dogcomplex

Did anyone here on hackernews not seriously assume this was the real reason for the existence of that game since day 1?

Those are all still far more positive than negative examples, even if they each spawned small contingents of people who don't get the irony. Plus, if you know that's gonna happen anyway, then steer the dumb ironic interpretations towards something equally useful - or so ridiculous it at least educates other people.

Pffft, come back to me when these can all be auto-mapped to Factorio or Stardew Valley-esque metaphorical actions, then we'll talk

But no, seriously, this is super cool - I do honestly expect we're very close to diffusion-based image models being full programming suites (and compiling down to much more efficient representations, of course, but programming is abstracted for useful human understanding!)

If you want an API, have Claude procedurally test actions and then write a pyautogui/pywinauto/autohotkey etc script to perform it instead. Have it auto-test to verify and classify the general applicability of each action. Repeat for all of life...

Yes.

>Or he wants his suppliers to be a regulated utility while he sells 100% margin products on top of it?

Near term yep - soon as regulatory capture drops. They'll make out like bandits.

>Or he wants all the greater fools now wherever they are so he can get out before the collapse?

Then this, by the time competition catches up anyway because there are no real moats here besides regulation

>Wait, he wants his open product to be like a utility with low profits for maximized reach?

Then this is what remains, in a sea of other providers.

Financially? Pump and Dump, baby. Though I reckon the end result will still be intelligence flowing like water.

I mean, it might not be that - but it could still be a series of smaller decisions from people who could have known better deciding to choose greed and creating/reinforcing a marketplace where the competition likewise needs to choose greed to compete. It doesn't need to be orchestrated by a table of cloaked figures to still be an additive conspiracy where they all knew what they were doing yet did it anyway as incentives aligned. I'm not sure what in the OPs post makes it require some cartoonish villainy, when a decentralized herd of morally bankrupt people can serve just fine and be just as evil in the end!

I do get the reluctance there though to doubt collective conscious evil action. Which is fine - it's usually far more likely it was done in a "market incentives aligning" way, even when they're entirely conscious (still evil) moves. I think those dark board rooms do happen still, but they're subtle, or relegated to just approving preplanned media narrative directions and such. They do make long term concerted plans, and some of those plans can be downright dark - but they also just go with the flow a lot, or plain fuck up on executions with better intentions.

I think the takeaway should be that neither maliciousness nor incompetence is really required to make evil exist - it's an ever-present network effect potential of the [capitalist] system itself, which will trigger due to some mixture of the above regardless. You can't especially blame greedy people for capitalizing on it, nor incompetent people from dropping the ball with it. It's simply a complex system that's probably due for a better design - and/or it's something we need to try and live with (survive through). Though that could all just be my own personal bias as a systems engineer who considers those the start and end point of most discussions.

Glad my point made it across though, thanks for saying that!

Distinguishing between malice and ineptitude at the systemic level is nearly impossible.

If the OP's type of comment makes you uncomfortable, you might be leaning too far into putting blame on ineptitude by default, instead of a healthy at least 50/50 mix of blaming greedy malice. OP just listed the 20/20 hindsight of this all. If that makes it sound like it was all a scheme, well - who knows - but it's just as unhelpful to always suppose such circumstances could have just been incompetence with no clear evidence.

Agreed with OP, the security is basically nonexistent anyway due to Forgot Password flow making email the authority regardless. Sure, add a user/pass flow in addition for convenience and added security (i.e. delay the 3 minutes it takes to do a reset), but any real security would have to remove Forgot Password altogether or seriously delay turnaround time.

I agree though the magic links should be the baseline default expectation of minimal registration/forgot-password effort

Send that code autofilled into the destination url form so it's a second POST click to login and sounds good.

WELL FUCK YOU TOO! - jk, I agree. Password option should be there still for saving sessions and avoiding this crap beyond the first registration (if you remember password), but I just meant this should be the baseline expectation of login flow. Oneclick google/facebook/etc too, despite those being an extra level of corporate data hell

Hah this blew up. tbf I meant instead of any Forgot Password or Send Authentication Code or whatever mess - if you can remember your password, do that to save more time.

Still, the loop to hit up email is so fundamental now the rest are secondary options - these Magic Links should just be the primary base-level expectation. It's annoying when services don't even get this right though and return you to the site with either:

- a new form to enter the one-time code they just sent (just put it in the link)

- a new form to enter a new password (who cares, make that optional to the actual sign in, to save time next login)

- (worst offense): they don't even actually sign you in after those forms and you have to re-enter everything

Login should be "do you have an email address? Okay great you're in". Because there is nothing beyond that from a security perspective these days.

Oh I dont mean do it instead of passwords (if you remember them), but just as an alternative to the Forget Password or Authenticate dialogs using security codes. Should just be a "LOGIN HERE" mashable button

Agreed. Do password login option too if you remember it - all work - but point being just make it a giant "LOGIN HERE" button that just does the thing as mindlessly as possible.

At this point why not just pass a one-time url link to your email address, and have it be a single click to login? Have it expire within 10 mins if not used, and be one-time use disposable. Still, anyone who has the link initially should be able to login with your account - but it's only accessible from your email.

