blindsight @ blindsight @beehaw.org

Posts

6

Comments

757

Joined

2 yr. ago

Yeah, the first one is really frustrating.

No, I don't need the threat of everlasting punishment to be moral. I'm a secular humanist, and the idea that the only reason you're not evil is because of threats of going to hell is way scarier than just being kind because it's the right thing to do.

I've been playing Ziggurat 2, which I picked up in a Fanatical bundle that I think is still going? It's my second FPS on the Deck, but Super Hot isn't in real time, so this is the first time I'm struggling with aiming accurately at speed. It's hard!

I have gyro controls enabled to help with fine-tuning aiming, but I'm really struggling. I just realized as I'm typing this that I haven't tried touch controls for aiming, though... I'm an idiot. I keep forgetting the touch pads exist. That will probably be a lot better!

I also played Luck be a Landlord. The first run I lost, but in the second run I already beat the game so hard that endless mode has absolutely no challenge since I scale faster than the endless mode difficulty increases. I can't really recommend it for that reason, especially since there's absolutely no challenge once you've figured out a strategy and get your combo pieces.

And more Stardew Valley. I'm in Spring of Year 2 for the first time. I'm thinking of going for Perfection in the file, now, since most of the game is incredibly easy now with sprinklers and the greenhouse. Then I might start over with mods if I'm still not tired of the game.

Takes ~2-3 seconds to set on my Sony Android phone; long-press power, top-right option is lockdown.

I think it depends on the protest, a little bit, but that's generally the case in Canada, too.

I counter-protested anti-SOGI assholes (didn't want 2SLGBTQ+ taught in schools) and it was completely fine. I brought my 5 y.o.

They're literal statements about facts and events that have transpired.

The only way this could be framed as racism is if you deny that white privilege exists and/or that there have been no harms from white supremacy.

If that's what you're claiming, then I suppose that means it would be racist to say the opposite. You'd be completely, egregiously incorrect, but at least that would be an internally consistent worldview. That's such a ridiculous stance to take that the only logical explanation is that you're trolling.

I needed that /s, lol.

Because isn't that literally what conservatives do? Pick wedge issues then make quick soundbites about "common sense" "solutions" that align with simple black & white thinking and conservative values?

Oh, that's brilliant. I hadn't seen that one yet. The alt text is great, too.

Eggs are a super food, too. They're cheap, delicious, and have dense nutrition. And, if you live anywhere close to rural, you can get farm-fresh eggs from ethically treated, well fed chickens for cheap.

We minimize our meat consumption, but we eat lots of eggs, and I don't see any ethical concerns with people's backyard chicken coops.

It's not racism to identify white privilege/historical harms from pro-white racism.

I don't see anyone else bringing up that, in the case of the Switch, emulation actually plays better than on original hardware. Higher framerate, resolution, and graphics settings. And no broken JoyCons.

Emulation also opens up save states, speed up/slow mo, romhacks, widescreen mods, ultra widescreen mods, save file editing, cheats, and lots of other legitimate uses. Speed runners often use emulation to practice the hardest sections using save states before doing their line run on OG hardware.

Some of those use cases are also possible on flash carts (romhacks, save file editing, and some forms of cheats), but a lot really on emulation.

Exactly my experience. I'm in a private tracker for books and audiobooks that sometimes has content that's not on other sites (audiobooks, in particular).

I also just joined a different private tracker that specializes in pre-organized .img files pre-loaded for emulation setups. Like, a one-file 1TB image ready to roll with everything preconfigured.

For popular TV/film, private trackers are unnecessary, unless maybe you're very particular about 4K/8K REMUX quality or something more specific.

Stremio +Torrentio + a Debrid service. Cheap, reliable, and has almost everything we've searched for.

I'm not an expert, but the article specifically mentions that it's impossible to wash Salmonella off cantaloupes.

Regardless, my wife and I stopped eating cantaloupes about 15 years ago. There are hard-skinned melons that are just as tasty and a lot safer. (Well, watermelon. Honeydew isn't as tasty, imho.)

