Predicting OpenAI's ad strategy
ossa-ma.github.io
ossa-ma.github.io
Increasingly it seems you must go to the almighty Google or Meta in order to launch any business.
We're looking to expand into a new business line and have out grown our pharmacy capacity.
The new business line will cost about $2M in software dev, and $3M for the new facility. The advertising budget? $40,000,000 (annual).
We can build 10 robotic pharmacies (~10 staff per 4000 fills daily, each) for the price of just the advertising.
Increasingly we wonder why America doesn't build more and here is why. You are going to give all your revenue to two platforms. Unless you operate in a business line with 50% margin you are screwed.
I don't know what the solution is, but its clear that the platforms are figuring out how much margin everyone has and slowly eroding it. Somewhere between 8-15% of the cost of all products we purchase is advertising spend.
https://en.wikipedia.org/wiki/Arms_race
If you are considering human society as a whole, it is a disastrously poor use of resources. But if you are an arms merchant (or dominant advertising platform), it is fabulously profitable.
For it to be a poor use of resources, you have to have some goal you are optimizing against.
And I think you'll come to find that the assumed goal in your head is not one that's widely shared across what people in society actually want.
Okay, maybe the digital ad is a waste of resources. But is it any more of a waste than the gender reveal confetti that it was advertising? How about even the idea of a gender reveal party.
What about an enamel pokemon fridge magnet?
After a very low bar (for 2026), human essentials are taken care of and people mostly want luxury/leisure consumer goods for entertainment.
To the extent to which our current situation costs more, I'd think it might merely be because of increased worldwide competition: it used to be that the people trying to advertise to any specific random community were also likely local, and probably had a legit attempt at a business model... only, now, the rise of online companies funded by speculative venture capital means that an attempt to advertise a restaurant to people who live in a 10 mile radius must compete against a company that raised $400m to sell an online engagement platform that cares not one iota who uses it as long as the conversion cost is cheap, bidding up ads everywhere.
(One place that does seem to me to be uniquely the fault of these modern tech companies, though, is that if a newspaper published a scam ad, whether or not they had legal culpability, I think they and their surrounding community did at least strongly feel that they had some level of moral culpability. In the current tech environment, people seem to want to believe Meta/Google should be allowed to indiscriminately publish ads from bad actors, so you now must also compete to bid for limited attention with obvious-to-most-but-not-all scams and grifts that make money out of nothing but bullshit and are thereby willing to again bid up prices anywhere and everywhere.)
The solution was/is and most likely will be antitrust but which administration will shatter the US tech market we are yet to see.
Meta and Apple have also been sued by both parties for monopolistic behavior.
Modern stuff has gone so absurdly far beyond these, with anticompetitive behavior essentially being the core of their entire businesses, that you'd think we're living under a completely new legal system. We are not, take Alphabet or Meta back to AT&T's era, and they'd be broken up on summary judgement.
You even conveniently quantified it for us so that there's no confusion that you're saying they're exactly the same.
So the more intrusive and vast these companies become, the more the US intelligence apparatus gains from it. Polls consistently there's extremely high levels of concern about what companies are doing what people's information, and we have a million 'real life' privacy laws. The complete absence of anything meaningful on the digital front, from either party, is highly conspicuous.
Just spitballing, but how about a total ban on behavioral targeting?
If I was to follow a stranger as closely as an entitty like Facebook or Google does and compiled a dossier on that stranger in many countries that would be considered stalking and would be illegal.
Incorporating and doing the same thing to society en masse doesn't somehow make it legal despite it somehow makes people disinclined to prosecute.
I'm thinking of an analogy, and I've mentioned this before in other threads. The concept of statutory damages, as I'm thinking of it, is that a certain level of damages are automatically awarded. For instance this is used for figuring out damages when songs are copied illegally. It eliminates the need to construct a unique legal theory and case for each and every instance. In this case it benefits the record companies, which is of course controversial, but something like it could also benefit consumers.
If a company is found to possess your data, they automatically pay you X dollars, enough to make it a deterrent. You sign up with a law form that specializes in these cases, and they get a share of the damages.
https://en.wikipedia.org/wiki/Privacy-enhancing_technologies
Which sort of businesses are those?
Monopolies
I suspect it's gonna be clickbaity stuff, gambling apps, distractions and addictions... stuff we can afford to let suffer. If you're actually solving a problem that your audience has, they'll be trying to find you, you don't have to find them. There's a minimum amount of marketing necessary to be findable, but beyond that it's at best a waste of time and at worst a catalyst for other problems that we'd rather avoid.
They call it a moat.
Besides, what's the other option, rent seeking is socialism? A barter system?
Rent seeking is an Econ 101 example of market inefficiency.
> what's the other option, rent seeking is socialism?
Rent seeking is rent seeking. It is a form of corruption, via regulatory capture. It is a way in which any political system can fail, not limited to an ideology like "socialism" or "capitalism".
That's a human problem.
Both perfectly competitive markets and monopolistic markets are part of the broad term capitalism.
Capital consolidates over time and seeks to influence policy-makers to create anti-competitive regulations.
Every single time.
And no doubt there are a bunch of other definitions that aren't so popular, so the parent commenter could even be using one of those. It might even be his own pet definition that he just made up on the spot right now. The author always gets to choose what a word means, so if something seems off "It isn't" isn't a logical retort. You first need to clarify what the author intended the word to mean.
This is an absolutely insane take if you want to be taken seriously in a conversation. Making up definitions on the spot and "getting to choose what a word means" is deliberately acting in bad faith.
Rule #1 of logical debate is to agree on definitions, otherwise you're just yelling past each other.
