Weeknotes 2024 W04: End of work

January 22​–​28, 2024
1900 words

I have left my job. Friday was my last day, and now I’m planning to take a break.

Initially, I wanted to not start looking for new opportunities just yet — give it a few weeks or so  — but I couldn’t help myself and I’ve already scheduled calls to discuss future opportunities.

Would you like me on your team? Do get in touch! You can grab an up-to-date version of my CV, too.

My last day at my job also coincides with the end of the performance review cycle, and I got the feedback that I, well…

… failed to meet expectations.

Okay, so, I know that saying “please hire me” and following that up with “I fail to meet expectations” does not inspire confidence. [Good lord what are you doing? —  Ed.]

Performance reviews are odd beasts. The core of my performance review is that I failed to deliver anything in the last six months, which is, well, mostly correct. Performance reviews rate people based on their output, not the effort that went into it, and they disregard the context in which the work happens.

I worked on two projects. One of those hasn’t had product requirements since late 2022, making it tough for me to meet expectations.1 For the other project, unbeknownst to me, someone else was working on the exact same thing, and in the end, their changes went in and my six weeks of work got scrapped.2

So yes, I failed to deliver anything in the last six months — but I was not set up for success, and then received the blame. The cause of my burnout is clear.

On the positive side, I received lots of praise for mentoring engineers and empowering them. I’ll accept that!

During my employment, I learnt a good amount, though less in the sense of “personal development” and more in the sense of knowing what doesn’t work for me personally. One of those things is that I do much better work when I’m working towards a vision; I prefer being given long-term problems rather than short-term projects.

My employer uses a 3–4 month cycle for planning. Each cycle begins with a few weeks of planning meetings, in which projects are defined for the coming cycle. That is then followed by the execution of those projects.

Where I believe this approach goes wrong is that projects are the only means in which work is planned. People are assigned to projects, but in between the cycles, when projects are finished and no new projects have been defined yet, there is a lull. I have personally experienced this, spending weeks not having anything to do. I’d have to hunt for things that I could potentially take on — but aren’t too big because the next cycle plan could drop any minute.

Having only projects as a means of planning is not enough. Products need to have long-term vision, a “north star” or so, that can be continually worked towards, even when there are no active projects.

Having this vision is helpful for defining projects too. Without that vision, the likelihood of projects being a radical departure to what was worked on in the last cycle is too high. Without a vision, it is unpredictable what comes next. As an example, one project was migrating data to a new system, while another project involved rewriting that very system that was being migrated to; in the cycle that came after, the rewrite was scrapped.

With a long-term vision present, you can continually ask crucial questions like these:

Especially the latter question is important. Software development is a process that continuously creates knowledge, and that knowledge is invaluable in reevaluating and realigning the vision and the projects that fulfill this vision. When done continuously, realignment is only slight and can be done quickly; surprises are few and far between.3

All this connects closely what I wrote about in Week­notes 2023 W37: Preferring problems: I prefer being tasked with solving problems, rather than being given a solution to implement. Tasks and projects are similar, though different in scope. It is only the vision that creates the set of problems to be worked on, and it is only the vision that can explain why what is being worked on is important.

My manager said he had considerable trouble finding projects for me to work on. There weren’t any projects that were particularly interesting to me, but that is because I was looking for something else entirely: long-term problems to solve, things to do that would move everything and everyone in the direction of the vision.

My disinclination to work on projects has been erroneously perceived as me not wanting to work at all. I do my best work and I am the most motivated when I get to work on strategic initiatives (vision), spending time understanding the problems and coming up with my own tactics (projects and tasks).

My dislike for generative AI systems is related to what I wrote above: such systems can only be given tasks to fulfill, and will never come back with questions for more input. It will happily attempt to complete tasks even when the task is ill-specified or the context is missing.

It is no wonder that the idea of replacing human worker drones with AI drones is appealing to some: no critical thinking; just do your job.

Once, a team asked for my help because they were understaffed and wouldn’t be able to deliver on time. I helped them out, but to my surprise, my changes were rejected, because the solution I provided did not look like something the team could have come up with. The way I tackled the work was to look for the underlying problem — the “why” — and solve that, but what they expected from me was that I do exactly what I was told, even ignoring my prior professional experience.

They wanted a worker drone. I can’t be a worker drone.

This made me wonder: do people that use generative AI systems even want such systems to be intelligent, or do they want systems that slavishly obey them?

It is abundantly clear by now (as it should always have been) that what passes for “AI” is not intelligence. Until such systems have critical thinking, can ask questions, can ask for context, and can point out that the lack of input and the presence of contradictions make it impossible to fulfill given tasks, those systems are not remotely close to being intelligent.

I can ask a generative AI system to write a short story, and it will do so without even asking me any questions about the genre, the theme, the underlying message, the characters, the setting, and so on. The end result cannot be anything but a hollow, meaningless string of words.

I can ask a generative AI system to write a one-line limerick, and it will happily try and fail over and over again, because it has no critical thinking, no understanding. Here, too, the end result is a hollow, meaningless string of words — and that is true not just for attempts to make one-line limericks, but for every single piece of generative AI output.

Life gets its meaning through the interaction with other people — people you can learn from, people who have different viewpoints, people who inspire you, people who are different from you, people who can challenge you, people who can make decisions and argue in favor of them, people who are different in ways that you could never have imagined.

Generative AI just sort of stands in the way there.

I continue to suffer from gremlins. At work, I got an invitation for a meeting that was three weeks ago. There is something fundamentally wrong with my Microsoft Outlook setup.

At least I am now free from Microsoft Outlook.4 Whew!

Entertainment:

Links:

Tech links:

  1. It felt like being led blindfolded into a room and being told to hit the dartboard with the darts you’ve been given. Is there even a dartboard in the room? ↩︎

  2. I am still at a loss to explain how this happened. The project was explicitly assigned to me but then reassigned without my knowledge. I actively communicate progress on all the work I do, and for this project my approach was no different. ↩︎

  3. I wrote about the wonders of continuous alignment last week, in Week­notes 2024 W03: Open channels. ↩︎

  4. “I cannot wait to stop using Microsoft Outlook” — me, last week, in Week­notes 2024 W02: No no no notifications. ↩︎

  5. Once Upon a Time in America, directed by Sergio Leone, written by Harry Grey, Leonardo Benvenuti and Piero De Bernardi (The Ladd Company, Warner Bros., Producers Sales Organization (PSO), 1984). ↩︎

  6. While it wasn’t storytelling specifically, I wrote about my struggles with Fallout 4 in Week­notes 2024 W02: No no no notifications. ↩︎

You can reply to this weeknotes entry by email. I’d love to hear your thoughts!
If you like what I write, stick your email address below and subscribe. I send out my weeknotes every Sunday morning. Alternatively, you can subscribe to the web feed.
Incoming links: Gremlins.