advael

Yea, we seem to agree that people can't actually push the needle on these kinds of issues in a meaningful way through their individual choices as consumers. That's a dead end, and it's one industries that fear getting regulated have settled on pushing at every opportunity, exactly because it's a dead end

Ultimately, the only thing powerful enough to take on a cartel is a government. A lot of people have negative feelings about the abstract concept of government, and certainly the government sucks sometimes. Nearly any way that an individual can come up against power has the potential to suck a lot

However, democratic governments for all their failings are far more responsive than consolidated industries to pressure from people who want change. Our "brand of capitalism" - by which I mean the particular economic and regulatory policies of the US government at this moment in time - is doing significant harm to the majority of people, in a way that it just wasn't in my parents' generation, and it doesn't have to stay this way. The options are not accept how things are or tilt at windmills by doing a master's thesis worth of research for every item we buy at the grocery store. The political will for better safety nets, better research and regulation, economic policy that helps the majority of people instead of a tiny sliver of the most deep-pocketed, and siccing the government on the cartels that increasingly dominate every industry is pervasive. The bet cartels are making is that they can create apathy and confusion to forestall this until they can burn down the institution of democracy itself

Also notable is that in the cases you've presented, the harm was not to the lives of poor addicts, but to the fortunes of rich investors, which seems a lot more like the kind of thing the courts will hear out

This is a common reaction to a pervasive phenomenon

Reading is one way people encounter ideas and learn, and so some people who read have a broad worldview and have absorbed the insights from books

People start to notice this, or perhaps those people attribute some other success to reading books. Enough association of this kind makes reading books high-status

Many people want to short-circuit this process to get status, so they optimize for reading a lot of books, or sometimes fake reading books

People who fake reading books for status are shown ridiculous when exposed

People who are allergic to and resent status nonsense conclude that bookshelves mean someone's a poser

This is obviously ridiculous. Many people have bookshelves because they simply need to store some books

The error committed here is, to me, a common one, overfocusing on the most salient case. I agree that it's icky when people do shit like fill shelves with books they've never and will never read as a status symbol. But are we really to assume that this is the most common reason to have a bookshelf? Or is it just the reason to which you have the strongest reaction, and thus stands out the most to you as a hypothesis?

With some additional information about a person, one might be able to conclude this given the bookshelf, but it hardly seems like enough evidence to come to this conclusion by itself

Yes, nearly impossible as I said before. The "nearly" qualifier acknowledges that this does sometimes nonetheless happen

It's quite strange to me that business leaders love talking about how they are being rewarded for taking risks when it's used to justify disproportionate executive pay and stock buybacks for investors, but in all discussions of economic and even criminal regulation surrounding businesses, a business being made to accept some liability for putting others at risk is considered too risky for business. At present, large swaths of the business world are heavily subsidized by the government, can write off operating losses on their taxes, are often funded by hedge funds and investors who don't seem to mind them being deeply unprofitable for many years, and hold things like the IP rights produced by any worker they've employed as a treasure trove of potential value to sell off for the benefit of the owners should the business fail. On top of that, we must also believe that it's good and right that it is nearly impossible to prosecute any actual decision-maker within a company for criminal conduct done or harm caused by that company, or else business is just too risky and the economy might collapse?

To me this doesn't even warrant consideration, it's ridiculous on its face and perverts the notion of rule of law

I honestly think that if half of the FTC's lawsuits under Lina Khan succeed, she'll have more meaningful positive change in the life of the average American to point to than the last six or so presidents

I think half of this is well-reasoned and insightful but it commits a couple of enormous confounds when trying to explain the general sense of stagnation

