HNHacker News
TopNewBestAskShowJobs

thomastjeffery

6,561 karma · joined November 18, 2016

Just another mind, hacking in to the winding maze we call life. Or just looking at what's already here from whatever perspective I can find.

Do you explore reality breadth-first or depth-first?

thomastjeffery@gmail.com

submissionscomments
thomastjeffery··on How the words we teach English language learners changed
> It's also odd to me to say an LLM isn't subjective.

I was saying the opposite. The core feature of an LLM is that it is subjective. The implications of a written expression (prompt) are not well-defined objective truth, but instead a vague probability. We can compute the probability, but that doesn't ever intersect with logical reduction or arithmetic; so the implications we get are just vague guesses on the progression of the story. With enough examples and training, the LLM can guess correct arithmetic answers, but critically, it does not actually perform the arithmetic that verifies them.

thomastjeffery··on How the words we teach English language learners changed
Increasing precision is good to the end that is disambiguation. The problem is that that end is not always what language is good for. Metaphor is an incredibly useful tool, because it allows us to convey meaning with imprecise definition, but precise implications.

The meaning of an expression is its implication, not its definition. The implication must be decided by the reader. This is either done by strict definition, or by arbitrary inference. Inference is a powerful feature of language, because it expands the set implications that can be expressed without abstracting the language. Without inference and metaphor, we must instead create a tower of well-defined abstractions that rule over us; having already determined what implications are relevant: objective truth. With inference and metaphor, the writer/speaker can articulate their implications with familiar language; relying on the reader/listener to infer from the expected relevant perspective: subjective truth.

Essentially the tradeoff isn't whether or not you get precision, but whether that precision can be shared by the writer and the reader, or be managed entirely by the writer. Defined abstractions allow the writer to be unambiguous to anyone (objective), while inference allows the writer to offload that responsibility to the context and perspective of the reader (subjective).

Does that make sense?

thomastjeffery··on How the words we teach English language learners changed
> It’s as if the world now requires you to be more precise about everything.

This is something I've been thinking a lot about. We have trended from subjective language to objective language. Why?

Computing. Software is written with objective language. Everything is clearly unambiguously defined. Blue is no longer a category, it's #0000FF. Logic must always reduce to a binary truth value. Most of what we have to talk about is somehow relative to software. Software even structures most of what we write! We don't just talk to each other, we tweet, email, message, post, search, etc. These structures each imply a specific set of phrase structures that can make sense.

Lately, it's hard to go even a day without reading some complaint that such and such was written by "AI" (an LLM). Why is this so obvious? Well, the core advantage that LLMs provide is that they don't compute. Inside an LLM, there is no arithmetic, no logical branches, no truth values. Phrases aren't generated to define or to resolve. They are generated to continue. Sure, we can direct the story to follow the steps of logical deduction, but that isn't anything like calculation. An LLM simply isn't invested in logic, precision, correctness, etc. the way we expect modern writers to be. It's not the em-dashes or the word choice that illustrates this, it's the fundamental perspective of the system.

We are sorely missing subjectivity. Natural language never was, and never will be, computable. You can't reduce a natural story to binary truth values without choosing an arbitrary perspective that resolves its ambiguity. The more precisely abstract our language gets, the more detached from reality our stories become. The more objective our assertions about reality are, the less relevant they can be.

My answer to this is to make the arbitrary choice of perspective a first-class feature. If we can explicitly decide what meaning is relevant, we should be able to weakly solve natural language processing. It seems like a pretty simple and obvious idea, but so far is easier said than done.

thomastjeffery··on F*: A general-purpose proof-oriented programming language
Then why are we all so interested?

Examples provide more than syntax. It's the semantics that we care about most.

thomastjeffery··on Flux 3 X Mimic: The Next Generation of Video-Action Models
Do you like horror? That's the only genre where risky projects are getting consistent funding. Obsession and Backrooms are both really good.
thomastjeffery··on Flux 3 X Mimic: The Next Generation of Video-Action Models
No Other Choice (by the same director) is incredible, and even relevant to this post.
thomastjeffery··on Ghost Cut – or why Cut and Paste is broken everywhere
Copy and cut are just redundant steps anyway. In Linux (generally), you can simply middle click to copy the current selection. Maybe we could add shift+middle click (or something) to do a move instead of copy.

The deeper underlying problem is that we treat arbitrary UI/UX implementations as objective truth. There is one decided way your computer can be interacted with, and you the user must accommodate it. This is what I want to change.

thomastjeffery··on Agnes Callard’s theory of the uni-context
Maybe not to your self, but it has a significant effect on the actual interactions you have.
thomastjeffery··on Agnes Callard’s theory of the uni-context
Yes, but the context itself is still omnipotent to everyone who participates with it. It's not a bar if it's everywhere. Wearing a mask to the everybar is a way to compensate for the fact you can't leave. You must fragment your self to compensate for the global scope of the bar, just like using a struct to reinvent lexical scope in a globally scoped program. The problem isn't really solved, it's just semantically moved.

What we are really missing is the ability for expression to be subjective by default. When we participate in socially global contexts, everything we read and write must be coherent to the generalized expectations of the entire group of people who are participating in that context. Instead of your words being taken out of context, they are constantly assumed into the context, implying your own interpretation is objectively wrong. The meaning of every expression is decided and relevant, even when it shouldn't be.

thomastjeffery··on Agnes Callard’s theory of the uni-context
Unfortunately that isn't a solution. When you keep separate identities, the only thing that can exist across platforms is your own participation. Everything you say and do is isolated to whichever identity and platform you are using in that moment. You still don't have the opportunity to exist completely, because your self has been fragmented. Even if you did manage to create a cross-platform identity, the product of your participation is fragmented, and every story you tell is objective to that platform's context. Even if you tell a story that links across platforms, you are still isolated to that specific cross-platform context.
thomastjeffery··on Agnes Callard’s theory of the uni-context
It's a shame the title doesn't just include the word: "uni-context". It's a really useful and interesting perspective.

