For me, I sort of imagine the sensation of eating something and observe how I react to it. If it's a strong positive reaction, that's what I'm in the mood for. Otherwise, if it's kind of a more bored reaction, I cast my mind out for another idea and try again.

For the apple test, while I can visualize the image of one, it's much easier to imagine the sensations associated with biting into one. The taste, the texture against my teeth, the stickiness of the juices, etc.

It's like a science fiction writer or an improv actor doing technobabble. If by chance it knows the actual answer it might use it, but even if not it still has to say _something_ that sounds plausible to a layperson. It'll never say it doesn't actually know, because their character it's acting as _should_ know.

> lacks basic features like dealing with compressed files – I don’t count a decades-old cumbersome wizard-style interface with countless steps to go through just to unpack a compressed file to be even remotely acceptable in 2026. Dolphin and Nautilus handle compressed files entirely transparently and much faster than Explorer does, and once you’re used to that, going back to ’90s style compressed file management almost feels insulting.

I can open a zip in a normal file explorer window just fine. Maybe this is something stripped from European versions for anti-monopoly reasons?

Anyway here's what I've done with my Win11 install to make it tolerable:

- Disable onedrive

- Make a secondary local admin account and use that exclusively instead of the the main Microsoft account one (since the library paths in the main account all have onedrive in them even if it's disabled)

- Disable web search in the start menu with group policies

- Remove task view, widgets, the search bar, etc. from the task bar

- Put the start menu in list view and remove every pinned item

- Left-align the taskbar

- Show file extensions in explorer

I'm sure there's a lot more I had to do, but I don't remember off the top of my head.

The Splinter Cell lighting stuff never made much sense anyway, since you'd be perfectly fine if you were standing in a shadowy patch while making a clear silhouette on the illuminated wall behind you.

The strangest misbehavior moment I remember from the first game was like this: There were two NPCs working on computers in a dark room. Behind them was an open door to an illuminated office. I walked into the office and shut the door, they didn't notice there was suddenly no light coming through. I turned off the office light, no reaction. I opened the door again, and then they noticed the office light had gone out.

In the same way, seeing an incompetent stage magician fumble before you see a very talented one perform flawlessly will help you understand that it's all sleight of hand. If you jump right to a professional performance, you might think it's real magic.

For a restaurant, a slop logo gives the impression that the owner doesn't care about the details and has no taste.

Beyond that, the use of generative models is a big moral issue for a growing number of people.

LLMs are pattern-matching machines. They keep the pattern going. Once "the agent disobeys the human's instructions" has made its way into the context, that is the pattern that it's going to keep matching. No amount of telling it to stop will make it stop.

The only possible solution is excising it from context and replacing it with examples of it doing the right thing. Given that these models have massive context windows now and much of the output is hidden from the user, that's becoming less viable.