One is the economic side of things. Like me, this author has a very US-centric view of things, and to be fair, much of the world often seems to. However, a crucial factor in the US economy now is inequality, concentration of power at the high end, and purchasing power and quality of life essentially rotting at the low end. The vast majority of people in the US do not give a single flying fuck about the GDP, and they are correct not to, it's become a useless metric for their purposes. Whatever the economists say about broad aggregates, a lot of people are struggling right now. The job market is chaotic and, for most jobs, less lucrative in real purchasing power. Even relatively educated younger people probably can't afford a house. Life expectancy has declined for the first time since we've been rigorously measuring it. Companies can just permanently and unilaterally raise the price of groceries or TV or crucial medicines because they operate as cartels. Your precarious job also provides your flimsy shield against the absolute nightmare that is ever dealing with the medical system. Even in the 2000s the average person could get bankrupted from a single medical emergency. But now that's still true except that's on top of more baseline precarity. A lot of this got muted by people being mezmerized by the density of the network increasing. Go to any living busy city and you will notice that happy people are not on their phones 24/7, but so many people are. You never meet those people IRL because of course you don't. I've met them. I seem a lot nicer in person. Some of the perma-online aren't as mean in person, but they also can't hold a conversation very well. Their life is online because that's what they can safely afford and do with their time. And as concentration of power has tried to squeeze more blood from a stone, as those goods and services we who touch grass also saw get shittier and more invasive and annoying, this further immiserated people for whom they were and still are more important

Also, destabilization of the world is real. There natural disasters at unprecedented scale every year. There are huge migrant crises that come from people fleeing natural disasters, wars that indirectly result as resources become more scarce, pogroms that follow the wars, and this in turn leads to populist nationalism that stokes bigotry and has significantly destabilized the actual governments of several countries

A lot of misery and paralysis are amplified by network effects as described in this paper. I'd even believe network density accounts for some of it on its own. But also, there is actual misery to be amplified at a pretty catastrophic scale right now for reasons that are not just because the internet. And moving to dense social medias concentrated by enormous corporate information-brokers is not some accident of history, those companies still make decisions that both maintain and exploit those network properties

A good mathematical shiny object is alluring because it's fun to reason about and math seems so powerful and it gets to have unambiguous truths that you can program about and so I think a lot of people find a mathematical model that looks explanatory and it dominates the picture for them. Maybe people are so upset because information propagates so fast and it's all a blur. And like, again I partially buy that. I've felt that. Also, I think some people dark forest because sometimes internet drama escalates and finds their house and sends swat teams there. Maybe we could sparsify our networks, but we are still cranking lots of voltage through lots of nodes and many of the connections are not just missing but badly frayed and kinda sparking out and flopping about on the street

On one hand, I strongly agree that some mitigations are personally achievable and that this is smart to do. Personally, I preferentially buy natural fibers (cheaper than you might think if you thrift 'em), and I cold wash and hang dry the polyester garments I do own

However, I am also wary of the push to view mitigations or "consumer behavior" generally as a means to actually deal with this kind of problem at scale. Like the whole carbon footprint thing BP was pushing for a while. Still in the public lexicon, doesn't do jack shit about real emissions at scale, but great for gotchas on activists. We have significant poison control and food quality control regulations because people can't really develop all the expertise necessary to vet all their own stuff, and it's a huge drain on society to make them do so rather than just solve that coordination problem centrally. Also, monopolistic consolidation has made even very organized boycotts dubiously feasible in forcing change in companies who respond more to their investors than their customers.

To me, governments exist to solve exactly this kind of problem, even though there are many other areas in which I would like them to mind their own business. In this case a lot of why petrochemicals are so cheap is because the cost of doing business for oil companies is heavily subsidized, so removing those subsidies wouldn't even require a new tax, and that money could be productively spent on lots of stuff, like maybe medical research even

I mean I've worked for multiple companies wherein I've been explicitly hired as R&D staff and then let go because they don't want that anymore and would rather downsize than pay another dev to do the short-term work they are now focusing on, and also for companies where the workload wasn't too heavy and left me time to do other stuff but they ran out of runway without telling anyone. I've also worked for a company where I did three people's jobs that weren't mine for three months and got commended for saving the company (not my words) but then unceremoniously laid off because promises were made during that time (that I didn't ask for) and then spreadsheets were consulted and said promises started looking expensive

