Articles in Notes

This weekend I finally got done something I probably should have back since I got my first computer nearly two decades ago: a script to automate the initial setup of a new OS installation. After all, I've been using Linux for most of that time and it wasn't even that hard to do. I guess I never felt fully confident in my setup itself to bother replicating it.

The biggest hurdle was that I have also always wanted synchronization of my most valuable files across all my devices... Which I finally got done this time too in a simple enough way, just how I like it. Sure, it's always been possible, but nice tools like rclone bisync and tailscale make it significantly easier nowadays.

The script is here. It ended up being a handy opportunity to practice my Bash too, and it's working pretty well for a minimalist like me. Still, suggestions are welcome!

I spent a while fixing up the genetic algorithm I wrote a while ago; up until the version that is up now, I didn't really like how it looked and the reason was the type hints. It simply didn't click to me why the classes from the implementation weren't compatible with the protocols in the definition and why I had to create the type variables (which I find really ugly, and hopefully won't be required anymore in the near future). Reading up on how to improve it led me to the concept of variance, of which I only had a vague memory from college.

Writing this note to ensure it doesn't become a vague memory again:

• If something cannot be replaced whatsoever, either by subclass or superclass, it's deemed invariant in its context. An example is identity: It makes much more sense to compare a movie with another movie instead of with a song, even if both cinema and music are considered arts.
• If something can be replaced by a subclass, it's deemed covariant in its context. An example is possession: If you need something with a camera, then any standalone camera, smartphone or tablet will do. A telescope, however, might not.
• If something can be replaced by a superclass, it's deemed contravariant in its context. An example is action: If you can run, it follows that you can do something that only needs for you to walk. Conversely, just because I see you walking I can't assume you can run a marathon.

The biggest movie award nominations came out today and to the surprise of nobody, the year's mediocre critical darling of choice got the most nods. I actually thought it was passable when I first watched it, but quite a few people called it awesome enough times to make me watch it a second time, to see if I had missed something. It only made me find it even more pointless and boring, and I ended up hating it and lowered the original rating I gave it.

But really, the worst sin of these awards is the total snubbing of "The Northman", which deserved at the very least a Best Director nomination. And while I'm very happy for "Top Gun: Maverick", "All Quiet on the Western Front", "Elvis" and, specially, "Elvis"'s frontman Austin Butler getting the recognition they of course deserve, his nomination only reminds me of how Taron Egerton was ignored three years ago for "Rocketman" while one of the worst performances I've ever seen (admittedly, by an otherwise good actor) not only got nominated but won. Really, it was the 2020 Best Actor shenanigans what made me lost what little faith I still had back then for awards, so at this point I'm just beating a dead horse.

All that said, credit where it's due: Certain uncomfortably revisionist movie everyone was calling a shoo-in for multiple nominations too was ignored, so that automatically means the academy has at the very least a saner head on its shoulders than some other award voters.

Well damn, I was tentatively excited about everything going on regarding large language models and how they're being used, until today it was brought to my attention how they behave when the questions are political in nature. Safe to say, impartial they're not, and the excitement I used to feel instantly became dread. The only silver lining I see is the belief that the people who would use these models to confirm their biases would already do so in other ways, and the people who know better would know not to treat them as source of truth regarding political issues.

Somehow, that thought did not make me feel better.

As a way to write more in my site, and in general, I've decided to properly set the notes category of this blog as a microblog "separated" from the main blog. Had to create some plugins for Pelican to accomplish it, and included the cool way to create slugs I read about in this great blog post by Brandur Leach.

To make sure whoever subscribes to the main blog doesn't get this more "stream of thought" bite-size content, the posts in this category are excluded from the main feed; it has its own feed anyway.

Added microformats, Webmentions, POSSE, backfeed and WebSub support to my blog. In fact, this note was syndicated automatically by https://brid.gy. Now waiting for Webmentions to display them.