queerlilhayseed

That’s how you do a god damn gritty reboot.

I really love this show. I’ll rewatch it every now and again. After my own mom died, “Free Churro” really hit differently. It’s a really beautiful dive into the thoughts and emotions that can come up when your mom is recently dead but your mom was also chronically an asshole.

The exploration of Beatrice’s and Bojack’s relationship thoughout the series, and the flashbacks to how Beatrice’s own trauma shaped her and her relationship with Bojack, and Bojack’s final acceptance that now, finally, he knows there’s no chance she’ll ever be the mother he wanted or the one he needed, really helped me come to grips with what I was going through, mentally and emotionally.

The rest of the show is also really good. I’m a big fan of Character Actress Margo Martindale.

I’ve gotten critiques that I write in walls of text, occasionally, but that just looks like a single normally-sized paragraph to me. Maybe I’m just old. I am using a desktop browser on a 4k monitor, so maybe that has something to do with it. Just eyeballing it it looks like several mobile screens worth of text, maybe it should be broken up for mobile viewers. I guess there’s a paragraph break for the last two sentences… but it’s seven sentences total. Is that too much for one paragraph?

Yeah, I think of it kind of like protocol negotiation. If I speak a technical protocol my conversation partner doesn’t, my options are either to get them using my protocol, I switch to a common protocol, or we don’t speak. Sometimes, if the topic is important enough, the first and third options are preferable. Other times, particularly in casual conversation, I opt for the second one. I still try to be precise, and that can mean an occasional specialty word and accompanying definition, but it’s costly to the conversation and if I’m not careful I can get bogged down in explaining minutia instead of getting to my point. Maybe it’s more like treating new concepts economically rather than treating jargon as a discrete protocol, though I tend to think of jargons as protocols, you can kind of add them into a common protocol a la carte. They’re just computationally expensive for people absorbing them if they’re also trying to follow a conversational thread at the same time, and too many can blow the budget and cause people to tap out, so I try to add only the ones I really need.

The internet is a place where humans can let there hair down.

My observations: it can be for both reasons, though I think most modern ones will be better than the human average unless you ask them to add typos. Which they can do, but they have a hard time being consistent with it, especially if they’re trying to pose as the same person over multiple responses (e.g. a reddit or fedi bot), they won’t have the same type and frequency of typos that a human would have consistently across posts.

For me it depends on the context. I love words, and it’s a mathematical fact that there are more long words than short ones, and I love using all of them. But, I also love communicating with people, and if I can use a word that most people know instead of a word most people don’t know and it works well enough, I prefer to simplify. I have a tendency to indulge my own vocabulary and I think consciously minding it makes me a better communicator.

I got the SNES Classic just for Donkey Kong Country. The other games were a nice bonus. I have some novelty electronics that I never use anymore but that one gets pulled out on the regular.

P.S. This is a hypothesis, I haven’t even designed the test for it, much less run it. What follow are my suppositions.

I think whether or not it’s a good idea depends on how similar all the models are. I don’t have a rigorous definition of “similar” but things like similar training data, similar design methodologies, similar QA processes would all contribute. Theoretically (I think), if they’re all dissimilar, they should each catch errors the others miss. However, the more similar they are, the more likely they have the same biases and weak spots, and your error rate from a response + verification may be the same or even higher than the error rate for just the original prompt, and you’d be unlikely to detect those errors using just two similar models. It can instill false confidence in the results because you’re doing something that should in theory increase the validity of the data, but in practice might make no difference or even make the quality of responses worse.

I think it’s tricky. It’s kind of like adding LLMs like vectors, and hopefully the effect can soften or at least reveal the shortcomings of individual models. Is it a good idea? I don’t know, I think there are good reasons to think it’s a waste of time and resources. I certainly think I’d need a better explanation of what use it would be before I spent more time building it. But I still think about what use it would be from time to time; I haven’t decided that it’s a bad idea yet.

One of the projects I started and never got to a satisfactory end state was basically that, plus a judging round. Every model would respond to the same prompt, then every model would evaluate every other model’s response for accuracy and completeness. Then the results would get logged to a spreadsheet.

It’s simple enough, but for N models it requires N + N^2 model calls so it takes forever to run any decent dataset on consumer hardware. If I had the resources and a way to run it that didn’t fry the planet, I think it would be a cool running set of comparative benchmarks. IDK if it’d be useful at all but I’m still interested to see the data.

