numpy-thagoras

I love this idea. Thank you for the examples!

I've been thinking of this as well:

Something like old school Facebook in UI, but functions more like MSN Messenger. You connect to your contacts via P2P, and download/upload updates to your social media network.

Yep, if you were to watch what happens at a 996 shop, it's people literally living their at-home life with their fellow employees for most of the time.

This is great for skywatching but at 32GB it only takes brief photos, and rarely video.

A real setup needs multiple 4K cameras, some kind of LWIR, MWIR, etc. as well as SDR with proper antennae for each of their respective performance envelopes.

I think it's good, but it will not be good at picking up the "targets of opportunity".

I accept any and all downvotes, but...

It's not flying saucers, it's weird orbs / spheres. There is volume of footage, and sufficient quality. I think we're past the point of "is it real" and more at "okay so what is this really and what is it doing".

If we could take Tier 1 reports from NUFORC, have tons of metrics available with sensor data, we can make a better guess.

That describes more than just one POS (Point of Sale) company I know of.

You don't need to load the entire world model in order to be effective at a task. LLM providers already do something similarly described with model routing.

The brute force got reduced down to fast heuristics, like Arthur Benjamin's Mathemagics.

There's a whole lot of undecidable (or effectively undecidable) edge cases that can be adequately covered. As a matter of fact, Decidability Logic is compatible with Prolog.

We would begin by having a Prolog server of some kind (I have no idea if Prolog is parallelized but it should very well be if we're dealing with Horn Clauses).

There would be MCP bindings to said server, which would be accessible upon request. The LLM would provide a message, it could even formulate Prolog statements per a structured prompt, and then await the result, and then continue.

A world model is a persistent representation of the world (however compressed) that is available to an AI for accessing and compute. For example, a weather world model would likely include things like wind speed, surface temperature, various atmospheric layers, total precipitable water, etc. Now suppose we provide a real time live feed to an AI like an LLM, allowing the LLM to have constant, up to date weather knowledge that it loads into context for every new query. This LLM should have a leg up in predictive power.

Some world models can also be updated by their respective AI agents, e.g. "I, Mr. Bot, have moved the ice cream into the freezer from the car" (thereby updating the state of freezer and car, by transferring ice cream from one to the other, and making that the context for future interactions).

I don't think he actually understands direct realism, idealism, or representational realism as distinctions whatsoever.

The default philosophical position for human biology and psychology is known as Representational Realism. That is, reality as we know it is mediated by changes and transformations made to sensory (and other) input data in a complex process, and is changed sufficiently to be something "different enough" from what we know to be actually real.

Direct Realism is the idea that reality is directly available to us and any intermediate transformations made by our brains is not enough to change the dial.

Direct Realism has long been refuted. There are a number of examples, e.g. the hot and cold bucket; the straw in a glass; rainbows and other epiphenomena, etc.

Good. The world model is absolutely the right play in my opinion.

AI Agents like LLMs make great use of pre-computed information. Providing a comprehensive but efficient world model (one where more detail is available wherever one is paying more attention given a specific task) will definitely eke out new autonomous agents.

Swarms of these, acting in concert or with some hive mind, could be how we get to AGI.

I wish I could help, world models are something I am very passionate about.

Most code is throwaway, even to you.

Most code will quickly be obsolete anyways, even for you.

Most code you write will be torn apart by often the harshest critic: yourself.

People who chase others' code just to copy it and not to understand it will not get anywhere, especially today.

Share for your benefit first, and share often. That's how I have found my gems, even for code I've written.

Yes, I believe that.

> If your statement was even remotely true then why is this not used in conflicts to devastating effect?

It has been, it continues to be.

Where have you been?

The source code had a very elegant and systematic use of intents (including prefetched intents) and a dependency injection container.

The pattern itself is a little bit different, has some conceptual overhead, but it's also fairly clean and scaleable.

Most of the time, I don't personally look at it as cheap labour because I am just ordering, e.g. 60,000 of something or 100,000 of something else.

It's cheap, yes. I can indeed buy 1,000 of something more locally or from other than China.

But when it comes to scale, needing vast shipments, then they are the ones who can actually ship it and do it reliably. It just also happens to be cheaper, too, which is more of a convenience or cherry on top, than the actual attractive part: vast scale.

Yeah I currently daily drive a 43" monitor and it has been a life changer since I got it in 2022.

I'm still happy with it, would kill for an 8K 43" 120hz monitor but that's still a ways away.

Is it the noise cancellation making a feedback sound, or is it the pressure differential in the ear canal pulling the ear drum back to produce a white noise?

He said that it goes away when he yawns, so I'm thinking it might be the pressure differential.

You write the requirements, you write the spec, etc. before you write the code.

You then determine what are the inputs / outputs that you're taking for each function / method / class / etc.

You also determine what these functions / methods / classes / etc. compute within their blocks.

Now you have that on paper and have it planned out, so you write tests first for valid / invalid values, edge cases, etc.

There are workflows that work for this, but nowadays I automate a lot of test creation. It's a lot easier to hack a few iterations first, play with it, then when I have my desired behaviour I write some tests. Gradually you just write tests first, you may even keep a repo somewhere for tests you might use again for common patterns.

Yeah, we really need LLMs to work swimmingly with Lean 4. It is currently hot garbage and it does not understand proof composition, exploring proof extensions, lemma search, etc. It does not explore an open-ended node to a mathematical knowledge graph by substituting various options.

I'd happily work with someone on a conversational theorem prover, if anyone's up for it.

"- moderation

- spam, which now includes scrapers bringing your site to a crawl

- good faith verification

- posting transparency"

And we have to think about how to hit these targets while:

- respecting individual sovereignty

- respecting privacy

- meeting any other obligations or responsibilities within reason

and of course, it must be EASY and dead simple to use.

It's doable, we've done far more impossible-seeming things just in the last 30 years, so it's just a matter of willpower now.

That reaaally does look like the Stonks guy.

Man we haven't change much, have we? :D

"...one where founders don't have to be particularly talented to hit the jackpot."

That's where we're at right now anyways.

"If tech companies are this stupid, it ought to be very easy to disrupt and usurp them by simply shipping--"

And that's how we got here.

The code rot issue will blow up a lot more over the next few years, that we can finally complete the sentence and start "shipping competing code that works".

I worry that mopping up this catastrophe is going to be a task that people will again blindly set AI upon without the deep knowledge of what exactly to do, rather than "to do in general, over there, behind that hill".

It's cool and all, and I like it. I have TidalCycles installed and have played around with it.

My only criticism is it makes music feel like CSS. In some ways it helps with theory, yes, but the DX is more like Tailwind.

This wasn't the first time a diesel-electric has caused havoc for US Navy expectations during a war game.

The article even mentions the added use of layered defence to try and counter this move.

Sketchy and Vancouver go hand-in-hand.

I live and work here. I play it straight and do my best.

However, the culture here pushes for a lot of sketchy things, often to make up for lack of productivity or some other deficiency.

I have prompt-injected myself before by having a model accidentally read a stored library of prompts and get totally confused by it. It took me a hot minute to trace, and that was a 'friendly' accident.

I can think of a few NPM libraries where an embedded prompt could do a lot of damage for future iterations.

I will upgrade when I see actual 10x optical zoom

Don't forget live photos as well