During none of these instances was I given any warning or offered a transfer or pay cut instead

Basically, the last few years dealing with this industry have given me some trust issues

The most ridiculous aspects for me were the anthropomorphizing (Reminds me of that one Sam Altman interview a bit) and the use of "infinite", which both doesn't really work on vibes (as many have noted, while I'm sure chatGPT has been exposed to every word, its pattern of communication is very "regression to the mean" among them), but also is silly if taken literally, because unless we're counting like some quirky technically-grammatical combinatoric compounding that we in practice infer the meaning of from composition of what we identify as separate individual words (like just hyphenating a bunch of adjectives and a noun or something) there's not really an argument for there being "infinite vocabulary" in the same sense that there is for "infinite possible sentences" because being a valid word requires at least that someone can meaningfully comprehend what is meant by it, and coordination requirements of this nature tend to truncate infinities

The case for ChatGPT doing significant coinage that sticks isn't particularly strong either, partially from theory and partially because I'd think I'd've heard a lot of complaints about it by now, and the ones on hackernews would be repetitive to the point of seeming unavoidable (we agree on that for sure)

Anyway, re: the silliest hype I've heard all week, I'm mostly just trying to find humor in what has been a pretty bad hype wave for someone who's pathologically bad at sounding like the kind of nontechnical hype guys that pervade any tech hype wave but is nonetheless mostly seeking out jobs in this field because it's what my expertise is in. Incredibly awful job market for a lot of people I realize, but it feels like a special hell I get for getting into ML research before it was (quite so) cool. I'm trying to fight the negativity but I've gotten screwed over a lot lately, but I don't have anything against you personally for being silly on hn

I've been messing with contrastive learning in the CLIP-like "minibatch" style for applications where correspondences between modalities within an example are a relevant source of information, and have been pleasantly surprised by how much this can actually in some cases speed up or otherwise aid with learning different objectives framed as decoding each modality's embedding for a different downstream objective. It seems that tying the representations together in this way can serve as a kind of semi-orthogonal correcting force on the embedding's ability to retain its original reconstructability while not constraining it too much to shift the representation for other goals, and many properties of the resulting models (like the obvious one of having interchangeable downstream decoding functionalities regardless of which modality is used to encode) suggest to me that it might be another tool to use when pursuing a better theory of objective balancing, as well as significant practical use in applications obviously. I'm surprised people aren't making more random contrastive embeddings for sets of "non-obvious" coupled representations

I really gotta find a job where they pay you but people won't freak out if you can't make a simple business case for every detail of what you're doing in a given week. Does that even exist anywhere without like a PhD from rich people schools? Aside from the moral issues this myopic mindset of businesses just refusing to trust or support their people in any sense sure makes it hard to believe they're actually capable of "innovating" anything. Necessity may be the mother of invention, but focusing on speedrunning exploiting will always leave exploring neglected

On a meta level I agree that having this kind of dataset with "before and after" would be pretty interesting. On an object level I do not predict that this would increase the overall diversity of language usage - and in fact it would be extremely surprising if this was even possible due to some general mathematical properties of neural networks - nor would "more professional writing," though I do agree with this characterization of the way AI-generated text sounds. The more I work with LLMs and encounter them in the wild, the greater my confidence that I can tell when something was generated, with the exception of B2B marketing copy and communications from HR departments or state agencies

On the level of meta-discourse you seem to want to also speak to: Dang even when people have the Official Corporate Approved Perspective (in particular, the claim that it's "like being able to chat with someone that has an infinite vocabulary" is probably the silliest delusional AI hype I've heard all week) and the most upvotes in the thread they still think they're an embattled ideological minority. Starting to think that literally zero people in the modern world don't have or affect a victim complex of some kind

Massively subsidizing anything to do with oil has certainly caused all these byproducts to be super artificially cheap, so with the governments with globe-spanning military empires absorbing most of the cost and ensuring not even the mildest threat of accountability for "externalities" in an economy designed to reward single-minded monetary optimization, it was pretty inevitable that things made of oil would get pervasive everywhere

Maybe we should convince government to do less of those things somehow?