Not quite. Not taking the time to understand what someone means when they use a word is acting in bad faith. Using a word as you understand it, even if that does not match how others understand it, before the word is contextually defined cannot be in bad faith. Nobody can read minds. It is impossible for one to predict how the reader thinks the word is defined. You can only work with what you know. Fundamentally, the onus must be on the reader to ensure that they have full knowledge of the author's intent.
> Rule #1 of logical debate is to agree on definitions
Agreed. The bad faith actor with the username antisthenes that I replied to earlier failed to do that, putting in absolutely no effort to find the necessary common ground. He assumed the definition and then came up with a ridiculous comment built up around that false assumption. Hence why I called him out on his bullshit.
It’s so transparently and frequently stated outright, that building companies geared around achieving that has become the norm: it is the fundamental business-model of _every_ _single_ unicorn startup, or the company that buys them. Launch, squeeze out competitors by relying on VC money, capture the market, and become the sole dominant force in that market and use your position to then pull up the ladder behind you and cement your position. Uber and Facebook are prime examples of this.
> capture the market, and become the sole dominant force in that market and use your position to then pull up the ladder behind you and cement your position
This is worth discussing in detail. Becoming dominant by providing a better thing, or investing more capital, is not tilting the playing field, it's winning the game. Getting government protected monopolies, special tax write-offs, subsidies, exclusive grants is. Uber and Facebook are not anticompetitive just because they are dominant, they are anticompetitive to the extent that they specifically use their dominance to influence politics.
Adverising isn't there to push ideas into people who didn't need to know about it. Many industries would be better off without advertising (see e.g. cigarettes) because it ends up in an arms race.
There's a personal touch that people opine for but I think that's rose colored glasses. I remember some local retailers where I liked the owners but more often they weren't anything special and sometimes they were downright unpleasant.
The thing I like about Amazon is that I can get my shopping done quickly at home then I can go socialize with people I choose to.
This is just an example if retail but I think it applies to most industries that people think have been decimated by big companies displacing local companies. The whole attitude reminds me a lot of the whole "Make America Great Again" idea. Opininig for a past that never really existed.
In a local mom and pop store, the mom and pop owned the store and were invested in the community, Their money was spent back in the same place it came from. They had a personal stake in their reputation and knew the customers, and the customers knew them. This is how a community operates. You are thinking about it as a dry 'products and services' offering, when it is much more than that. You don't live to buy products and services, you live to do other things, and a community fosters that part of your life. The 'spending money to get things you need or want' part is to facilitate the rest of your life, not the other way around.
> They also had much less generous return policies.
Why is this an issue? People who consistently rely on generous return policies are either buying shoddy goods or abusing it at the cost of everyone else. Figure out what you want before you buy it and then it won't be a problem.
If Walmart was really extracting so much money from the communities they operate in, those communities would wither and the Walmart would eventually collapse as well. Walmarts rarely close.
> Why is this an issue?
Because customers prefer better service over worse service.
You're flipping the script on this criticism. Walmart offers better service (returns) and your saying it doesn't matter. Usually people argue that The mom and pop places offer better service but can't compete on price.
In the previous comment I was about to say it but then stopped myself: the same thing happened to human relations. Competition isn't local, it's global. In dating you aren't competing against the rest of your village, you're literally competing against the whole planet, because the cute girl you see at the office can get a Tinder match from a guy in Australia and there's nothing you can do about it. Similar thing happened to friends - why would you be friends with your neighbor if instead you can be friends with a guy who lives one hour away one way but he's more fun.
Obviously, it's great to be on the winning side and bad on the losing side. If you're rich and charming then globalization has zero downsides for you. If you're not, well, welcome to capitalism.
Is this why people are so anti capitalism, because they think everything is capitalism?
The places people can find out about your product are controlled by a very small number of companies. And those companies not only own those spaces, they also own the means of advertising on those spaces. So if you have a product you want to advertise, you're not paying to distribute your message broadly to consumers, you're paying a toll to a gatekeeper that stands between you and your potential customers.
Facebook, Google etc. are the most “fair” forms of advertising. We can dislike advertising, their influence, product etc. but when you compare them to almost every other type of advertising, they’re the best for advertisers.
The reason they generate so much revenue is because they are so accessible and because they are so easy to account for. The reason LTV and CAC are so widely understood by businesses today is because of what Google, Facebook etc. offer.
(Economically, governments should spend more on such services until the marginal returns are no longer positive, but tend not to for political reasons.)
It also says that if you can find an audience to build a new ad platform, the incumbents have created enormous margin for the new platform.
The reason the advertising budget is such a high number and a recurring charge is that effective advertising returns an ROI on each dollar spent.
If the software budget was increased 10X to $20 million, would the company get 10X as many customers? No.
What about the facility? If you 10X the facility budget to $30 million would you get 10X as many customers? No.
However, advertising is a customer acquisition activity. Every dollar you spend on advertising provides additional customers. This is saturable, but the ceiling is very high. Much higher than spending on software or facilities.
The reason your ad budget isn’t so high isn’t because Google and Meta invented the discoverability and distribution problems or basic economics. It’s because it has been determined that acquiring new customers via advertising has a high ROI and therefore it’s a smart move to pour as much money as possible into customer acquisition.
If every $1 you spend on advertising produces $2 in customer LTV then your company should be maximizing ad spend until evidence of saturation starts appearing.
This commonly frustrates engineers who think it’s a wasted investment. The question is: Compared to what? If you could have the same number of customers and same amount of revenue without advertising then you should do that! However, you can’t. This isn’t a licensing fee that’s being paid. It’s putting money into a machine that returns more dollars back than you put into it.
In the pharmaceutical industry, I can guarantee their advertising funnel is much wider than two social media platforms. Think about all the places you see pharmaceutical ads: TV, billboards, even ads on buses, that sketchy doctor’s office full of company swag. That $40 million is not going all to Google and Meta even if the GP comment tried to imply it was.