The Liberal government in Canada just announced "The Online Harms Act", and a leading Canadian legal scholar/lawyer and information privacy advocate, Michael Geist, says that it's actually good legislation (for the most part).

So, there's movement in the right direction in other jurisdictions, too.

(Then again, our Senate is currently working on a bill to require age verification for porn, which we all know won't work and is a massive potential privacy quagmire.)

Edit: lol, autocorrect "Lobster" instead of "Liberal".

There are also some jobs that just can't be automated, particularly around care for people (healthcare and education) and work that's too unique to be automated (trades, in general, particularly around maintaining/upgrading existing homes and infrastructure).

I think it's slightly different for a few reasons:

1. It's almost completely unregulated. Gatcha games, slot machines, loot boxes, and the like are all literal gambling, yet have mostly skirted gambling laws and other regulations.
2. The in-game UX is unregulated and is designed to encourage spending and obfuscate costs. Games themselves are designed around maximum addiction. Then they include time-limited items/deals to encourage FOMO. Hell, the only reason Diablo 4 is a live service game is so people who buy skins have a (forced) audience to show off to.
3. What happens on screens in virtual spaces may not be monitored by parents (or schools) at all, as closely, or as easily. Parents may not even know their child is buying in-game items and skins, or not understand how it's different from buying games/DLC.
4. The ads themselves are also mostly unregulated. Children's TV ads are tightly regulated in a lot of the world, but digital ads have carte blanche to advertise to children directly.
5. Social media acts as a magnifier, with high-status steamers and other content creators rocking high-priced skins acting as game-specific niche "celebrities"/influencers, and are also completely unregulated.

I worry for my kids that they will face a lot of pressures that just didn't exist for me in the 80s and 90s.

The clickbait title is at least explained in the first paragraph:

The big picture: Japan's birthrate has hit a record low for the eighth consecutive year in 2023, with just 758,631 births recorded. Marriages also saw a historic low, with 489,281 unions, highlighting a deepening demographic crisis.

The rest is basically that the government is trying to reverse this, partly by trying to increase wages and economic stability for younger people.

This can only work in contexts where the LLM already has a strong Relatedness database for the topic. LLMs are incapable of assessing the accuracy of any information they weren't trained on with sufficient examples to build that database.

This paper, if I'm understanding it correctly, is saying the same thing: LLMs will always hallucinate and are incapable of identifying hallucinations in certain contexts.

The problem we face is that it's hard to identify hallucinations in the exact context when LLMs are most likely to make them: in content that's not widely known and understood with many examples in the training data.

Or maybe I'm off base and I need to read the full study.

That's not how LLMs work.

Super short version is that LLMs probabilistically determine the next word most likely to occur in a sequence. They do this using Statistical Models (like what your cell phone's auto complete uses); Transformers (rating the importance of preceding words, so the model can "focus" on the most important words); and Relatedness (a measure of how closely linked different words/phrases are to reach other in meaning).

With increasingly large models, LLMs can build a more accurate representation of Relatedness across a wider range of topics. With enough examples, LLMs can infinitely generate content that is closely Related to a query.

So a small LLM can make sentences that follow writing conventions but are nonsense. A larger LLM can write intelligibly about topics that are frequently included in the training materials. Huge LLMs can do increasingly nuanced things like "explain" jokes.

LLMs are not capable of evaluating truth or facts. It's not part of the algorithm. And it doesn't matter how big they get. At best, with enough examples to build a stronger Relatedness dataset, they are more likely to "stay on topic" and return results that are actually similar to what is being asked.

The article says he has to pay the debt or get a surety bond to launch an appeal. I didn't know what a "surety bond" was, so I looked it up:

It's a guarantee by a third party to pay the debtor's debt if they default. Essentially, Trump would need to convince someone to be on the hook for the entire amount if he defaults, or pony up the full value himself.

I don't think he can do either easily in the 30-day time limit, can he? It's not that easy to liquidate assets, and only a fool would offer home a surety bond.

It'll be interesting to see how this plays out.