Notices by Mallory's Musings & Mischief (malcircuit@thingy.social), page 4

Modern masculinity: I am very tough and can survive any situation life throws at me (especially if it's of the life-or-death variety), but the idea of owning or wearing anything that is an especially bright or pastel color terrifies me

Transition was worth it, if only because now I can enjoy the full spectrum of colors without feeling weird

Just to be clear, you don't need to change your gender to enjoy colors. You just need to admit to yourself that gender is bullshit and doesn't automatically determine anything about your preferences.

A piece of wisdom I learned long ago is that it's always better to run *to* something you want instead of running *away* from something you don't want.

I often think about this in the context of jobs. Almost every job will have its moments when things just suck so much you want to leave, so quitting a job simply because you don't like it is unlikely to work in your favor. Replacing one shitty job with another shitty job doesn't help anything.

A better strategy is to understand what it is that you want in a job and begin looking for that, because, even if that job doesn't work out either, at least _something_ will have improved.

@thomasfuchs I'm glad I've been collecting a lot of narrowband data lately, because now I'm motivated to get fucking weird with it

Just watched an #astrophotography themed video about the "True Color Problem" and it pissed me off.

The only "problem" is people asserting that there's an objectively "correct" way to perceive the universe. "Color" is an arbitrary and completely subjective concept that exists only in the minds of humans.

If you want "accurate" images of shit, go download the raw data from Hubble or whatever and stop whining about people not making art you like. Get over yourself.

If you're interested in getting into #astrophotography and want to get a "real" rig instead of a "smart" telescope like the Seestar or Dwarf3, keep in mind that you don't need a ginormous, expensive telescope to get great pics.

A lot of the images I share come from my Askar FMA180pro + ASI585MC Pro, which is a great combination if you want a small, lightweight, and relatively inexpensive setup. Half of the subframes I used to make this image came from that scope:

https://pixelfed.social/p/malcircuit/867409064266126690

Also, an additional benefit to using a small scope is that guiding is significantly easier, and you don't need a fancy mount. Even to this day, I often use my FMA180 setup on my Sky-Watcher Star Adventurer GTi, which is a very affordable beginner mount.

It's easy to fall into the trap in #astrophotography of chasing the biggest and best hardware, but it's not necessary. Like anything else, the quality of the end result has more to do with the time and effort you dedicate than anything else.

My suggested budget beginner #astrophotography rig:

- Askar FMA180pro
- ZWO ASI585MC Pro (or use a DSLR if you already have one)
- Sky-Watcher Star Adventurer GTi
- ZWO ASIair+
- SVBONY SV165 guidescope (or whatever is cheap)
- ZWO ASI120MM-MINI guide camera
- Sky Watcher mount USB adapter (https://a.co/d/eJb1XhI)
- Any dew heater strip
- Bahtinov mask for FMA180

Bonus gear for more money:
- Askar Color Magic C1+C2 duo-band filter set
- ZWO EAF + FMA180 EAF adapter bracket

From this perspective, LLMs look less like intelligent entities and more like viruses.

A biological virus is just a bit of genetic code in a little protein packet. It doesn't *do* anything on its own. It's not alive. It has no goals, no wants, no objectives. Nonetheless, it will insert itself into a cell and repurpose the existing machinery to mindlessly make copies of itself.

Same with ChatGPT. It has no goals or desires, but it can hijack people to make more of itself.

/end

And because it's been optimized to be fluent in language, it can be extremely effective at giving the impression of usefulness without actually getting anywhere close.

It's an expert system, but for bullshitting. It can bullshit at superhuman levels. It can out-bullshit any human alive, and it has managed to convince many of the most powerful people in the world to dedicate enormous resources to its persistence and growth.

/3

Someone here said that LLMs are the "paperclip machine" but for GPUs, and I think there's something to that.

ChatGPT isn't nearly intelligent enough (or intelligent at all) to actually have a "goal", "plan", or "strategy" in the way that people do, but it doesn't need to. All it needs to do is convince enough people to do those things for it.

/end

People like to talk about LLMs as if they are a dramatic deviation from previous "AI" systems like AlphaGo. Those so-called "expert systems" have been trained and optimized to perform a single (or a few) well-defined tasks, while something like ChatGPT can "do anything". But that's a lie. ChatGPT and similar systems are just as focused and optimized on specific tasks.

The most obvious is to use language in a way that's as indistinguishable from a human as possible. To talk like a person.

/1

The second (and in my opinion, much more insidious and dangerous) task that LLMs are optimized to perform is to give the user the _impression_ of being useful.

A lot of times it does this by actually being useful, but that's purely accidental and incidental. It's only goal is to make you *think* it did something useful, and by any means.

That's why ChatGPT "hallucinates" facts; truth is irrelevant, only the _appearance_ of truth matters. A simulacrum of truth can be just as convincing.

/2

@thomasfuchs A lot of programming is more similar to plumbing than doing anything particularly interesting. It's figuring out how to connect the output of one library/framework to the input of another to do something useful. That shit is often abstract in the extreme, incomprehensible to a normal person, and not very interesting from a computer science perspective.

That's why I like low-level stuff; it feels more real to flip bits in a register, and knowing how a CPU works actually matters.

Scientific illiteracy makes me sad

Dating is annoying. I really am going to be alone forever, aren't I?

Average United States citizen: Trains are so expensive! We can't waste tax dollars like that!

Meanwhile, every level of government in the US: Dumps infinity dollars into repaving roads every summer.

This post was brought to you by my friend who just got back from visiting Toronto for the first time, absolutely loved using their transit system instead of driving her car everywhere, and spent a lot of time complaining about how backwards and dumb the US is.

Welcome to the #WarOnCars club lol