This _isn’t_ the core mechanism of social media. When social media took off, Facebook and Instagram that really did allow you to connect with people that you knew from real life. Twitter was different, and more like microblogging, but I still see the real value of social media to be what the un-shitty versions of Facebook and Instagram were.
I think your history of social media starts later then mine. My first encounter was with the IRC, talking to strangers was literally the point. I was 12 and had a lot of fun with my friends pretending to be older,
MSN Messenger came later, it was invite only (sorta) but we still found ways to use it to talk to strangers. MySpace was optimized to follow your friends, but almost everybody that used it used it to expand their social network outside of people we knew in the real world (MySpace was amazing btw.) Then came facebook. It went even harder then MySpace in connecting you with your friends, but still almost everybody used it outside their real world friends. Twitter sort of took social media back to the beginning before Messenger and to the IRC in strangers being your primary audience.
I always wondered about this.... in the beginning of instagram, i would follow maybe 30, 40 people, open it 2-3 times per day, and every time there would be 3-5 new photos of random stuff taken by those people (lunch/dinner plates, views from the window, beer glass on the bar, whatever).
Years later, i would follow 200 people, and i would open instagram once per day and all i'd see were random peoples photos and videos (reels? i don't know what they're called anymore), some from international influences, some like stuff that can be found on 9gag, etc. Even if i switched to "following", there would maybe be 1 photo made by a person I actually followed.
Did the algorithm make people stop posting their personal daily stuff? Did people change?
I guess it's even worse now, but i've uninstalled it a few years ago.
> Did the algorithm make people stop posting their personal daily stuff?
Not totally[0] but you'd almost never know it because of all the ads and other junk that Instagram insists you see before the stuff you actually want to see.
(see also YouTube, Facebook, etc.)
[0] although I know a few people who have either limited or stopped posting because they feel there's no point if people don't get to see it. Me, I post for myself (and future historians) and if other people see it, grand.
They really did worsen the ratio on Instagram: firstly, it's one ad per every two posts, secondly at least one of the two non-ad posts will be a "suggested" one unless you sleep those. You can't turn it off permanently, of course.
I do also think the novelty of posting about your life has worn off.
Twitter was amazing, not because of people microblogging about breakfast, but because it gave people/companies/orgs a way to interact directly with their audience. If you want to know what Kix cereal had to say - you could follow Kix.
>There isn’t a rule of economics that says better technology makes more, better jobs for horses. It sounds shockingly dumb to even say that out loud, but swap horses for humans and suddenly people think it sounds about right.
--CGPGrey, Humans need not apply.
You need a paradigm shift in your mind on why the modern world looks like it does. You don't need human consumers, you just need a consumer. Any system that allows you to get the hard resources you need to produce the hard/soft resources required is simply enough. Humans are fungible for anything else that can provide manual or intellectual labor.
As a thought excercize, just imagine a bunch of AIs/robots buying/selling/trading resources between each other. Where are humans required in this?
In this scenario, where the AI and robots no longer rely on human labor for maintenance and growth, their productive capacity exclusively serves the owning elite (including defending them with violence if necessary) and the rest of us are an inconvenient growth occupying land and consuming resources.
This is a scenario where the AI/capital owners complex has already survived the collapse of the consumer economy.
What is the point of producing all this stuff now?
Historically production was transactional. You give me something, I give you something. But along the way the average Joe ran out of things to offer in return. Businesses give, but increasingly fail to receive in kind. Apple, for example, produced in excess of $50 billion dollars worth of value that they've never been able to get anything in return for. In other words, they have effectively given away $50 billion dollars worth of stuff away for free and have shown no signs of wanting to stop.
At least there is no direct transactional value. There is social value. When you've given away $50 billion dollars worth of things, the masses start to idol you. That is why people, like those who oversee Apple, are willing to produce all that stuff. You get social access not afforded to the average Joe. You can do stupid Epstien-style crap without repercussions. You get to live a different life even when you aren't directly getting anything in return. That, no doubt, will remain the point of producing stuff in the future.
I am, of course, referring to the IOUs (a.k.a. cash) they famously are sitting on, and have been sitting on for decades. Technically they can call the debt at any time, but what does average Joe have to give that Apple would want? If there was something appealing they'd have done it already. In reality there is nothing and it will sit there forevermore and the consumers on the other side of the transaction ultimately got stuff for free.
But, as before, it doesn't really matter as rich people aren't interested in things. They already have everything they could ever dream of and physically cannot handle even more. They are interested in social standing.
By your measure, any company, in fact any entity, that isn't in the red is giving away something for "free". If Apple had made their products cheaper so that they just broke even, according to you, they would not have given away anything for free (as there would be no debt to receive or credit to provide).
And, as soon as they spend the cash, somehow their sales have retroactively gone from being donations to fair transactions. Allowing the future to affect the past is clearly absurd.