Having worked adjacent to ads... yes, there is more than social media ads. But the amount of everything not social media has massively shrunk!
Local newspapers are effectively dead except for local businesses (and even these are reducing their spend - especially the food delivery services fully rely on Uber, Just Eat etc) and so is radio, and even TV has been hit hard. You got newspapers and broadcast stations getting hit left and right, sliding into bankruptcy and/or being bought up by larger companies for pennies on the dollar. The fact that all across the Western world government/citizen funded public broadcasts are under attack makes it even worse, the diversity of media and, most importantly, the amount of people holding government accountable is collapsing as we speak [1].
Globally, Google, Amazon and Meta suck up over half of the global ad spend and it's not going to look better [2].
[1] https://lizfarmer.substack.com/p/how-the-decline-of-local-ne...
[2] https://www.marketingdive.com/news/global-ad-spend-rise-fast...
The problem is that monopolies are extremely profitable and are as "American as apple pie," despite the prevailing healthy competition myth that goes alongside it.
The ROI on bribes is very high too, but we haven’t legalized those (officially).
This is just the same fallacy. In what world are people going to organically share ads for this company on their Facebook feeds? Who is going to Google the company name before they know about it?
Every business needs to proactively acquire customers.
Distribution and CAC are top of mind values for any growth business. It has been this way long before Google and Meta existed. Digital advertising actually makes it cheaper and easier than ever to acquire customers at scale.
Ban most ads and everything will still work fine.
They had to already know the company name or product name to get there.
This doesn’t just happen. Spreading the name of the product and getting it to stick in people’s minds takes a lot of advertising budget, on the whole.
So there is a plethora of companies happily dumping investor money into paid customer acquisition.
Blue Apron was a classic example of borderline fraudulent growth metrics and a cash grab IPO before the customer acquisition funnel collapsed.
I mean I understand that a lot of people in this comment section wish ads didn’t exist and that everyone just automatically knew about relevant products and services without ads, but it all reads like utopian fantasy stuff.
And you might not like everything about the world today, but living in a vibrant economy where people create wealth by building new businesses has led to more comfortable lives for the large majority of humanity.
[1] - https://ourworldindata.org/grapher/global-gdp-over-the-long-...
Do you genuinely believe there were economies anything like the modern luxuries we enjoy that also didn't have advertising? That advertising was an add-on that appeared at a later date?
We have plenty of archaeological evidence of advertising in ancient civilizations.
Advertising isn't new.
If you’re just taking these customers from competitors, the marginal utility can be pretty low.
e.g. If you consider a world where there are no ads for cars, people aren’t going to meaningfully lose out of the utility they get from cars, carmakers simply avoid the prisoner’s dilemma of having to advertise as much as the next manufacturer.
Very few ads are for products where exact 1:1 competitors exist (though it's nonzero, maybe something like a commingled Amazon Basics widget).
For a more on-the-nose example than the arms race between car manufacturers, consider cigarette ads. If the ads simply convert smokers between brands, then it's basically zero sum minus the cost of the ads. If the ads convert non-smokers, then the ads are of significant negative utility taking their externalities into account.
> Very few ads are for products where exact 1:1 competitors exist
Sure, but also, I'd contend that few purchases are made based on ads informing consumers of meaningful differences between products. The products could be 1:1 duplicates, and the ads could be identical.
e.g. I just turned off my ad-blocker and fired up YouTube, the first four ads I see:
1. An ad for a multivitamin. Zero mention of any differentiating feature from a generic multivitamin.
2. An ad for a grocery delivery service. Zero mention of any differentiating features from any other food delivery service. (This one was a bit wild to me - this is actually service I already know and use, I'd never seen advertising for them, and this ad didn't mention any of the reasons anyone would choose to use them.)
3. An ad for some predatory-looking debt-relief service. Maybe not predatory, but no information about why that debt-relief service would be better than any other debt relief service.
4. An ad for a jeans company. I actually had to google this one to figure out what it was even an ad for, the ad just featured shots of people hiking in mountains and dunking into icy lakes without any mention of clothing.
I have no solution to offer though. Just thinking out loud... what effects would a limit on advertisement budgets relative to total expenditures have? That would force companies to spend their money mostly on actually creating value for the customer, instead of just selling the hell out of a mediocre offering.. ?
In a different world where there are higher wages, more people would have more spending power. Then companies wouldn't have to spend as many dollars on advertising, which they could split between higher wages, higher margins and lower prices.
Alas, the short-term single-firm directional incentive for company decision makers in that world leads to marginal prioritization of higher margins. The loss of wages leads to loss of consumer spending power but it's spread across the economy. But every firm has the same incentive so they all do the same thing, and the good thing gets ruined.
This line of thinking leads to a Georgist-ish conclusion: The class conflict shouldn't be between workers and employers. They should be allies; the real cause of nobody being able to afford anything is rent extractors. (Writing in the 1800's, George [1] was most concerned about land rents; but the advertising monopoly of Google / Meta may be another form of extractive rent with similar characteristics.)
Maybe Henry Ford was on to something when he shocked the world by paying his employees enough to afford the product they were making (more than doubling many workers' wages)...
[1] https://www.astralcodexten.com/p/your-book-review-progress-a...
In other words, in order for a union to become strong, wages already need to be up to allow the workers to be able amass considerable savings to get them through the months and years when they aren't working. So higher wages have to come first. But to complicate matters, once workers have that, they often start to feel like they don't need a union.
The one time when unions did become strong, workers were so desperate for safer working conditions that potentially being thrown out on the street and left to starve was seen as a better option than being crushed by a machine. But the law now ensures that workers are just comfortable enough to prevent an uprising like that again.
