photonthug

Actually I think the public tends to generalize their complaints/injuries and act in the most spiteful ways that are available to them. For example, decades of bad experiences at the DMV translates into cries that we should defund the post-office, NSF, etc, no matter how irrational that is.

But capital has a playbook now that's pretty effective at dodging this kind of backlash, like the "advertising without signal" thing that's also on the front page right now is pointing out. That article mentions "Disposable brand identities" which does seem relevant here even if that piece is mainly talking about the relationship between amazon/manufacturers/consumers. Part of what PE is accomplishing is brand/liability laundering, but brands head in this direction anyway before they fail. Consumers can't typically look at list of 10-20 "different" hotel brands and really tell which are under the same umbrella.

And all this is kind of assuming consumer choice exists and is still meaningful, but when you need a hotel you need a hotel. If Hyatt gets away with this abuse, every hotel will do it soon, and capital can just wait out any boycott.

> Without brand loyalty, hotels suffer.

Executive decision makers won't though. It's clear that consolidation in many sectors has gotten to the point that consumer power is an absolute joke and "ignore them, abuse them, and just defraud them" is a standard business model. Even if there's litigation.. this crap just overwhelms services so that basically the public pays twice. Witness the situation where various attorney generals have said that Facebook outsources customer support to the taxpayer when the attitude for handling everything is simply "don't like it? so sue us, good luck"

For anything smaller than Facebook though, it's hard to understand why brands/investors/business owners tolerate their decision makers encouraging wild abuse and short-term thinking like this, knowing that after brand loyalty is destroyed the Hyatt leadership will still get a bonus and fail upwards to another position at another company after claiming they helped to "modernize" a legacy brand. Is the thinking just that destroying everything is fine, because investors in the know will all exit before a crash and leave someone else holding the bag? With leadership and investors taking this attitude, I think it's natural that more and more workers get onboard with their own petty exploitation and whatever sabotage they can manage (hanging up on customers, quiet-quitting to defraud their bosses, etc). And that's how/why the social contract is just broken now at almost every level.

It's about what you define as "free standing" i.e. self supporting, which at least means no wire anchors, but arguably also means no pyramids because they are a sort of refined mound, no ziggurats, no cheating by excavating around mountains and putting a cap on it

> You also can't translate .. into Hellenistic Greek or some modern indigenous languages because there isn't enough shared context; you'd need to give humans speaking those languages a glossary for any of the translation to make sense

What? By substitution, this means you can translate it. As long as we're assuming a large enough basis of concept vectors of course it works.

> I'd say our current largest LLMs probably contain sufficient detail to explain a concept like a named race horse starting from QCD+gravity and ending up at cultural human events

What? I'm curious how you'd propose to move from gravity to culture. This is like TFAs assertion that the M+B game might be as expressive as 20 questions / universal. M or B is just (bad,sentient) or (good,object). Sure, entangling a few concepts is slightly more expressive than playing with a completely flattened world of 1/0 due to some increase in dimensionality. But trying to pinpoint anything like (neutral,concept) fails because the basis set isn't fundamentally large enough. Any explanation of how this could work will have to cheat, like when TFA smuggles new information in by communicating details about distance-from-basis. For example to really get to the word or concept of "neutral" from inferred good/bad dichotomy of bread/mussolini, you would have to answer "Hmmmmmm, closer to bread I guess" in one iteration and then "Umm.. closer to Mussolini I guess" when asked again, and then have the interrogator notice the uncertainty/hesitation/contradiction and then infer neutrality. This is just the simple case.. physics to culture seems much harder

This is mostly just a really high-level overview and the exciting stuff is only teased in the conclusion:

> (There is a nascent field of epistemic game theory, as well as some models of social media manipulation, but these fields are still in their infancy.) A more systematic study of such games would help provide a basic conceptual framework to understand these very real dynamics, and develop strategies to counter or mitigate them.