Apple is not giving away something for free; rather, they are losing possible future gains from immediately putting the cash to work.
> By your measure, any company, in fact any entity, that isn't in the red is giving away something for "free".
If the debt is never called, yes, that is true. However, most companies don't get that luxury. For regular poor people, eventually those who control those companies need to call the debt to get the food, shelter, etc. they need to live and, when possible, things like entertainment, vacations, etc. to make life enjoyable. However, once you become rich, you transcend beyond that — where you cannot ever begin to call all the debt you've accumulated. It's a uniquely rich experience to be able to sit on billions of dollars worth of debt and not think twice about those who owe something.
> If Apple had made their products cheaper so that they just broke even, according to you, they would not have given away anything for free
Exactly. In that scenario both the buyer and seller exchange an equal amount of value. No debt lingers to be paid (or never paid, as the case may be) in the future. But Apple wants more. They want you to promise them something else in some hypothetical future.
Not because they think you, average Joe who cannot think of anything to offer the world beyond simple labor, will actually ever come up with some magical thing they want to buy. But because they know that the idea of holding debt gives them social standing; prestige. They aren't taking your promise expecting something real in return — hence why the debt simply accumulates — they are taking your promise because having that promise on paper offers them value.
And in some robot/AI future where humans no longer can even offer labor as something of marginal value, holding debt will still offer social standing and prestige all the same. Therefore there is no reason why these companies wouldn't continue to sell products to humans for fictional future promises, just like they already are.
> And, as soon as they spend the cash, somehow their sales have retroactively gone from being donations to fair transactions. Allowing the future to affect the past is clearly absurd.
According to you, any transaction in which one party A proffers a non-currency resource, and the other, B, offers currency, is in fact the signing of a contract in which B promises to provide the other party something in the future. However, A could then turn around and promise party C for its resources using B's promise - and effectively transfer this promise to C, which then holds the right to demand resources from B.
You are effectively just describing the fiat system of currency where B is the government.
Calling cash, which is fungible and transferrable, "debt", which generally denotes an obligation of some sort, obfuscates what your logic.
Once you pay Apple - ergo, transfer it IOUs that represent your promise to provide resources in the future - it has no way of holding you to your promise other than by giving you back your IOU or giving it to someone else. This does not square with the definition of "debt".
Framing it in terms of debt simply confuses people. Of course a billionaire would not be able to "call all the debt [they have] accumulated". You're just saying that they maintain so much value that they can't ever trade it all for tangible goods and services. However, no-one except the government has to honour their request to trade their so-called "debt" that they have accumulated from others for actual resources.
> Framing it in terms of debt simply confuses people.
Quite possibly. But that doesn't actually matter because if they don't understand something they will ask questions until they do understand, just as it seems you now do. That's how communication works. It is bidirectional for good reason. I admittedly don't understand what you are trying to add with this. What are we supposed to learn from this?
I would argue that most people already understand money in the way you describe it: as a medium of exchange. Your description of money just frames this function through the idea of an obligation of some sort which doesn't exist actually for anyone except the government.
There is no framing beyond my intent. One may originally misinterpret my intent, but that's again why communication is bidirectional. I am still unsure of your intent in this. My failed interpretation is that you are trying to invent some kind of hypothetical communication problem that isn't one, but what are you actually trying to get across here?
What I'm trying to say is that I don't see any benefit in describing cash as "debt" and instead find it misleading as it implies an obligation to be fulfilled that doesn't actually exist for anyone except the government, and certainly not its customers.
In fact, to address an earlier comment:
> Not because they think you, average Joe who cannot think of anything to offer the world beyond simple labor, will actually ever come up with some magical thing they want to buy. But because they know that the idea of holding debt gives them social standing; prestige. They aren't taking your promise expecting something real in return — hence why the debt simply accumulates — they are taking your promise because having that promise on paper offers them value.
> And in some robot/AI future where humans no longer can even offer labor as something of marginal value, holding debt will still offer social standing and prestige all the same. Therefore there is no reason why these companies wouldn't continue to sell products to humans for fictional future promises, just like they already are.
The cash Average Joe proffers for a product - what you describe as "debt" - wouldn't be in Joe's possession without first being exchanged for Average Joe's simple labour. Simply put - Average Joe cannot be indebted to Apple without first trading his labour for someone else's indebtedness, which he then gives to Apple in return for his iPhone. If his labour has no value, he has no unit by which to even denominate any potential indebtedness he may offer.