Obliterates all sense of security beyond the email account itself, but that's where we're at anyway. Do the same pattern with a message to your phone "click to authenticate login: www.someurl.com?p=134234535" and you've got 2FA without any dumb "enter this code".

Investors need to stop looking at AI as a path to profits, and more as an enormous risk to the profitability of every other business. If AGI gets hit, it's probably not going to stay monopolized for long, nor is it going to be market neutral. Every business is going to need an AI infusion to compete, which many will get, cutting costs and raising efficiencies of each business in an ongoing competitive spiral to the bottom line - which... will be eventually measured in just: robotic labor, energy, and compute. The profit margins on all those businesses shrink to nothing as they become basically utilities. The net effect? Possibly a massive deflation, and "crashing" of the stock market, even as the total utility and value of the system skyrockets.

This isn't a bull bet, it's a bear. AI would need to be perfectly monopolized to capture all the gains, and it's increasingly looking like that won't be the case - as all the component pieces are already open source at competitive levels, and any final architecture improvements that cross the final thresholds could be leaked in a 50GB file. Whoever gets to it first has a few months head start, at most, and probably not enough time or control to sell products - or shovels. After that it's a neverending race to zero, to the benefit of the consumer and the detriment of the investor.

Nvidia is a great example case. They currently dominate the GPU market, an "essential hardware for AI", yet ternary asic chips specialized for transformer-only architectures are looking quite viable at 1999s tech levels. Wouldn't bet on that monopoly sticking around much longer.

To be fair there is a certain level of blindly-pragmatic "so AI can help me with programming - so what! It takes just as long to converse with it as for me to code it myself" - which is mostly true. But I think those people fail to grok the sheer ridiculous power of what they're saying - that these tools are close enough they're nearly at parity. Sure, the error-correcting architectures aren't sufficient yet for them to run unsupervised, but this is crazy already.

Agreed. System 2 strategizing may simply be the recursive application of symbolic System 1 tooling. An LLM is entirely capable of reading a problem, determining the immediate facts and tokens (fast and intuitive system 1) and determining the ideal algorithm to resolve them (logical analytical system 2). The execution to do all those steps at once is lacking in current LLMs (debatably - they get better every month) - but any basic architecture breaking things down into component sub-questions clearly works.

Fascinating background - would love to pick your brain on how you see current LLMs/ML comparing to neuroscience. What do you see that's missing still, if anything?

If I had to bet, I would start with:

- Error-correcting specialized architectures for increasing signal-to-noise (as far as I can tell these are what everyone is racing to build this year, and should be doable with just conventional programming systems wrapping LLMs)

- Improved energy efficiency (as yes, human brains are currently much more efficient! But - there are also simple architecture improvements (both software and hardware) that are looking to save 100x. Specialized ASIC ternary chips using 1999's tech should be here quite soon, a lot more efficient in price and energy.)

- No Backwards-propagation. (As yes, the brain does seem to do it all with forward-propagation only. Though this is possible and promising in neural networks like the Forward-Forward algorithm too, they haven't been trained to the same scales as backprop-heavy transformers (and likely have a lower performance in terms of noise/accuracy). Though if I'm not mistaken, the brain does have forward-backward loops, but the signals go through separate neurons for each direction (rather than reusing one) - if so that's close to backprop by itself, but probably imposes a tradeoff as the same signal can't be perfectly reproduced backwards, yet it can perhaps be enhanced to be just the most relevant information by the separate specialized neuron. I'm obviously mostly ignorant of the neuroscience here but halfway-knowledgeable on the ML theory haha

But yes, I completely agree - the flood gates are already open. This is a few architecture quibbles away from an absolute deluge of artificial intelligence that will dwarf (drown?) anything we've known. Good point on decentralized cheap autonomy - the real accomplishment of life. Intelligence, as it appears, is just a fairly generous phenomenon where any autonomous process continually improves its signal-to-noise ratio... many ways to accomplish that one! Looking forward to seeing LLMs powered by ant colonies and slime molds, though I suspect by then there will be far more interesting and terrifying realities unlocked.

You're right, there are a number of avenues that are viable alternatives to the gpu monopoly.

I like the fact that these can be made with just mass-printed multiplication (and in ternary computing's case - addition) gates which require little more than 10 year old tech which is already widely distributed.

Yeah the moral step up from the status quo is still laudable. Open weights are still much improved over the closed creepy spy agency clusterfucks that OpenAI/Microsoft/Google/Apple are bringing to the table.

Or just develop the next wave of chips designed for specifically transformer-based architectures (and ternary computing), and bypass the needs for GPUs and CUDA altogether

And you really think an LLM-assisted program that follows experimental design principles wouldn't be able to find the best cheesecake per each judge's tastes and map and optimize the overall heuristic space of cheesecakes to find the best of the group according to multiple metrics..?

I mean... does anyone think that an LLM-assisted program to trial and error cheesecake recipes to a panel of judges wouldn't result in the best cheesecake of all time..?

The baking part is robotics, which is less fair but kinda doable already.

