White Mirror: Story Zero

Nathan Dileas

meta note; I write these in in raw html in with no spell check, no editor, and no llm.

q1

What does “Fact imitates fiction” mean? In context of the previous paragraph, can you find a real example of this happening in your world?

This is one of those questions its impossible to answer without sounding like the avatar of the god of the copybook headings. As simply as I can: The stories we tell and are told, shape reality. Abel gets told he's a real boy. Cain gets shamed. You know how that one ends. Obviously this can't change the bare facts of reality. But when the most important thing is the "human factor" (feels somewhat too specific nowadays) sometimes this is enough to make a difference.

Real example; I do a lot of my own car maintainence. Knowledge is power in a very clear way with these kinds of tasks. Imagining "what would car guy nate do?" in this situtation is a reliable way of getting around something or more importantly, choosing what not to do. Picking a persona and somewhat intentionally inhabiting it is wonderful for peace of mind and confidence.

q2

“The LLM was trained on a Borgesian corpus” -> this refers to Jorge Luis Borges. What is he known for? Why would Lucilius train an LLM specifically on this one author?

I don't know much about Borges. He's one of those people aspiring and perspiring literati seem to be obsessed with. I think he was midcentury, modernist, metafictional.

Why? First, it's not trained only on Borges (paraphrased "all the books I've ever read, everthing I've written"), it's a metaphor. This is a message from the author to us; either he is signalling high status, or communicating something about the nature of the work at hand. As a pratical matter, it sounds like a terrible corpus. Don't you want your super LLM to be more widely read than you? For the sake of argument I will pretend he is fine tuning.

q3

If living in fear is a self fulfilling prophecy, and we know that, why can't we just NOT do that anymore? What's the bottleneck?

Because change is quite difficult and gradual for most people. Just like any large shift in beliefs, it requires proof. The greater the claim, the greater the proof. Some people have been living in fear their whole lives, and have seen it reinforced over and over. The bottleck is condition and information.

If you could wave a magic wand and fix, say, all appliances everywhere, instantly, perfectly. They'll keep running without fail, not needing parts or electricity. That's condition. Information; around me (NE US), some people are going to keep hiring furnace maintainance for quite a while - they know that the chimney is going to get blocked or the heat excahnger will crack and they will be cold and miserable for a few days at best. They need to become convinced that there are in fact, no downsides.

In real life, the physical and information problem is enormous. The Memome is heavily contested. Why aren't we talking about religion when it comes to the Memome? Christianity, Islam, the enlightenment, nationalism, communism all seem to be primarily memome phenomena. Fiction which became fact, in bigger ways that anything to do with AI (yet).

Personally, I try to live in as little fear as possible. But to be honest, there are real things to be scared of in the world - car crash, cancer, dementia, nukes, etc. And sometimes fear is a useful signal. The problem is when it's an unreliable barometer, and we run inside to hide from nonexistant thunderstorms.

Also; assuming the consequent. There isn't any real proof that fear is a self fulfilling prophecy, even if we believe it is.

Concluding

I have to say, I'm disappointed by the book so far. I'm perhaps too hardheaded at times, but there's been a couple glaring errors relating to programming which annoyed me. Several of the stories aren't really WHITE mirror. More like, near future, god in a box shows up and yaps at Lucilius while thin metaphors for LSD experimentation or the simulation hypothesis play out.

Is the whole thing more or less about LLMs? There are other things in the world. The world is a huge, amazing place. As a description of a utopia I much prefer the WTC epilogues. Simple lack of physical threat and implied leisure isn't enough.

Obviously llms and broader applications of AI will heavily affect the future. But I think we have almost no information on which way. I think that in 20, 40 or 80 years, life will be pretty much the same as today. That's my lifespan. Planning too far beyond that isn't my task, but the generations below me. If a powerful actor could guarantee no nuclear war and a comfortable life for my family in that period, I would certainly switch my allegiance. But it would take more than being able to generate some nice-sounding and scary-sounding ideas.