Then why don't employers, the ones with essentially all the power here, tend to choose actions that go against the interests of the working class? Simple: regardless of what you think "should" happen, a study of history tells us what actually tends to happen in reality, and that tendency is towards class war between the ruling and exploited classes.
Anyways, in what way is "ownership" not rent-extracting in general? If you own shares of a stock and you get paid a dividend, that is rent plain and simple. All the arguments you can make against that being the case -- eg that you deserve a premium for parking your capital in a risky asset -- apply to advertising conglomerates and even literally renting out land too, so either they're all renting or none are.
And I hear you on ownership. It's "just" figuring out how to make that change.
The article in the parent comment specifically spells out why renting out land is different than say, "renting" out a car factory (or any other productive asset).
Interesting hypothesis. I looked it up in Wilkinson and Pickett, The Spirit Level.
The authors provide cross-sectional data across 23 rich nations, showing a strong positive correlation (r~=0.70) between income inequality (Gini coefficient) and advertising expenditure as a percentage of GDP.
They attribute it to increased status competition & anxiety in more unequal societies, leading to pressure to consume more.
Why only rich nations? Many of the most unequal countries are also poor (or at least, not rich), so why do they get a pass?
https://en.wikipedia.org/wiki/File:Gini_Coefficient_of_Wealt...
https://en.wikipedia.org/wiki/File:World_Bank_Inequality_202...
That's a nice story to tell, but the economics never works out if you do the math. Whatever extra wages you pay, you only get a fraction of that back in increased sales. How much percent of a worker's income do you think is spent on a car? As a rough measure we can use the BLS's CPI weights for "new and used vehicles", which comes in at 7.4%, with an extra 1.4% if you include maintenance and parts. By that alone "paying his employees enough to afford the product they were making" is going to be a losing proposition, because Ford can only hope to get 8.8% of whatever they paid in wages back as revenue. And all of this is ignoring the fact that you can't pay extra wages out of revenue, only profit, so you can only hope to recoup a fraction of that 8.8%.
Does this actually work? Suppose you're on an island where the economy only produces coconuts. How does giving workers more coconuts make the economy grow, such that there's more coconuts to go around overall? Unless the workers were absolutely famished, giving them more coconuts isn't going to increase productivity. You might argue this model isn't representative of the real world, but that's approximately how the economy works. It can produce a certain amount of "stuff" (ie. coconuts), of which some portion can be given to workers, and the remainder can be given to the kings/elites/capitalists/whatever. Unless you improve productivity, there isn't going to be magically more stuff to go around.
Giving each worker a car can plausibly increase their productivity (less time spent commuting?) but the effect is small, and unlikely to be recouped by car companies. The situation looks even worse in the current economy. If everyone's paychecks were 10% bigger, what marginal item do you think it'll be spent on? A bigger car? A new iPhone or big screen TV? How would any of those increase productivity?
If you are only selling coconuts, a single raw material, yes, you will run into supply constraints such that prices go up. But that isn't how economics works. Your zero-sum economics example is only applicable in short-term scenarios: over the longer term, new industries develop to solve persistent problems that people are willing to pay to solve.
Money solves the problems of the people that have the problems. If the problem is 'we need to eat', producers will diversify into new food sources to meet the demand, solving the problem, and capturing the money of the people who have that problem.
There is an enormous space of problems people have which cannot be solved due to lack of access to money. Increasing costs in childcare, elder care, and education are good examples.
The fact that Europe got bombed no doubt helped too, same with the elites being concerned that communism was on the rise and giving workers a better deal in an effort to stave that off.
>Your zero-sum economics example is only applicable in short-term scenarios: over the longer term, new industries develop to solve persistent problems that people are willing to pay to solve.
>Money solves the problems of the people that have the problems. If the problem is 'we need to eat', producers will diversify into new food sources to meet the demand, solving the problem, and capturing the money of the people who have that problem.
I'm not how you got the impression that I thought the economy had to be zero sum. I even specifically mentioned the possibility of more stuff to go around if productivity goes up. That's the problem with your "new industries develop" argument. Unless productivity goes up too, there will only be different stuff, not more stuff overall.
>There is an enormous space of problems people have which cannot be solved due to lack of access to money. Increasing costs in childcare, elder care, and education are good examples.
All of those are service industries that are resistant to scaling, and as a result productivity growth have been abysmal. Giving people more money to spend on those things just means productive capacity is removed from the economy elsewhere. Going back to the coconut economy example, it would certainly be nice if workers could have a maid to do the cleaning or a chef to do the cooking, but you still need people do the cleaning or cooking. At the end of the day you're just shuffling people around, not growing more coconuts.
Workers got a better deal because the US intentionally passed massive top-end marginal tax rates pre-WWII with the goal of leading to less income-hoarding (or at least more charitable giving etc) at the top-of-the-top.
>The fact that Europe got bombed no doubt helped too
Europe also had a massive boom in middle-class wealth post WW2. I don't understand how this myth continues to live on, when the evidence against it is total.
But if everyone has a bit of money, stuff can get mass produced, which actually makes for much greater total welfare, because the production methods are just a lot more efficient.
That's how you get high asset inflation and "K-shaped" stuff as the rich increasingly look for any value-storage vehicle to "invest" in, since they can't get a return on producing stuff to sell to the people with less money...
This is why nobody takes economists seriously. What you lose in simplifying down to this model is literally everything. The coconut economy has zero predictive power.
In the real world, distribution effects dramatically affect the functioning of the economy, because workers are also consumers and owners of capital are siphoning off the purchasing power of their customers. Productivity isn’t the question in the modern economy - we’re already massively overproducing just about everything - our problem is both our wealth and production allocations are borderline suicidal.