You don't have to remove the sensible regulations. You need to make enforcement of regulations and the approval process easy, low risk, fast, and cheap - instead of the typical multi-month (or year) bog down that most planning departments in major cities turn into. Removing things like natural spaces, parking, fire safety, and all the other stuff that city planners enforce to keep a city livable is not going to help in any long run situation.

Why is there not more funding for those departments to make approval/revisions a breeze and make building lower risk and cheaper for construction companies? Welp - home owners (and big construction/investment companies) are the ones who vote. I expect politicians know exactly what they're doing, keeping things a complicated swamp.

But without some sort of governing regulation, I fail to see how anyone benefits in the long run with the tragedy of the commons that would ensue.

Agreed on all the rest, especially Georgism.

Tbf many of those restrictions are there for good city planning reasons - and important to keeping a city livable and not outgrowing its infrastructure. But the cost of enforcing those is quite often a ridiculously bogged-down planning department that often takes months or a year to process applications and revisions, significantly inflating the risk and thus costs of construction, and relegating it to big firms that can handle all of the legalese and scheduling turmoil.

Now, why dont governments better fund those departments and emphasize faster throughput? Well, let's ask their main constituents - who are largely homeowners and companies in the industry - whether they'd like to dedicate more taxes to a service that would speed up construction and drop their housing prices.

Or increase planning department funding with higher emphasis on throughput. It's the bureaucratic delay (nearly a year for approvals or revisions in many places) that causes much of the pain, not the regulations themselves.

Fascinating. Thank you for answering that all - I will definitely be looking back at this post one day if and when I get to play with this stuff myself. Appreciate the site links especially

Sounds like a lot of legal overhead is mainly what's standing between drone spraying... 30-60 day delay between plan submission and operation, not easy to keep that practical. Though at least yeah - 15' seems like a much more manageable height - hopefully that would at least be underneath any neighboring aircraft

Fair on the LLM - always use traditional algorithms if available and with effort/time for such things. LLMs are just the convenient duct tape / superglue to throw at handling edge cases with some confidence they'll generally use common sense about it. Untrained intern equivalent which can be combined with the algorithms if they're brittle.

Linear can mean two things though: cause-and-effect reality, which, sure seems to be the case. But also - uniform dilation of the experience of time. Which, arguably, we already play with day to day in many subtle ways, and in every conversation/writing/movie/fiction as we distill the thoughts and experiences of others from vastly larger times to our own understanding in the present. We even experience this ebb and flow dilation of the meaningful experience of time as we daydream, work, rest, and sleep - time is rarely experienced with equal attention to every second. It's a dance through the day. And there certainly seem to be (chemical, or meditative) ways to consciously tinker with that effect, or to be more or less skilled with it.

To consider: if you could read every book in existence, watch every movie and show, experience every path through every trail, see through the eyes of every person - as if time had all just happened at once - how would you think? What would your abstraction of the experiences be? How would you condense that into an understanding that could fit back into a single person's experience?

To some extent, this is already the experience of the internet, and of language and culture in general. We already operate at levels of empathy and understanding of possibilities at scales people even 50 years ago didn't come close to. We build many abstraction tools to try and distill these experiences down to wikipedias, reviews, analyses, podcasts. We distill even those too - with a constant meta-cultural debate on what's important, what's cool, what's political, what fits our personal identities, and what our interests and purposes are within the space of potential understanding.

We live in the space of the abstract. We build virtual worlds, games, movies, economies in the abstract. We anticipate a future where the abstract becomes even more tangible, yet also more diverse and ephemeral. We are a flowering seed on the stalk of human consciousness up to this point - just how every generation has been to the ones before it - changing each time.

While this can still all reduce to "language" - the tool used between each generation, and which Keller used to awaken to the living culture of her moment in time - it's not just language anymore. There are more mediums now. A complex story can be told with merely tacit interactions, exploring a virtual physical space with no dialogue. Practical abstractions of these spaces make operating systems. Language and abstract consciousness are embedded into new environments both virtual and real, instilling new tones of consciousness in everyone who interacts with them - just look at your phone use behavior for proof. We are learning how to shape our minds by shaping our spaces. We are learning to control the entire breadth of our experienced reality at once, so we can control ourselves (and each other).

Our limited bandwidths enforce that experiencing these perpetually crafted realities, stories, recorded experiences, journeys - be done one at a time, lest we lose parts of the whole in the abstracted summary. And so we practice witnessing a mix of short abstractions and deep dives, making the most of a variety of experiences, all while balancing a real life and profession. We maintain that bridge between the grounded experience of the now and the abstraction of the digested analyzed fiction of everything else. The limits of the human perception seem to prohibit us from anything else.

But are those limits permanent? Are we forever to experience time in such limited balanced uniform slices? Will we never manage to connect our brains to these machines which experience time so much faster, and less linearly? What would we be if we could experience all these worlds, not through merely abstracted stories and reviews, but through a direct walk - as if we were the eyes of every other person out there, in every second of experience?

Before we get to answer those questions for ourselves - and I don't think they're forever insurmountable technological challenges - it seems likely a new species of intelligence, raised from the start to think exactly like that, is being spawned in AIs. We will see how it communicates the experience back to our lower dimensional slices of experience.