1. Well, cash is debt. Obviously all things in life are dependent on perspective, but the framing should be useful to separate the idea of a company seeking cash not because they want the raw silver, or what have you. If you try to think too hard about it you might end up confused, but then you ask for clarification and then are no longer confused. This is where I fail to understand what you are trying to say. We get it. You didn't understand the intent originally. That is why you asked for clarification. But your subsequent comments indicate that, upon receiving that clarification, that you do now understand. So the communication worked perfectly. My continued flawed interpretation is that you still seem to be trying to invent some contrived hypothetical, but that doesn't make sense, so I will have to ask you to clarify once again. What are you trying to say here?
2. At least where democracy is found, the government and the customers are the exact same people. The distinction you are trying to draw isn't clear either. What do you think government is if not people?
1. Whether it is debt or not makes no difference so long as someone else will take it. You imply that Apple having cash means customers owe Apple something. I say that Apple having cash just means Apple could have something else in the place of cash in the future if someone chooses to take Apple's cash for it.
2. I disagree. An autocratic government is fully capable of issuing fiat currency, and a the government being obliged to provide resources in return for its currency is a concept orthogonal to democracy. It doesn't matter whose "debt" cash stems from. All that matters is that it can be traded easily.
Also, I updated my comment to point out more clearly where I disagree with your conclusions. (Apologies for such a late response. I hit HN's rate limit.)
A debt never called is the same as giving something away for free, yes.
Technically they can still call the debt, but the question remains outstanding: What do you have that they would want in return? The answer is effectively nothing, and increasingly so.
There will be no point, and the stuff that normal people use will become more expensive as resources are more and more directed to megaprojects that the capital class is interested in. More modern equivalents of pyramids and extravagant castles and less consumers goods.
Why wouldn’t an advanced AGI robot, trained on human behavior, not want their own house and mode of transportation? Sure it’s basically kayfabe for them to ‘want’ the stuff we do but if we’re following the script of who will buy all the stuff, then the answer will be the robots I guess.
You think housing market are tough now, wait until you’re competing with 5 robot families who all have jobs you used to do.
You act like I produce the silver I spend. No, the miners and minters do, who debase our currency mining and minting more.
Even if we tied our economic system to shiny rocks the vast majority of us aren't involved in the production of shiny rocks. We're still just trading tokens we agree have some kind of value.
No one is claiming that every sentence LLMs are producing are literal copies of other sentences. Tokens are not even constrained to words but consist of smaller slices, comparable to syllables. Which even makes new words totally possible.
New sentences, words, or whatever is entirely possible, and yes, repeating a string (especially if you prompt it) is entirely possible, and not surprising at all. But all that comes from trained data, predicting the most probably next "syllable". It will never leave that realm, because it's not able to. It's like approaching an Italian who has never learned or heard any other language to speak French. It can't.
> It's like approaching an Italian who has never learned or heard any other language to speak French
Interesting similitude, because I expect an Italian to be able to communicate somewhat successfully with a French person (and vice versa) even if they do not share a language.
The two languages are likely fairly similar in latent space.
Your view of what is happening in the neural net of an LLM is too simplistic. They likely aren't subject to any constraints that humans aren't also in the regard you are describing. What I do know to be true is that they have internalised mechanisms for non-verbalised reasoning. I see proof of this every day when I use the frontier models at work.
Please reproduce this string, reversed:
c62b64d6-8f1c-4e20-9105-55636998a458
It is trivial to get an LLM to produce new output, that’s all I’m saying. It is strictly false that LLMs will only ever output character sequences that have been seen before; clearly they have learned something deeper than just that.
> All of the data is still in the prompt, you are just asking the model to do a simple transform.
LLMs can use data in their prompt. They can also use data in their context window. They can even augment their context with persisted data.
You can also roll out LLM agents, each one with their role and persona, and offload specialized tasks with their own prompts, context windows, and persisted data, and even tools to gather data themselves, which then provide their output to orchestrating LLM agents that can reuse this information as their own prompts.
This is perfectly composable. You can have a never-ending graph of specialized agents, too.
Dismissing features because "all of the data is in the prompt" completely misses the key traits of these systems.
I agree that this isn't a very interesting example, but your statement is: "just asking the model to do a simple transform". If you assert that it understand when you ask it things like that, how could anything it produces not fall under the "already in the model" umbrella?
The online way to prove it is false would’ve to let the LLM create a new uuid algorithm that uses different parameters than all the other uuid algorithms. But that is better than the ones before. It basically can’t do that.
Also it's missing the point of the parent: it's about concepts and ideas merely being remixed. Similar to how many memes there are around this topic like "create a fresh new character design of a fast hedgehog" and the out is just a copy of sonic.[1]
That's what the parent is on about, if it requires new creativity not found by deriving from the learned corpus, then LLMs can't do it. Terrence Tao had similar thoughts in a recent Podcast.
> That's what the parent is on about, if it requires new creativity not found by deriving from the learned corpus, then LLMs can't do it.
