2024 in Review

Last year was tough, but all manner of things ratcheted the overall difficulty to eleven, and here we are.

I had to start with this because it’s been a major theme of the past two years, and it’s not going away–plus I have been working extensively in this area in both my day job and personal projects.

And yet, other than my news summarizer, I still haven’t found a captivating use for generative AI–and by that I mean large language models (image and video generation are a completely different kettle of fish). I use LLMs daily, spend far too much time fiddling with ollama and other APIs (i.e., almost literally all the models, besides all three major hyper-scaler offerings), and have a homegrown AI stack that allows me to build anything I might want (thank goodness for OpenAI-compatible endpoints).

However, I still haven’t found a project that I can point to and say, “this is what AI is really good for.”

In fact, I am seeing more and more ethically dubious uses of AI, many of which are being pushed by companies that pay lip service to AI ethics while shifting accountability (and any human impact) to customers–much like arms dealers do.

This is disheartening, and I am now more cautious about how I engage with AI. And, of course, my main frustration is that I really wish people had a better grasp of the actual limitations of AI.

As I wrote before, this has easily been one of the worst years of my career, even though I am now back to a role that emphasizes a semblance of order amidst the general chaos of the technology industry.

But let’s face it: I am still recovering from major burnout.

The main outcome is that I’ve changed my attitude towards work (and, by extension, technology). I still want to derive satisfaction and pride from my work. However, I am now much more guarded against overcommitting or allowing work to encroach upon my personal life.

And, of course, I’m frustrated that work has negatively impacted me for two consecutive years, with things deteriorating to the point where I derive more satisfaction from a quiet, orderly week than from a challenging one.

I decided against travel for work unless absolutely necessary and successfully stuck to that decision for now, even if it means turning down some interesting opportunities.

My home office setup has remained mostly unchanged except for acquiring a new chair: an IKEA Markus, which is fine and comfortable (the adjustable lumbar support, while grudgingly so, is fine for me), but I suspect I may remove the armrests in the future.

I did consider getting a standing desk again, but it’s hard to forgo the convenience of my current 180x90cm desk, and replacing it would be a hassle–plus I am still happy with the setup in general, even if my Arturia KeyLab 61 (which sits right in front of me every day) has become an accusatory dust magnet given how little inspiration I have to use it.

So far, I have resisted the allure of the new Mac Mini M4 due to budget constraints and the fact that the M2 Pro I got last year is still more than adequate for my needs. Additionally, in the extended ranking of things I would like to upgrade, the iPad Mini was higher on the list, and this year’s model was a significant disappointment.

My homelab and its little menagerie of various machines is currently on a consolidation trajectory after a brief period of expansion that took it up to seven nodes, and will eventually converge onto more powerful host hardware that can run the AI, media, and development stacks I’ve been fine-tuning over the year.

That consolidation work (which sometimes involves rebuilding a few VMs or containers) is a slower, more thoughtful process than sprawling throwaway dev environments across new machines, and I’m now at a point where I want to set up self-hosted CI/CD for a few private projects (another project that was cut short by professional turmoil).

I haven’t been writing as well or as much as I would have liked. Since I have little (or nothing) positive to say about the industry, AI or the current status quo (and the impending turmoil in the US) writing reviews has, in a way, been the thing that kept me going, even as I become much more selective about what I take on.

They also require time and effort to do “right”, or at least to a minimal standard that I find acceptable, and to be honest, it has been a long time since I’ve had the stillness of mind to sit in my home office listening to music and write about more generic technology-related topics like I used to do.

I hate myself just a little bit for watching so much YouTube (which has many rabbit holes to fall into when you’re too tired to do anything but live vicariously through other people’s wacky projects), but I am very happy to have watched Arcane, Shōgun and Slow Horses this year (as well as steadily plowing through Frasier with my kids nearly every evening¹).

I’ve managed to keep up with my reading, although this year I had to resort more to re-reading comforting (i.e., marginally less dystopian than reality) books than exploring new fare.

Gaming has taken an interesting turn this year. I’ve not only maintained the streaming setup I pioneered with the Logitech G Cloud, but most of the gaming that happens in our house is now streamed from either the Xbox or the AM18 I reviewed earlier in the year.

After a few months of use as a Proxmox/backup AI server, I set the AM18 aside as a dedicated Bazzite box that streams to my Apple TV.

And following Nintendo’s attempt to eradicate Switch emulation, I decided not only to snapshot a working environment but also to transform the AM18 into a Switch and PS3 emulation box. Setting up emulated versions of many games that were gathering dust in the cupboard was a fun endeavor and provided a great excuse to revisit titles I haven’t played in years.

Things were going well here until my work reorg, and I had actually purchased a soldering microscope and a hotplate to begin doing more advanced SMD work, with a couple of projects already in mind (and one effectively in hand). I acquired a Spacemouse, designed parts and enclosures, and ordered components for prototypes–then I lost the energy and enthusiasm to do anything of consequence.

I have been tweaking the SK1 a bit (and I am still satisfied with it, even if it has a tendency to heat creep and get filament stuck in the extruder with the enclosure closed), but the 3D printing I have done over the past six months has been mostly utilitarian.

I actively dabbled a bit earlier in the year, but it’s mostly still not happening, even though I am trying to get around to it in a rather roundabout way by doing more electronics and building more synthesizers.

After my recent role change I started walking a few kilometers a day, which has indeed made a difference. Not having daily evening calls with the US has also markedly improved my sleep quality (I can still manage with less than six hours of sleep, but whatever I do in the hours leading up to sleep has a huge effect on how I sleep). Even though I am currently fighting a losing battle with holiday “diet”, I have managed to be in a slightly better place than I was a year ago.

My current challenge is that I miss the years when I was moving about all over town during the average day. I am trying to figure out how to incorporate more exercise into my day, but I still juggle too many calls and meetings–across a smaller set of time zones, but still enough spread to make it difficult to find a good time to exercise.

But I’m going to focus on keeping this up, somehow.