The other relevant word is "objectivity". There so many systems in our society whose context we surround ourselves with, it starts to feel like every subject is objective. The reality is that every subject is subjective.

I think one of the big drivers for this dynamic is that our social systems are facilitated with software, and we always make software as a uni-context. An application is a fixed context.

If we can figure out how to introduce subjectivity into software, that would be extremely useful for both computing and society.

thomastjeffery··on Precursor
So we're just going to let tech monopolists make accessibility impossible, are we? Fuck that!
thomastjeffery··on Precursor
That's even worse
thomastjeffery··on How we can reduce traffic congestion
I'm actually a fan. It's an excellent exploration of sarcasm: what you literally say in a sarcastic tone might have a metaphorical meaning, or it might be a twice application of sarcasm that loops back to literal meaning, like in deadpan comedy.
thomastjeffery··on How we can reduce traffic congestion
I second guessed my usage, but decided it was a good way to communicate my lack of experience and confidence in the specifics.
thomastjeffery··on How we can reduce traffic congestion
A random factoid that will always stick in my brain is that the damage a vehicle does to a road is proportional to the weight on an axel...to the fourth power. I suspect rerouting semis to the roadway is by far the greatest factor.
thomastjeffery··on What Emily Bender meant by "stochastic parrots"
There is more here then pedantry, but if you aren't interested in it, then go right ahead and live your life.
thomastjeffery··on What Emily Bender meant by "stochastic parrots"
I think it's a good distinction to make between having and being, which seems to be what the whole "stochastic parrots" bit was intended to make all along.

It doesn't make sense to say a model is in possession of its self. That's exactly the sort of poetic anthropomorphization that Bender was criticising here, and a good reason to not refer to an LLM as "an AI".

thomastjeffery··on What Emily Bender meant by "stochastic parrots"
But the stochastic parrot (LLM) is the world model, isn't it? What's the difference?
thomastjeffery··on What Emily Bender meant by "stochastic parrots"
How can you articulate a criticism without a repetition of your criticism becoming "inherently pejorative"?

You can't. That's just the way the news goes.

Essentially, what you are saying is that because some people somewhere frame a statement as pejorative, the statement itself is inherently pejorative. By that logic, every criticism ever articulated is inherently an insult.

thomastjeffery··on Footage Shows Cop Stalking Woman After Surveilling Her with a LPR
How about accountability? How could that possibly be too much to ask?
thomastjeffery··on Resetting Xbox
They can't afford to buy every other successful studio, which means that their anticompetitive moat has to be competitive. Otherwise, they could have made the whole thing profitable the usual Microsoft (monopolist) way.
thomastjeffery··on Resetting Xbox
...more like

> We can't afford it, and it's a bad investment anyway

thomastjeffery··on Rayfish, Peer-to-peer mesh VPN with no server to trust
It's because people are too obsessed with providing complete instructions to incorporate any package manager into their instructions.

What we are really missing is an explicit progression from new software to maintained packages across distribution. As it is, each distro expects each package to have a maintainer, and very few people actually want to do that across several distros just to release their software. Generally, the expectation is to instead just wait around for people to make and maintain those packages by virtue of their own interest in your software, but it takes a while, and discoverability isn't automatic.

thomastjeffery··on What's wrong with EU age verification? (Nothing)
And none of them are in my living room, nor do they belong there.

The government does not get to put child locks on my liquor cabinet, or even keep me from leaving booze out on the counter. They get to restrict businesses from selling to people without identification. Note that this has already always been the case on the internet, too.

thomastjeffery··on Ask HN: Is "no source code was copied" still a sufficient copyright defense?
> I'd say it's not really so much a theory as a definition of the word "value"

And that's my point. It's a tautology that demands our participation, while at the same time literally devaluing participation itself.

If you want to understand my position, then you need to step out of that tautological frame, and consider labor as valuable. If you avoid doing that, then you are just avoiding what I'm trying to communicate here.

thomastjeffery··on Most arguments are about ego, not ideas
They say "arguing", but really this is about bickering. Arguments are constructive. Bickering is just engagement. I argue with people so I can construct my worldview, and maybe sometimes even construct the world around me. And, being honest, I occasionally find myself bickering, too; though I do tend to avoid it well.
thomastjeffery··on Data centers trigger voter backlash
Maybe that is the case. I'm definitely not an expert, just sharing my understanding of the situation.
thomastjeffery··on Data centers trigger voter backlash
I was glossing over that detail, yes. You can pay the municipality for filtered water, or dig a well.

If you dig a well, you probably own the lot, and also own associated water rights. Otherwise, you are paying for water processing and municipal reservoir management by the gallon. That's how it works for people in houses.

Your mistake is assuming a datacenter is more like that, and less like an agricultural consumer. It's the other way around. A data center and/or power plant handles its own water processing, which only involves water rights.

thomastjeffery··on Age verification is just a precursor to automated attribution of speech
That, and it defines children as perpetrators instead of victims. What right could a citizen ever claim in a world where even children are guilty?

There is a huge difference between protecting children and prosecuting/punishing children. Age verification can only be an implementation of the latter.

← PreviousPage 2 of 34Next →