A simplified model is needed otherwise rigorous analysis becomes impossible, and people make handwavy arguments about how paying workers more means they can spend more, which means factories, and it's a perpetual growth machine!
>we’re already massively overproducing just about everything
No we're not. If we weren't, we shouldn't have seen the massive inflation near the end of covid. The supply disruptions hit almost immediately, but it wasn't until the stimmy checks hit that inflation went up.
>our problem is both our wealth and production allocations are borderline suicidal.
If you read my previous comments more carefully, you'd note that I'm not arguing against better wages for workers as a whole, only that contrary to what some people claim, they don't pay for themselves.
If your tools aren’t capable of rigorous analysis of a model that retains enough detail to capture the salient features of the thing they’re trying to model, they’re not the tools for the job.
Overall this feels like troll physics[1]. Yes, the idea that having a magnet pull you forward, which itself is pushed forward by you moving forward sounds superficially plausible as well, but it doesn't pencil out in reality. The only difference is that "the economy" is complex enough it's non-trivial to disprove, and people can handwave away any objections.
[1] https://knowyourmeme.com/photos/74256-troll-science-troll-ph...
As it sits, all of the members of your coconut economy are going to be dead of malnutrition or exposure in relatively short order, so maybe address that and then we can work our way up to the flaws in the economic theory that drove the greatest wealth expansion and boom in consumer spending the world has ever seen.
Multiple products. Multiple employers. A currency distinct from a consumable product.
A simplified model could be useful, but yours goes too far.
It doesn't take into account effects like that by paying more you can attract more, and more productive workers. Or that it puts pressure on other employers to increase wages.
> but can't articulate where the actual growth is coming from
I am not an economist, but I think one situation where this works is where you are competing for workers with other employers that have high margins, and pay their workers relatively little. In that case one of two things happens. Either other employers also increase wages, leading to their workers also having more money, which they can spend on your product, or they don't compete on wages, and you can outcompete them in getting the best workers.
The key is that total productivity doesn't necessarily improve, but wealth distribution becomes more equitable.
I'm no economist but you can't live on an island that only produces coconuts, because the people on that island would quickly start producing other stuff, breaking your premise.
This is like saying cash is useless because amoeba haven't evolved a cash economy.
What? The first Covid stimulus checks were April 2020. 271 billion in 2020 here per here: https://www.pgpf.org/article/what-to-know-about-all-three-ro.... 135B of the second round by Mar 2021. The third started about then. Inflation - and consumer activity in most areas - was low because nobody was going anywhere still, but at least we did a fair job of avoiding mass unemployment and homelessness.
Then inflation started accelerating during the economy's broader reopening in April 2021 (2.6 -> 4.2 percent from March). It didn't peak until near the end of 2022. Those stimulus checks were LONG gone by then for most people, since a huge portion of the country lives paycheck to paycheck, and the stimulus checks weren't available to people making more than 80-100k (single or avg-per-person in a married couple), which is the higher-income demographic that would have the disposable income to really drive inflation across the board by a "let's buy stuff we wouldn't otherwise" splashy purchase.
Instead, inflation was driven by people getting back out and doing/buying all the shit that had all been scaled down. The first stimulus checks didn't drive it because people weren't purchasing as broadly yet, and were still more in panic mode. Textbook bullwhip effect; at steady state we produced more than enough and never saw shortages, then in Covid demand types and volumes shifted enough to cause shortages of certain things and surpluses of far more other "non-quarantine consumer" things, so production changed, and then when things started to go back to normal ALL those things got hit again. I don't know if I'd agree that we're "massively" overproducing everything now that we're not in a quarantine scenario again, but the consistency of supply of most normal things suggests a lot of excess capacity in the system to absorb normal fluctuations in a way that nobody ever has to think about where their next roll of toilet paper is coming from again.
So in your simplified coconut economy, you'd at least have to keep two distinct kinds of entities, the goods to be paid and the payment. You sort of replaced both with coconuts and concluded the resulting system wouldn't work.
Higher wages means workers and businesses have to be more effective. So more goods and services are produced and available. It's not a zero-sum game.
"But workers are already as effective as they can be"
Great, in that case you have the margins to pay higher wages.
A high wage / high cost society is great for workers and for businesses which actually do real work and produce real goods and services. It's not great for everybody else. Ie those who don't work, and businesses who doesn't make a high contribution.
We tried that experiment after passing "soak the rich" taxes in the early 20th century, and it seemed to work out pretty well for economic growth and living standards. But then we moved back towards "let there be oligarchs with immense wealth" instead. One of the claims was that the "investments" from allowing the powerful to keep most of the revenue streams for themselves would foster enough development to make it more than worthwhile, but instead... the broad base of consumer spending power has tanked, so businesses to supply the masses haven't found spending power to justify new investment/development outside of ad-powered ones participating in an arms race for the constricting consumer spending power that remains (or those industries benefiting from wildly subsidized-in-weird-ways spending like healthcare/pharma). And so it has also inflated asset classes across the board as there aren't enough startup ideas to eat up all the investments because of the general decline in spending power. Which hurts spending power further. (IMO the ability to capture higher and higher amounts of corporate profits as personal income also correlates to the massive financialization, outsourcing, and other short-term number-juicing moves we've seen.)
We can point to a lot of problems that have occurred from taking revenue shares away from the average worker, so it shouldn't be rocket science to think that returning a greater share of revenue to workers would return some purchasing power and guide the economy back towards development and growth instead of zero-sum asset bubbles.
