flambojones

I worked in mobile robotics for a defense contractor in the early-mid 2000s and we had a homegrown Lidar (though we always called it ladar) that was large, heavy, and cost $250k to make. I remember a year out of college driving 2 hours to a military base with it in the backseat of my car and being paranoid at every bump I hit. These things seem like a dream.

Having been at Microsoft when we had SDETs for everything (and I miss it greatly, though the way we could write a feature and then just toss it to test was ridiculous), I think things have swung too far away.

On one hand, engineers needed to take more ownership of writing things other than low-level unit tests.

On the other, the SDETs added immense value a ton of ways, like writing thorough test specs based off of the feature spec (rather than the design spec), testing without blind spots due to knowledge of the implementation, implementation of proper test libraries and frameworks to make tests better and easier to write, and an adversarial approach to trying to break things that makes things more robust.

I've also worked with manual QA for product facing flows, and while they added value with some of their approaches to ensuring quality - poking at our scenarios and tests, and looking more closely at subjective things - they often seemed to work as a crutch for the parts of code paths that engineers had made too difficult to test.

I've never seen anywhere attempt to replace the value that SDETs delivered with what engineers were tasked with. I'd argue it's not necessarily possible to fully replicate that when you're testing your own things. But with services now, it also seems like product/management are more willing to have slightly few assurances around quality and just counting on catching some in production in favor of velocity.

I've never seen places that got rid of QA

Sounds like an opportunity for the person who sent it to learn to improve their prompts. I've set my default claude rules to emphasize doing small-contained commits and use a stacking tool like Graphite to help them get reviewed.

Ahh good call. My $200k vs. your $200k doesn't really matter with a given year.

Salaries aren't a one-time expense, so is the amortization rolling? Like, year 1, you pay me $200k and deduct $40k. In year 2, you pay me another $200k, do you get to deduct $40k for year 1's salary and $40k for year 2's salary?

I guess another way to ask is, does this mean that if you keep someone for 5 years and don't change my wages, is their yearly salary effectively fully deductible? If so, does that create incentives to try to keep employees longer-term in order to make them more cost-efficient?

+1 - a universal way of doing this via markdown beats learning whatever app-specific hotkeys I have to do to make an H3 or whatever. I don't have any particular feelings about markdown as a format for data, but as a universal set of shortcuts, I've found it to be a huge productivity boost.

I’ve completely been there. I burnt out hard, not from overwork, but from this type of stuff. I started to think maybe I enjoyed software engineering but not the business of software engineering.

I took a break and did a lot of reflecting. I bought a workbook in career changes and started working through it. After a bit I started getting the itch to code again, so I did a Udemy course in a programming area I had done previously and really enjoyed but had gotten away from.

After a while I realized I was ready to go back to work. Before I did, I tried to figure out what made me so miserable previously. In my case it was the poor decision-making processes with unclear ownership that made me feel like I was always seeking permission to get things done. I also decided I didn’t want to work on large projects with untyped Python. When I started interviewing, I decided not to be picky or make too many assumptions in the early stages of the process. Once I started interviewing, I really honed in on these things and asked a lot of questions about it. I ended up at a small startup that’s doing reasonably well with really high velocity. I feel like I have a conversation about some design thing in the morning and I’m well on my way to implementing it in the afternoon.

I went through this for a few years up to about two years ago, though the cause of my burnout was less overwork and more politics and things like that. I don’t feel comfortable giving advice, but here’s what I did:

1. Therapy - it turns out a lot of my burnout was due the ways that I had grown to handle hard situations breaking down. I needed to process how I dealt with success and failure and hard situations. Ultimately it’s similar to the “care less” advice but I couldn’t have gotten there without therapy.

2. Time off - this was a tremendous luxury, but I took about 6 months off. For the first month I did nothing but help my kids do remote Covid school. After that I started trying to work through where I was at. I started meditating, reading books on self-compassion, burnout, and career changes. I started exploring potential career changes. After a while, I started getting the itch to program and worked on some personal projects. This was really useful for processing that I still enjoyed programming.

3. Figure out why the previous job didn’t work. I spent a fair bit of time thinking about why the previous job was hard for me. Turned out it was the slow pace of decision making and diffuse accountability that made driving projects really stressful for me.

4. Find a job that checks the boxes… once I realized I still enjoyed writing code and had figured out what I didn’t want, I was able to optimize my job search and ask the questions I needed to during the process to find something we’ll-suited for me. It’s possible to be wrong here, either about the causes or about what the new job will offer, so I gave myself some slack here; nothing is perfect. 5. Medication - even after I started my new job, I still dealt with some burnout-y feelings, after conversation with my therapist, I got a prescription for a relatively low dose. This wasn’t magic, but it has helped me reset my baseline feelings and approach things with more ambivalence.