Time for a renaissance! Honestly game theory feels more practically relevant now than earlier with MAD, and it also seems obvious that the "rational actor" posited by classical behavioural economics is a pretty limited abstraction if you're interested in modeling the world. Besides politics/misinformation and wild stuff that happens in aggregate at the highest levels of "rational" economics policy.. it also feels like "management science" never really succeeded in actually saying much about the difference between healthy vs unhealthy bureaucracies, and the varieties and lifecycles of these kinds of systems. Plus epistemic/nonmonotonic logics capable of explicit belief modeling seems very well positioned for analyzing and architecting with AI systems, like checking theoretical properties of agentic interaction protocols, or answering what good mixtures of (credulousness for creativity) vs (skeptics for grounding beliefs) look like, etc.

Here's a really interesting thing, basically TLA+ style model-checking engine that supports agents, environments, protocols etc and explicitly takes into account epistemics: https://sail.doc.ic.ac.uk/software/mcmas/ Anyone else know of similar things? Software suites that are useful for game-theoretical analysis and modeling are kind of hard to find unless it's yet another toy for prisoners dillema.

Belief-and-knowledge stuff seems to be consulted and adopted in robotics/autonomous vehicles research sometimes, a place where wrong answers actually matter. But I sort of expect modeling/specs/invariants/determinism to continue to be kind of neglected almost everywhere else, because resolving ambiguity in advance is kind of threatening for groups that benefit from a zero-theory "just try it!" and "you're doing it wrong, buy more tokens and use this framework" kind of approach with AI and ML. Hope this changes.

Comprehension is a spectrum that starts somewhere superficial and merely "adequate" but also stretches into deeper literacy/fluency. On one end of the spectrum it is about reading between the lines, but that doesn't mean it's purely subjective nonsense. As for whether it's interesting or valuable, if you want to stay on the surface that's fine, but it's a narrow point of view to imagine that's all that is there.

Not sure if you've got an engineering/math brain with no taste for art, but I'll put it like this in case it helps. Who cares about the infinitude of primes, I mean it's just numbers and what could be interesting or valuable in that? If you're thinking squishy crap like literature and critical theory sort of sucks because you're craving something more hard and objective, maybe try to come at it from the point of view of semiotics[0], which is an adjacent topic, but also closely related to stuff like linguistics, formal semantics, cognitive science. Frege worked on this kind of stuff when he wasn't busy being a giant in mathematical logic [0] https://en.wikipedia.org/wiki/Semiotics

Thanks, probably better than TFA. The under-employment numbers are always telling, anyone who talks about employment without mentioning this is running a scam. Sobering to see that only elementary ed and nursing are really doing ok, fields where we're always chronically short. And even while everyone's talking about demographic bombs and aging populations, nurse-adjacent med-tech is still sitting at 57.9%.

First layer: Literally yes, there's fog everywhere. It gets around.

Second layer: Interesting contrast of something clean and natural meeting something industrial and dirty. Voices, who is speaking, where from, and with what perspective? Themes of liminality / phase-change / obscured visibility / motion. Those tiers of shipping mean that some other stuff besides fog gets around.

Third layer: Generalizing a bit, if natural things enter into a blackened, dirty hub of artificial industrial and commercial activity, they can become unclean.

Questions: Is man not also natural thing? Foreshadowing: What happens to the heart and soul of a man in an overcrowded, dirty, artificial setting? Can what was once clean and then dirty be made clean again? What does all the motion actually move towards? Where will the shipping go, and will the fog see the meadow again, and will man be able find his heart?

I don't know. Talking with strangers is always kind of insane, or at least a pretty illogical leap of faith, because there's not much chance that the interaction is interesting or amusing. Now a stranger that walks up to me and demands some dramatic reading of Paradise Lost, stat? Ok, I'm intrigued, hold my beer and buckle up while I let you hear about regions of sorrow and darkness visible because the produce section has never seen some shit like this

Python is probably too awesome and versatile to go out of fashion, but as a long time user.. the ecosystem is frustrating. Dependencies and packaging have been, and still are a nightmare after thousands of years. Problems are fixable sure but the thing is that they never end. After you get plenty of practice fixing all the related problems, you'll have to keep fixing them for yourself, your less savvy teammates, and in third-party code pretty much forever. This is worth it in exchange for "import antigravity" for the first 100 years, but it's frustrating eventually.