Oh neat. Yeah, if something like that had existed (and I’d been aware of it) I probably would have used it instead of building my own shoestring version.

Nothing so fancy. I just made a little python script to prompt the first model, wait for a response, then prompt the next model with the initial prompt + the response, and so on. It was very hacky and slow.

Yup, ollama, various models. I initially downloaded it because I, along with thousands of other people, wanted to see what would happen if I made models debate with each other after RAGging them with various books (The Prince, The Art of War, The complete works of Shakespeare, etc.).

The results were uninteresting and I abandoned the project pretty quickly. I’ll sometimes use them for code analysis but they’re too slow on my rig to be really useful.

It’s very confusing when it happens to someone you know because they usually aren’t horrible people. Nor are they particularly stupid, at least not more or less than your average human.

I think people in left-of-center spaces like to say things like “The Republican party is a cult” as a sort of disparaging cliche without really appreciating what a cult is or how they operate in real life. They see things like “Donald Trump ate a live baby on camera today, also his support among Republicans is at 90%” and assume that 90% of Republicans are cool with infant cannibalism, when in reality those Republicans just don’t see the same news we do. They are actually in a cult, which means (among other things) that they are conditioned to believe the cult leader (Trump, at the moment) and his approved mouthpieces, and only them. They’re reminded, constantly, that the left is deranged and willing to say anything to make Trump look bad, so when Trump does something so heinous it actually does penetrate their media bubble, it gets brushed off as more evidence of the left’s lunacy. It’s a very resilient form of conditioning and it’s been going on way longer than the internet’s been around, and it’s… really hard. Because they aren’t horrible people, mostly. They’re mostly scared, and broke, and confused about the world, and they flock to people like Trump because they have been told by their handlers that he’s their best bet. Once he’s dead or too toxic even for the Republicans, they’ll be reoriented toward the next leader.

The problem of the dead internet isn’t that there are no human users, but that the human users are isolated from each other or herded into ideologically suitable echo chambers, where misinformation and lies can be harder to resist because they already have momentum. It’s also hard to prove because we’ve demonstrated the inclination to do it to ourselves even without malign orchestrating influences like giant corporations.

An indicator of a dead internet wouldn’t be that no one in your IRL experience uses the internet, but that either A) their experiences are extremely congruent with your own (you’re both in the same bubble) or B) their experience of the internet reality has no shared basis with your own (only one of you is in a bubble, or you’re both in separate bubbles). Which… does happen to me occasionally, especially with older folks. A lot of people are caught in the dead internet of facebook, and are being groomed and manipulated like cattle. The export products of the bot farming industry are influence, votes, hatred of minorities, etc. I suspect a lot of the MAGA elements of my family are deep into dead internet traps, though of course it’s hard to get an accurate picture of their media diet because they don’t trust me enough to share it.

Do I think this means the internet is now Certified Dead? No, but I think it’s a sliding scale of deadness and it’s somewhere between 0 and 100 percent. Where on that scale we are is difficult to pin down.

the actual essence of the community and the thoughts and ideas are reflected not in your posts but in your ideologies

lizlemonmegaeyeroll.gif

Very impressive that you can divine the essence of a community, not reflected in posts, just by reading a bunch of posts and making assumptions about them. That’s a neat trick.

This reeks of manipulation. Console makers have an interest in making exclusives look attractive to studios. Players are interested in exclusives because… why? If you already have an Xbox why do you care if a title is exclusive or not?

I don’t trust it.

I guess there are some spices that cost the same as a week of food, but I don’t think there’s anywhere on the globe where that’s true for everything that falls under the “spice” umbrella. A box of craft dinner is about $1.10 where I live, a 1oz. container of dried thyme is about $3.50, which is fairly typical for grocery store spices around me. There are spices that are very expensive and those are out of reach for a lot of people, but I think most almost everyone has access to some cheap spices.

A lot of them are easy to grow / are invasive species that are hard to stop growing: I have mint in my yard and I’m not sure I could kill it if I tried. Every year I cut it back and every year it grows back strong.

Spices are pretty cheap in the US if you don’t mind them being not-very-good, most grocery stores will have a small selection. Still, it’s doable and worth it to keep a small collection of spices even if you’re on a beans and rice budget.

If I try to eat exactly the same recipe for an extended period of time, I find I am much more vulnerable to a fast food craving just for a change of pace. So it works out to be cheaper to get the spices if I account for how my animal brain needs variety.