundaries or constraints. Encourage them to show their work, talk about where they got stuck, and how they might keep building on top of it to improve upon it next week.
You'll notice that one thing a lot of these things have in common is the idea of novelty. Doing something new that you haven't done before is often the easiest way to get yourself a little out of your comfort zone and practice learning something a little new or a little riskier than before.
These certainly aren't the only ways that you can build up a muscle for a bit more risk. But it's a good start. Given the rate of change in the way we learn and the way we work today, we're all going to need to get a little more comfortable feeling uncomfortable.
And that's a good thing. Because scary means you're learning.
Last night I crashed my own app's server by attempting a database migration. This morning, I rebooted, and we're back online. Such is life. As it turns out, I learn faster by trying stuff (and breaking stuff) than just sitting around reading tutorials. For today's post, more on how I’ve been building up my tolerance for risk and why I think that matters more than ever in the AI era. https://hardmodefirst.xyz/why-we-need-safe-spaces-to-fail
Last night ended with an app server crash during a database migration, but a reboot restored service this morning. Embracing trial and error has become vital in the age of AI. Find out how this mindset aids in learning from risks and mistakes in today's post by @bethanymarz.