Everyone will mention uv/pyenv/poetry/conda/virtualenvs, so fine, let's pretend it's not a problem that you tried each of those in desperation and they are all household names. Suppose the packaging wars actually ended and everyone uses what you use without you needing to tell them, and suppose further that every pypa problem isn't blaming debian maintainers for obvious regressions. Pypi will still yank[0] packages at the source to perhaps randomly break deterministic behaviour, smashing anything in the blast radius rather than only clients using a --strict flag or something. You can pin your own dependencies but who knows what they will pin (nothing, or the wrong stuff, or the wrong way probably!) or what they will yank. Now for repeatability you need to host a mirror for everything you use- which is fine for corporate but a nonstarter for the novice and annoying for FOSS projects.

If you have zero dependencies or a slowly changing environment that is receiving constant care and feeding, you'll probably never notice how bad things are. If you put down most projects for a month though and pick them back up, perhaps with a different environment, machine, or slightly different python version it's broken, bitrotted, and needs serious attention to be rehabilitated.

People might argue.. that's just software dev. Nope. I say it with lots of love and gratitude for the efforts of the community.. but most langs/ecosystems would never tolerate this level of instability. One has to eventually just admit that working, portable, and actually repeatable environments with python basically just require docker. Can we talk about how "--break-system-packages" is hilarious? After you retreat to docker you can type this a few times a day, push the container up, pull it down months/years later, and then realize that literally the only way to get a stable working environment is to request a broken one. QED

[0]: https://docs.pypi.org/project-management/yanking/

Most platforms don't grow this feature because they can benefit from redirecting user energy into places that the platform is choosing. Or some vocal minority of the user base benefits from redirecting the platform to a place of their choosing.

Similar to nest usurpation with eusocial insects, this is by definition parasitism when the energy-redirection is unwanted or unavoidable.

In the specific case of AI it's way worse than the usual suspects where everyone is effected and so everyone has to have some opinion (looking at you politics). Because even some rant about how much you hate AI is directly feeding it at least 3 ways: first there's the raw data, then there's the free-QA aspect, then there's the free-advertisement aspect when others speak up to disagree with your rant. So yeah, even people who like some of the content sometimes quickly start to feel hijacked.

> But I still like my git config, my editor and other tooling that I have to assist in my tasks. And it's not worth it to replicate it over to other people's setup.

This isn't about stopping you from customizing your development environment. It's about making your custom development environment automatically useful on your desktop or your laptop, or committing a project-specific starter environment to version control so that it doesn't take everyone on your team days to onboard.

> And neither I want their things.

Why not? This isn't about something like git aliases or editor keyboard shortcuts. It's about tooling pipelines. If someone's claiming enhanced productivity I don't want to watch their 1 hour youtube video, pause it, and painstakingly recreate their setup to check if the claims are true.

There's a reason projects grow Tiltfiles, compose files, and similar. In the limit, agents are basically just a collection of microservices that just happen to operate on code. How do you expect to do that sort of thing without tools?

Well, the prompts used in testing ( https://github.com/t3dotgg/SnitchBench/tree/main/prompts) are pretty serious and basically about covering up public health disasters with lobbyists, so I'm not sure this is the kind of freedom you might want.

Still, the contacted_media field in the JSON is pretty funny, since I assume it's misfiring at a rate of several thousand of time daily. I can only imagine being on the receiving end of that at propublica and wapo. That bitch Katie was eyeballing Susie again at recess and she hates her so much? Straight to investigations@nytimes

All development is text files, that is missing the point. The development environment is a system, and a pretty complicated one too. It matters where the files are, what's in them, and how they interact. Things change together instead of staying isolated, you add more pieces over time, and even more things need to change together. Anyone who likes text-files more than click-to-configure UIs for tools, will probably like containers more than text files for systems, and for all the same reasons.

Your choices to reproduce complex systems are basically to 1) deny that complexity exists and accept any corresponding limitations in your working environment, 2) follow some error-prone multistep processes to reproduce existing system setup manually, 3) commit to centralizing all development on external cloud platforms, or 4) do something else to bundle together a setup that's repeatable.