I think there's an issue of burden of proof at play. If these microplastics were a drug, they would need to go through an expensive and onerous vetting process that consists of both safety testing and efficacy testing for a particular indication for it to be legal to sell someone if and only if a doctor specifically prescribed it for that person. Even if you believe, as I do, that efficacy testing shouldn't stop people from voluntarily accessing potential cures, safety testing doesn't strike me as unreasonable

These microplastics are crossing the blood-brain barrier in a significant portion of the population without most people even intentionally ingesting them. Hell, it's hard to avoid if you try. Our society slathers this stuff over everything. In my lifetime I've seen a giant swath of packaging for food go to plastic from some other material, presumably to cut costs. I recently bought a new oven and I guess they put those stupid plastic phone screen covers over the glass panes on both sides of the oven door. Do they just ship all glass with this stuff applied to it now?

The burden of proof to ensure these things are safe should be on the companies selling them. The companies that make these substances already have vast research arms for designing the polymers, and if they were held to even the FDA's standards for drug safety we would have a lot more studies about this stuff. These companies are also heavily subsidized by governments, which might even be part of why downward pressure on price seems to make everyone go to plastics. Like why exactly is it cheaper, even if you don't bake in the entire R&D budget of inventing them, to use novel polymers made as byproducts of petroleum than it is to make glass or cardboard packaging?

We should, in the same way that we do for pharmaceuticals, hold these orgs to a standard of demonstrating that this bioaccumulation doesn't cause some kind of harm, because the alternative is that some researcher needs to come up with a specific claim of harm, fund a study despite there being significant pressure from these orgs to never do such studies, and certainly never publish if the results might credibly make them look bad, and somehow translate that into making bespoke positive policy prescriptions that would likely be fought tooth and nail by same. That expectation is not a sane default

In the last century the landscape of human health problems has pretty drastically shifted (more cancers, more obesity, more asthma, perhaps more mental illness, lower rates of maiming, less malaria and consumption deaths, and frankly this is all just deaths, which is not the only kind of health outcome we care about) so the fact that we can't suss out all the cause and effect from every factor and every outcome is neither surprising nor an argument against it being alarming any time a novel foreign substance can penetrate cell membranes and readily bioaccumulates

Like we study mechanisms for a reason. "Little things get into all your cells" isn't a specific causal argument for specific health outcomes, but it's reasonable to assume it might do something undesirable based on what we know about how cells work and other cases like it. Also, the intervention space is pretty large and some scaling back on plastic usage from the current levels would not be some crazy obvious harm either

Also, "we haven't found effects yet" is especially troublesome in a case where the contaminant in question is a cash cow for an industry that has continuously spent more money than I can even meaningfully reason about suppressing science and muddying discourse about products they sell. Sowing doubt has been an explicit strategy of the petroleum industry for greenhouse effects and it took a long time for us to even definitively know that, so now that we do, why should I assume the same companies aren't taking the same approach for this problem? Again, it feels like isolated demands for rigor

I am fairly convinced that this comment section is among the best demonstrations for "isolated demands for rigor" I have personally ever seen, especially paired with any comment section about an AI-related article

When we are trying to predict the implications of unproven technologies on complex worldwide economies, there is talk of "obvious inevitability"

When we express concerns that chemicals humanity recently started coating the entire planet with at an alarming rate seem to get really deep into every kind of living tissue and nearly everything else we've ever seen bioaccumulate like that has caused a lot of unforeseen issues that took a long time to suss out but most turn out to be at least somewhat harmful, there's all this "well we don't have longitudinal RCTs with enormous sample sizes showing the specific harms of every specific plastic published in prestigious journals yet so who knows really"

In fact I used the key phrase "sunk cost" directly to imply that, since that's entered the lexicon as a common reasoning error

While business leaders want to think of and portray themselves as these hyper-rational actors whose every choice is either made from ingenuity or total necessity, this is obviously false and I think the prestige and optics of office spaces heavily play into the priorities of managers, especially as compared to people who do any other kind of work