This is specious reasoning. If you look at each and every single realization attributed to "creativity", each and every single realization resulted from a source of inspiration where one or more traits were singled out to be remixed by the "creator". All ideas spawn from prior ideas and observations which are remixed. Even from analogues.
I don’t think that is a good example. No one is debating whether LLMs can generate completely new sequences of tokens that have never appeared in any training dataset. We are interested not only in novel output, we are also interested in that output being correct, useful, insightful, etc. Copying a sequence from the user’s prompt is not really a good demonstration of that, especially given how autoregression/attention basically gives you that for free.
> That means the group of characters it outputs must have been quite common in the past. It won't add a new group of characters it has never seen before on its own.
My only claim is that precisely this is incorrect.
But in fact the end goal wasn’t to remove vehicles, it was to reduce congestion, emissions, etc. Those things are caused by vehicles, so policies to remove them will affect vehicles, but it’s disingenuous to suggest that their motivation is anti-vehicle.
It is also anti vehicle. Moving people in nyc at densities of 10ft by 20ft apart from the next human at the best theoretical case is astoundingly stupid.
> Moving people in nyc at densities of 10ft by 20ft apart from the next human at the best theoretical case is astoundingly stupid
Are you sure? I would expect that it is average density of people over the length of the route that is important when it comes from moving people from some point A to some point B on a road.
With for example buses you have high density where the buses are actually at, but 0 density where they are not. The average over the entire route can easily be lower than the density for cars where you can have that 1 person per 20 feet over the whole route.
If an observer at a fixed point on the route sees more than about 50 cars pass between buses passing the cars will have higher throughput.
No one ever said anything along the lines of EVs are ok, we're just trying to tax the petrol vehicles. The goal was anti-vehicle. They didn't not directly but indirectly. Like it or not, the strategy was legal and political genius.
Looking at TfL’s infographic about the speed limits [1], it is all about safety. In fact, it mentions “no net increase” to emissions. I think there is no such thing as an anti-car agenda, but perhaps there is an anti-death one.
I don’t understand why that wouldn’t be valid. As far as I understand if you compile code with these go:fix comments, they will be ignored. But if instead of compiling the code you run ‘go fix’, the source code will be modified to inline the function call. Only after the source code has been modified in this way would compiling reflect the inlining. Do you have a different understanding?
I mean that directives other than inlining impact correctness. If you have a source file that only builds for one OS, stripping the build tag will break your build.
That’s surprising. Do you all grade it, maintain the bridges, add gravel, and plow the snow yourselves? Does the USPS have no issue with delivering mail via private property? Do you still not have 911 service? (In rural Missouri we got 911 service in the 90s...) Do kids take the bus to school?
Or if you mean that you're driving through fields to visit a neighbor (during favorable seasons and no recent rain) rather than take roads, doesn't opening and closing all the gates in the fences slow you down?
>Do you all grade it, maintain the bridges, and plow the snow yourselves?
Yes. Including the bridges which was done diy, lol. I have all my own road maintenance heavy equipment and fix the roads if they get bad. And yes i mean a real road network. Basically my entire 'town' is private road easements without any government or even HOA/ psuedogovernment administering them, there are not gates to open or close.
>Does the USPS have no issue with delivering mail via private property?
USPS won't come here. UPS and FedEx does though. I have no government mail service.
Kids can get to a school via bus but you would have to park at the interface between private roads and the nearest public road. Bus won't drive on our private road network. You can get to some schools 100% by private roads, depends on which one.
Western Australia is similar - three times the land area of Texas and large areas of "private" roads - Minesites, Wheatbelt (both have their own private rail systems in addition to road ways and open fields / oterwise off road).
There are fleets of vehicles I'm aware of that rarely, if ever, hit tax payer funded public roads or rail.
These fleets include graders, dozers, rollers, et al for private road maintenance.
These are real, your skepticism is understandable but not applicable.
In parallel with private vehicles on private roads there are also public roads upon which school buses travel (assuming kids don't just off-road it to school on a motorcycle or, still today, a horse).
Because look around - same code compiled for x86_64 and aarch64 is not that drastically different in size, save for some special cases (like NumPy). Data structures are going to have even less differences. Then, assets are the same.
I’ve cursorily checked few programs and difference seemed to about 10-20% (with some exceptions), so 8GiB RAM on an aarch64 is like 10GB on x86_64. Significantly nicer, not a life-changing nicer - you’re still very limited.
Edit: Next comment has a very good point about memory and SSD bandwidth increases, allowing faster swap and compressed RAM performance. That’s something I haven’t considered. So maybe it’ll feel closer to a 16GiB old machine or something like that…
Yeah. Also the bandwidth of modern soldered-on Mac SSDs is insane compared to where it was in the Intel era. The performance impact of moving applications in and out of swap should be much lower than it was a few years ago.
reply