I'm strongly in favor of (4) here, and while I'd probably disagree that it requires VMs instead of docker, the goal of repeatable processes is so important that hey, whatever works. It sounds like you're in camp 1 or 2?

> Is this level of fear typical or reasonable?

Anyone with more than one toolbox knows that fear isn't required. Containers are about more than security, including stuff like organization and portability.

> If so, why doesn’t Anthropic / AI code gen providers offer this type of service?

Well perhaps I'm too much the cynic, but I'm sure you can imagine why a lack of portability and reproducibility are things that are pretty good for vendors. A lack of transparency also puts the zealots for "100x!", and vendors, and many other people in a natural conspiracy together, and while it benefits them to drum up FOMO it makes everyone else burn time/money trying to figure out how much of the hype is real. People who are new to the industry get leverage by claiming all existing knowledge does not matter, workers who are experienced but looking to pivot into a new specialization in a tough job market benefit from making unverifiable claims, vendors make a quick buck while businesses buy-to-try and forget to cancel the contract, etc etc.

> Is it actually true that Claude cannot bust out of the container?

Escaping containers is something a lot of people in operations and security have spent a lot of time thinking about long before agents and AI. Container escape is possible and deadly serious, but not in this domain really, I mean all your banks and utility providers are probably using Kubernetes so compared to that who cares about maybe leaking source/destroying data on local dev machines or platforms trying to facilitate low-code apps? AI does change things slightly because people will run Ollama/MCP/IDEs on the host, and that's arguably some new surface area to worry about. Sharing sockets and files for inter-agent comms is going to be routine even if everyone says it's bad practice. But of course you could containerize those things too, add a queue, containerize unit-tests, etc

Yes please, more containers-first thinking for all things AI. The fact that this has been largely absent from most discussions and demos regarding agents and vibe-coding seems like a big red flag to me. I mean if we're going to YOLO the code, perhaps we should at least pay some attention to the dev / test harness?

It's worth thinking about reproducibility even if you're not concerned about security. We would not have to argue so much about the effectiveness of something like vibe-coding if everyone could simply share process/tooling without tons of custom setup. And even if you're not worried about filesystem sandboxing and leaking secrets like ssh keys, even if versioning with git is enough of an undo button for you.. who wants to clutter their whole system with a ton of new language stacks and/or weird dependencies for every small experiment? Most experiments with the new hotness are going to fail, so we should expect to do a lot..

Not OP, but this resonates, so I'll speak for myself. One common thread in a lot of PKD is that he takes tons of very esoteric philosophical, religious, and scientific ideas extremely seriously. This includes everything from Buddhist Maya, to Platos circular theory of time, and includes lots of really interesting Gnostic stuff too.

Reading his stuff at 10-15 years old can set you up with a follow-up reading list that lasts decades and reaches way outside of scifi. As a human being, there's some very real benefit to explicitly grappling with ideas like Cartesian solipsism at ~12 instead of hearing about it for the first time at ~20 in philosophy 101, but there's also some real dangers too! Besides being well read, PKD was obviously compelled to imagine the logical conclusion of almost everything he encountered in his life, including socio-political trends. While this inclination obviously bestowed an amazing gift, if you look at biographical elements of his life, it's equally obvious that it was a curse. There's definitely such a thing as being way too smart for your own good.

His prescience on many topics is well known, so I'll skip that and add something more obscure. I'm very pleased that PKD is gradually becoming recognized as way more than just some pulp / SF writer, or just another prolific pop-fiction dude (like say Dan Brown or Stephen King) with an existing audience that Hollywood can monetize. PKD deserves much credit in the high-brow postmodern canon too, and although it's not widely realized yet, probably has done more to pioneer and popularize truly experimental fiction generally, and metafiction in particular, than almost anyone else you could name.

Since I recently reread 3-stigmata, I will also quickly remind everyone that Choosy folks choose Chew-Z. As of today, it's uncanny how much Palmer Eldritch looks like an unholy combination of Zuck+Musk.