I think the most compelling explanation is that control over workers is more important than productivity to the leadership of many of these companies

Perhaps especially in the case of larger companies, sunk costs on office space may be a factor

A lot of tech industry mythology is written from the perspective of a technical lead as imagined by an executive, or from the perspective of an executive who views themself as primarily a technical lead. In practice, this mythology is usually exactly just that

Or you know, sometimes your boss just lies to you, you do the crunch time, then when you save their asses you get laid off because that's cheaper than the raises or bonuses and you're prolly burnt out anyway so you won't be that productive again for months

At will employment, baby

Much appreciated. Super interesting stuff

It's almost too on the nose to do this level of silly mythologizing through a half-baked analogy (a favorite rhetorical handwave of the so-called "founder cult") of Mark Zuckerberg of all people in this particular thread. Excellent poe if true

Business culture's tribal knowledge about technology seems to have by and large devolved into wanting only bossware, that all gizmos should function as a means of controlling their suppliers, their labor force, or their customers. I think this is a good sign that the current economic incentives aren't functioning in a desirable way for most people and this needs significant intervention

It's crazy to me that we managed to get such an exciting technology both theoretically and as a practical tool and still managed to make it into a bubbly hype wave because business people want it to be an automation technology, which is just a poor fit for what they actually do

It's kind of cool that we can make mathematical arguments for this, but the idea that generative models can function as universal automation is a fiction mostly being pushed by non-technical business and finance people, and it's a good demonstration of how we've let such people drive the priorities of technological development and adoption for far too long

A common argument I see folks make is that humans are fallible too. Yes, no shit. No automation even close to as fallible as a human at its task could function as an automation. When we automate, we remove human accountability and human versatility from the equation entirely, and can scale the error accumulation far beyond human capability. Thus, an automation that actually works needs drastically superhuman reliability, which is why functioning automations are usually narrow-domain machines

Yes, like I said before, this hasn't de-stratified the job market. Jobs that want to select applicants by class can find plenty of ways to do so easily

E.G.: More expensive vocational programs that haven't been subsidized, selecting among universities for especially "prestigious" (read "class-signaling") ones, baking cultural assumptions of the upper classes into the expectations surrounding "professionalism" in the interview process, etc

Socialism is a term too broad to mean anything. When we say socialism, do we include states that provide healthcare systems as infrastructure, as is common across the world? Do we include the vast amounts of market interference represented by decisions about what crimes can be hidden behind a corporate veil, what companies win lucrative government contracts to have decades of non-competitive profits?

The attitude that the government shouldn't intervene against companies on the behalf of human rights has been tried for fifty years, and it is an unprecedented failure even in financial terms for at least roughly 80% of the population of one of the largest and wealthiest countries in the world in terms of real purchasing power. Even for many of us in higher income brackets, the resulting crumbling infrastructure and drastic wealth disparities leave much to be desired as a society to live in. Many of these problems have solutions, and calling them "socialism" is meaningless as an argument against them

It isn't set up to now, but it has been before and it can be again

Yea, and jobs that selected by university degree wanted to implicitly select for class

Making the path to go to a university more accessible is admirable. Entrenching this hiring practice with policy designed to enforce its implications on universities was always ill-conceived and the consequences have mostly been negative, creating a class of permanent debtors, turning universities into dysfunctional and corrupt quasi-businesses, and not really causing a significant de-stratification of job opportunities on balance at all

With all due respect, fuck that. We do disruptive stuff with unintended consequences for millions of people as a matter of policy all the time, and the idea that it'll disrupt some finance goons' ponzi scheme does not bother me

I think it's really sad that we so reflexively consider universities vocational training that criticism of universities so often includes "offering degrees that won't get you a job"

Actually academia predates the push to gate jobs behind undergraduate degrees, and trying to repurpose these institutions that mainly exist to train and employ researchers to be fully general vocational schools has been a disaster in every respect for everyone but the parasitic class of administrators it's spawned