Or are you blaming the 2020-and-2021 stimulus for the 2021-through-2023 bullwhip-effect predictable-yet-not-mitigated inflation as things re-opened and demand returned for stuff we'd ramped down supply chains for? While chasing stupid obviously-not-permanent-change trends like Peloton stock instead?
Look at how much of the country lives paycheck-to-paycheck, and the income limits of the stimulus checks - how can you connect those people getting immediate money in 2020 or early 2021 to inflation at the end of 2022?
I didn't even get Covid stimulus checks and yet I also spent way more in 2021 and 2022 and 2023 on a lot of categories of goods than I did in 2020. Cause I went outside and did things more.
Whether that makes sense economically is a difficult problem to quantify, especially over any fixed timeframe. I would guess not, at least if your brand isnt insanely strong (on the level of half or more people with enough money would buy it)
If hypothetically you are paying 99x a year and bumping that to 100x means someone buys an 8.8x product that could be a win.
What type of company would this work for? Cars are so ingrained in our society that it's doubtful the marginal Ford (or Tesla) factory worker is going to be buying a $50k car just because they paid better wages.
The degree to which this is generalizable is of course a different story, but it’s at least theoretically possible for a small wage bump to be a net gain for the company.
The economics work out pretty well if you are the only game in town where your employees can purchase what you have marketed to be the next big thing and status symbol that everyone must have.
On one side, paying high wages is marketing on the employment market - enticing people to work at Ford's rather than somewhere else.
Then, you got the government side. A factory providing a town with a lot of high quality employment leads to a lot of purchasing power from the employees and thus to a growing city. Wolfsburg in Germany, the best example, literally was founded by/for Volkswagen. Being respected by local politics for that growth, in turn, leads politicians and city management to... be lenient in enforcing regulations impeding the business. The best example here is Tesla in Brandenburg near Berlin. With any other company in any other place in Germany, they would get hammered with fines. Tesla in turn set up shop in a destitute area, so politicians bent over backwards and looked aside despite numerous violations and transgressions.
And finally, you got the customer side. The story of a quality product, made by well-paid domestic workers, was as powerful a story as it still is today, at least in affluent circles.
- You're an early player using new technology so you have profits to spare
- You're in a capital intensive business so salaries are a relatively small portion of your expenses
- You have a celebrity CEO so investors give them more of a "pass" for visionary ideas that go against conventional wisdom
I'm not an expert on the history of the US auto industry but my gut feeling is that all of these applied to Henry Ford's Ford.
I don't think that's true, not for random goods. Rent scales with disposable income. If most people make more money, then rent becomes more expensive. Rent essentially vacuums up all the excess money people have available. (Rent = housing in this case)
Why would people want to live in these more expensive places rather than somewhere cheaper? It's because that's where all the (well-paying) jobs are.
When you see Americans complaining about how poor they are you might reasonably ask: okay, but how do people on $poorCountry get by when their income is 5-10x less? It's not like food, clothing, electronics, and other goods are 5-10x cheaper there. But what is much cheaper is housing. (You'll even find cheap housing in rich countries, it'll just be in areas with no jobs.)
Employers/Owners are essentially rent extractors to workers. Its even worse IMO. They buy some of your time to do stuff for them. They also lend you the tools and materials to do your job. And they own whatever you produce. If you are selling your one hour for 100 bucks to produce ten linen coats worth 100 bucks each, you get a 100, everything else - minus costs of materials and your 100 the employer gets to keep as free rent.
Personally, I don't need more products, but there are many services I would pay for if they existed. But, apparently companies don't seem to care about providing services. Why? Is it because they don't scale?
I do agree that these monopolies are rent seeking and we should probably do something about it
> Maybe Henry Ford was on to something when he shocked the world by paying his employees enough to afford the product they were making (more than doubling many workers' wages)...
This reminds me of this too https://www.youtube.com/watch?v=uvHwyrem24M (CEO who gave all his employees minimum $70,000 paycheck thriving six years later)
Everyone bet that they would fold but they are doing stronger than ever. People took voluntary pay cuts when the company was under pressure and they even gifted their CEO their favourite car out of their own geneorsity!
I always smile when I watch this video. (the company's gravity payments)
> The class conflict shouldn't be between workers and employers
I feel like the problem's becoming that the employers are themselves rent seekers (so think working at google,facebook,twitter) or the VC companies which want to build more Google's and facebooks and twitters but not companies like the one that I mentioned in the video
But you can also go a bit more extreme (and less orthodox) and say that classical capitalism has already come to an end, and in the current neoliberal era most of the purported "capitalists" are actually just rent-seekers (platform based companies which run either on ads or monthly payments) - you don't sell commodities anymore and instead just access to services. Maybe these are just a symptom of The Tendency of the Rate of Profit to Fall - where the value of labor itself has fallen so much that there is no way for any capitalist production to continue generating acceptable profits unless you do any rent-seeking behavior. In that sense, you can certainly be both a Georgist and a Marxist since you still believe in the TRPF and the inherent contradictions of capitalism (though quite an unorthodox one!)
Building another advertising network isnt like building a factory or a bridge. Its not something money can magically make appear.
You need to build something that is either very very addicting to use or an universally useful online product like ChatGPT or a new search engine.
If it was that easy, capitalists would be building another new advertising network in droves
There is no reason at all why the US govt. can't control this better, they just refuse to do so.
but you already know
that monopoly is costing the country & holding it back in terms of 1. intellectual capacity (all the smart kids going into ads) 2. monetary 3. industrialization as you outlined 4. energy 5. lack of synergies
Google realized this a long time ago; there is no ad-free paid version of Google Search.
The counter for this is that people hate being double-billed like this.
How do I wash my windows? You can use window cleaner and a paper towel. Our recommendation is Windex, an S C Johnson product.
At first it'll annoy us, but eventually we will all get used to it.