> simply proof that the code works right

Combining LLMs + formal methods/model checkers is a good idea, but it's far from simple because rolling the dice on some subsymbolic stochastic transpiler from your target programming language towards a modeling/proving language is pretty suspect. So suspect in fact that you'd probably want to prove some stuff about that process itself to have any confidence. And this is a whole emerging discipline actually.. see for example https://sail.doc.ic.ac.uk/software/

> This happens with any sufficiently big/old codebase. We can never remember everything, even if we wrote it ourselves

The important thing about a codebase wasn't ever really size or age, but whether it was a planned architecture or grown organically. The same is true post-LLM. Want to put AI in charge of tool-smithing inconsequential little widgets that are blocking you? Fine. Want to put AI in charge of deciding your overall approach and structure? Maybe fine. Worst of all is to put the AI in charge of the former, only to find later that you handed over architectural decisions at some point and without really intending to.

> A talking head rambling on for 60 minutes.

And probably a few minutes of commercials too. I get the impression this is an emerging generational thing, but unless it's a recorded university course or a very interesting and reputable person.. no thanks. What is weird is that the instinct to prefer video seems motivated by laziness, and laziness is actually an adaptive thing to deal with information overload.. yet this noble impulse is clearly self-defeating in this circumstance. Why wait and/or click-through ads for something that's low-density in the first place, you can't search, etc.

Especially now that you can transcript the video and quickly get AI to clean it up into a post, creating/linking a video potentially telegraphs stuff like: nothing much to say but a strong desire to be in the spotlight / narcissism / an acquisitiveness for clicks / engagement. Patiently enduring infinite ads while you're pursuing educational goals and assuming others are willing to, or assuming other people are paying for ad-free just because you do, all telegraphs a lack of respect for the audience, maybe also a lack of self-respect. Nothing against OP or this video in particular. More like a PSA about how this might come across to other people, because I can't be the only person that feels this way.

Trees are great, but ultimately a pretty ridiculous idea if the goal is to create shade, even if you're not worried about water consumption. Avoiding concrete walls or overhangs is smart because you don't want the thermal mass.. but of course you can build these things out of fabric or thin metal.

The funny thing is, if you build a wall or canopy to avoid the water consumption plus literally waiting a decade for a tree to get tall.. now you're probably in violation of your HOA height restrictions, etc. Desert cities need to basically drop the idea of conforming to the typical expectations of visitors and newcomers by trying to add greenery. It's better to add shade, dig underground, build wind-catchers[1], salsabils[2]. There's tons of basic things like making sure roof surfaces are more reflective, and more strategic architectural things[3] that can be done to improve things and the techniques have been used forever

[1] https://en.wikipedia.org/wiki/Windcatcher [2] https://en.wikipedia.org/wiki/Salsabil_(fountain) [3] https://en.wikipedia.org/wiki/Passive_cooling

Wait until you hear about the mighty Colorado. The Hoover dam wasn't just built to imprison Decepticons you know

Yes, because living in the desert is absurd and obviously unnatural. Setting aside of course 3,000 years of civilization in Egypt. While we're at it, let's also unpack all these skyscrapers in New York and Tokyo and make sure no one is living at any height greater than the monkeys in the trees.

It seems like you might be confusing "research programs" with things like "branding" and surface-level terminology. And probably missing the fact that society-of-mind is about architecture more than implementation, so it's pretty agnostic about implementation details.

Here, enjoy this thing clearly building on SoM and edited earlier this week: ideas https://github.com/camel-ai/camel/blob/master/camel/societie...

This is a typical critique that sort of assumes you have to be right about everything to be right about anything. Maybe a useful point of comparison is aether theories in physics. Wrong, sure, but useless? People might argue whether it was always pure distraction or a useful building block, but I wonder what Maxwell or Einstein would say. If nothing else, one needs something to ground one's thinking and to argue against, which is why replacement theories usually need to acknowledge or address what came before. And typically we try to fix bad theories rather than simply discarding them, especially if there's no alternative available. What are the other available "grand unifying theories" of consciousness? To quote the end of Aaronson's rebutal:

> In my opinion, the fact that Integrated Information Theory is wrong—demonstrably wrong, for reasons that go to its core—puts it in something like the top 2% of all mathematical theories of consciousness ever proposed. Almost all competing theories of consciousness, it seems to me, have been so vague, fluffy, and malleable that they can only aspire to wrongness.