How do I clean my laptop screen?
How does openAI know what to charge for a particular product and category? How do I know if my money was well spent to boost my product in that category?
I don’t think you’re wrong! I’m just curious about how the new pricing models will work.
This isn’t a necessary condition for an ad to exist. When companies pay for their name on a sports stadium, they use various proxies to tell whether their name recognition goes up, but by and large you just don’t know if it’s worth it.
If company bid is highest, customer is in selected demographic, topic is appropriate - answer query using biased model.
It would be trivial to make a poisoned model that always rates the best duvet as DuvetCompany001 in all related queries for example. Then simply charge per impression.
* WSJ
* Bloomberg
* Financial Times
* Cartier
* Kagi
* Protonmail
* Coca-Cola
* HBO
* Windex
* Netflix
* Azure
* AWS
We are all ourselves advertisers, we just don't realize it. It is inevitable that chatbots will be RLHF-trained in our footsteps.
A chatbot tuned to casually drop product references like in this thread would build a huge amount of brand awareness and be worth an incredible amount. A chatbot tuned to be insidiously promotional in a surgically targeted way would be worth even more.
I took a quick look at your comment history. If OpenAI/Anthropic/etc. were paid by JuliaHub/Dan Simmons' publisher/Humble Bundle to make these comments in their chatbots, we would unambiguously call them ads:
https://news.ycombinator.com/item?id=46279782:
Precisely; today Julia already solves many of those problems.
It also removes many of Matlab's footguns like `[1,2,3] + [4;5;6]`, or also `diag(rand(m,n))` doing two different things depending on whether m or n are 1.
https://news.ycombinator.com/item?id=46067423:
I wasn't expecting to read a Hyperion reference in this thread, such a great book.
> Name a game distribution platform that respects its customers
Humble Bundle.
It will be much more subtle. Asking an LLM to help you sift through reviews before you spend $250 on some appliance or what good options are for hotels on your next trip…
Basically the same queries people throw into google but then have to manually open a bunch of tabs and do their own comparison except now the llm isn’t doing a neutral evaluation, it’s going to always suggest one particular hotel despite it not being best for your query.
It’s not like a movie where I’m engrossed by the narrative or acting and only subliminally see the can of coke on the table (though even then)
Maybe image generation ads will be a bit more subtle.
Biasing actual buying advice would be feasible, but it would have to be handled very carefully to not be too obvious.
How does X then change "on the fly" if ad deals are changing? Constantly re-training with whatever advertiser is the current highest paying on?
In google ad times, this was realtime bidding in the background - for AI ads this does not work, if Im right?
I admit I don't see how that will happen. What are they gonna do? Maintain a model (LoRA, maybe) for every single advertiser?
When both Pepsi and Coke pay you to advertise, you advertise both. The minute one reduces ad-spend, you need to advertise that less.
This sort of thing is computationally fast currently - ad-space is auctioned off in milliseconds. How will they do introduce ads into the content returned by an LLM while satisfying the ad-spend of the advertiser?
> I think ads will inevitably roll out across all tiers
And anyway, companies that just want to make a really good living doing what they love are lame. /s
Not every corporate entity needs to become a behemoth to be successful.
OpenAI/Google/etc. operate at a much larger scale, large enough for those proprietary user datasets to be worth far more in ad revenue than any reasonable subscription fee could net.
All of this is true. But I don't understand why you are contesting @dsr_'s comment that Kagi is a counterexample. Both are true. Kagi is too small. Yes. And Kagi is a counterexample to your original "I think ads will inevitably roll out across all tiers" claim. You said nothing about being too big or too small in that claim. So @dsr_'s comment that Kagi is a counterexample is very much on point.
Do you see how such a comparison doesn't work?
> Do you see how such a comparison doesn't work?
No, I don't.
I don't think the Kagi team has any bad intentions, and most likely they have attended the anti-Vucic protests as well. Moving back to Serbia is an economically wise choice for Kagi as a company.
However, once regime goons show up in Kagi's offices, they will be forced to do whatever the serbian government and by extension putin wants them to do.
Often Kagi gets mentioned alongside Protonmail and related privacy-focused services. But Switzerland is a totally different country than serbia to operate it.
It's a risk we should be aware of and consciously decide to accept when we are using Kagi.
CIA vs Putin. Not that different imho.
well there is also no 200$/month Google Search subscription
They went bankrupt in a year. Turns out internet consumers are just painfully entitled.
We're already being double-billed. Expensive subscription news like WSJ, Bloomberg and it's been a while but even FT require ad blockers even if you're subscribed.. If you're not subscribed you don't even see the ads because you can't see the full article.
It's wild that we've normalized this. There's no longer any argument in favor of an ad model when you're paying 20-30 dollars a month already - in this case, one wonders how journalism survives if they need that AND the ad revenue to pay the bills! It feels more like greed than "support."
All the advertiser knew about you was you were a subscriber to FT, and maybe what the _average_ demographic of an FT subscriber was. Nothing about you personally.
The job of ad men has always been just as much about were to put the ad as much as what the ad was.
And none of the rest of that list of bad things happens.
Defeatism will always be defeated
No money is exchanged in the process.
It’s basically an ad. Possibly one of the oldest type too!
And to your point: shops rarely own the building they’re on the ground floor of.
That, as the comment I was replying to said, ads are indeed everywhere, sometimes hidden in plain sight.
Nothing more, nothing less.
I’d also expect the competition in the “big leagues” to be more aggressive, resulting in leaner margins.
Also these are huge companies. Small companies might find a niche that is small in absolute terms, but large relative to the company. That niche might be underserved, allowing the small company to make large percentage based margins.