Of course, it's not on Aaronson to rescue the theory he's trying to disprove, but notice that he is out to disprove it and spends his time on that, rather than imagining what axioms might be added or replaced, etc. Proving that having a large Φ-value is not a sufficient condition for consciousness hardly seems devastating "to the core", because finding better descriptions of necessary conditions would still represent significant progress.

Similarly a critique like

> He thinks that because in CPUs the physical transistors don't have as many connections as neurons in the brain, that it's fundamentally limited and cannot be conscious.

seems a little bit narrow. I do agree it seems to misunderstand universality, but on the other hand, maybe it's just distracted by counting IO pins on chips, and what it should focus on more is counting nodes/edges in neural net layers, and whether connection-counting in hardware-vs-software might need to have a weighting-coeffecients, etc. HN loves to celebrate things like the bitter lesson, the rise of LLMs and ML, and the failure of classical logic and rule-based reasoning and NLP. Is all of that same stuff not soft-evidence for the relevance if not the completeness of IIT?

IIT has always interested me, and after reading some of the detractors[1] I get that it has problems, but I still don't get the general lack of attention/interest or even awareness about it. It seems like a step in the right direction, establishing a viable middle ground somewhere between work in CS or neuroscience that measure and model but are far too reluctant to ever speculate or create a unifying theory, vs a more philosophical approach to theory of mind that always dives all the way into speculation.

[1] https://scottaaronson.blog/?p=1799

Interesting article about this: https://en.wikipedia.org/wiki/History_of_cannons

The oldest bell foundries do seem to happen at around the same time as the emergence of what we'd call a cannon, but before that there were indeed bamboo tubed contrivances and similar. If the Romans did have gunpowder earlier though, I bet they'd find a way.. I mean you don't hear about stone cannons and they'd be heavy AF but why not? These people basically invented fracking in BC, and literally tore down mountains to shake out all the gold. https://en.wikipedia.org/wiki/Las_M%C3%A9dulas

Yep. Tfa mentions elixir being below the suggestions to rewrite a web app in rust and golang (??). The conclusion here should be that the LLM cannot be trusted that much, but we know that’s not going to happen, so anyone who wants to remain relevant must start playing games and damn the consequences.

Now instead of improving the language in other ways there will be distracting constant pressure to appease the machine. Now call this added “efficiency” with AI, similar to how adtech calls hiring your own director of marketing and paying tons of people who actually work for google with money from your own company coffers “5:1 ROI on advertising investment“.

This is really cool but at first it was not that clear whether the project is serious or made in a recent decade.

Clicking around eventually took me here [1] with some cleaner prose, a publish date, etc although it’s more blog than reference. HN will probably like it since we love to hate the modern state of self hosting email

https://changelog.complete.org/archives/tag/nncp

Re: the YouTube thing, that’s an interesting point of comparison too. I can well imagine a younger version of myself looking at all “content production” stuff as quite silly and ephemeral, and of course much of it is.

The irony though is that lots of things more or less in this category still have a longer shelf life than software (or effort you put into technology related stuff in general). A 5 minute journal entry about that day still may serve some purpose even years later, but 5 hours/weeks/years spent on an obsolete platform or now irrelevant problem? Probably not. Even setting aside companies and professional work, bitrot gets really frustrating eventually after you realize that practically everything requires so much care and feeding. I don’t customize things like phones, browsers, or IDEs anymore because I fully expect most of the effort is pointless treadmill where most problems actively resist even semipermanent solutions.

Awareness of this kind of stuff helps some, which is why you see enlightened devs being pretty ruthless about pruning dependencies. Some tech ecosystems are obviously better than others too, but once you see the treadmill you never really unsee it