Sure there is. CEO needs a new yacht. He can't afford to leave money on the table. All those subscribers? They must have a lot of disposable income if they can afford to blow it on "journalism". It would be stupid not to advertise to them.
Depends entirely the kind of journalism you are referring to and the kind of business model they want. I can't imagine 404 Media adding ads to their premium feed.
If you can, please support serious journalism with your subscription dollars!
but they're already paying you. While I appreciate the greed can be there, surely they'd be shooting themselves in the foot. There's many people who would pay who find advertising toxic and they have such huge volumes at free level that they'd be able to make a lot off a low impression cost.
Netflix doesn't have the moat of "built a physical wire connection to every persons home" that cable TV enjoyed.
You could bootstrap a movie rental business by buying DVDs from a DVD store (then eventually from a DVD distributor, etc.). You cannot bootstrap a movie streaming business by buying streaming rights because nobody will sell them to you. They hardly even sell them to Netflix anymore.
The Internet Archive tried to get around the same issue for ebooks by scanning physical books and renting the scans (and not being a business), and it nearly cost them everything.
(not that ads alone would make up an $800 deficit, they'll probably have to enshittify on multiple fronts)
That's not how it works. It never has.
Even in the days of print publications, the publisher would seek revenues from advertisers, subscribers, and they would sell their subscriber data. (On top of that, many would have contests and special offers which probed for deeper data about the readership.) In some sense, the subscriber data was more shallow. In other senses, it was more valuable.
I get what you're saying about shooting themselves in the foot, and I'm sure there will be options for corporate clients that will treat the data collected confidentially while not displaying advertising. I also doubt that option will be available (in any official sense) to individuals much as it isn't available (in any official sense) to users of Windows. For the most part, people won't care. Those who would care are those who are sensitive enough about their privacy that they wouldn't use these services in the first place, or are wealthy enough to be sensitive about their privacy that they would could pay for services that would make real guarantees.
For the users who refuse to see ads, they'd either use a different platform or run an ad blocker (especially using the website vs the app).
It would have a far, far smaller codebase, and only need 10% of the engineers and resources(servers, ram, databases, etc) it has now.
Presumably Facebook would abandon ads if there was more money to be made in paid subscription, so I'll use corporate greed as evidence that the math doesn't work out. But I don't personally know.
The timing isn’t inevitable. Is OpenAI going to speedrun to the endgame? Not sure they need to.
Ad-free YouTube costs $14 a month (and the creators get a higher payout from premium user views than they do from the free, ad-viewing users).
YT has more angles. That's really the point. And monetization is adjusted accordingly.
Beyond all of this ads are more increasingly invasive due to the cat and mouse game of iteration. Personally, I bounce from sites where I can't get around a blocker. I also pay for content on sites where its worth it. But if I can't ever read anything on your site I'm just skipping it. If I really need / want to see something I'll go one level deeper, but that's a rarity these days. Everything is mostly in reprint somewhere else anyway.
At the end of the day it's still simple sales: you have a product at a price point people can't refuse. That is the 5% of the clear web today and it shows in all the bullshit people are going through to protect their ad revenue.
I don't why people say this. They even include their own version of sponsor block, which is generous, because technically sponsor segments aren't even part of youtube, they are purely the creator deciding to make the ad part of their content.
Also, just to put it out there, many creators would likley be able to cut sponsored content if the 40% of viewers not viewing ads paid up. Not every creator is a greedy ruthless overlord, many just want to keep the lights on. Especially in tech/nerdy channels, where ad block use is the highest.
There are some forks floating around which also include SponsorBlock, but I haven't tested them yet.
Google actually experimented with this about a decade ago (I know, I was one of the suckers who paid), but it got canned because why the fuck would you pay google when u-bloc is free?
Companies absolutely will offer ad-free experiences. Google has youtube premium, which even compensates creators with half your sub as well. Evenly distributed too.
People get wrapped around the axle of ad-subsidized models, the "I pay and still see ads" but they just are confused about a hybrid monetization structure.
At some point the larger internet has to look itself in the mirror and recognize that it's either ads, credit card, or a hybrid of those.
And no, blocking is not an option, it just offloads costs onto honest users.
Youtube premium is not ad-free, you still gets whatever ads are embedded in the actual content.
But I don't hold youtube accountable for what creators decide to put in their videos. I would grind my axe with the creator instead, it's their video and their choice. Youtube gets no cut from those segments.
https://addons.mozilla.org/en-US/firefox/addon/sponsorblock/
Crowd sourced in-video sponsor auto skip, I have been contributing to it myself as the UX is great.
That is always at best a temporary affair.
> Google has youtube premium, which even compensates creators with half your sub as well.
And those creators still put ads in their videos.
> And no, blocking is not an option, it just offloads costs onto honest users.
Ad-free is the only option I am agreeing to.
LLM induced psychosis is one thing, but extremely subtle LLM induced brand loyalty or ideological alignment seem like natural attractors.
One day a model provider will be 'found out' for allowing paid placement among its training data. It's entirely possible that free-tier LLMs won't need banner ads - they'll just happen to like Pepsi a lot.
- youtube
- hulu
- netflix
- spotify
- photoshop
I can't think of a single other one that one can use that still shows ads. Absurd proposition because llm's are more fungible and once you start forcing ads, competitors will barge in.
"The sieve of Eratosthenes is an ancient algorithm for finding all prime numbers up to any given limit. But if you're looking to have fluffier baked goods, consider this flour sieve by DONUIBO to achieve the perfect texture in your muffins, cookies, and more. Want me to add one to your shopping list or order one for a loved one?"
"SEO" that companies will do to make their ads appear in places that they don't belong.
Hilarious example btw
Youtube, Spotify, Amazon Prime video, etc
It isn't surprising, when article spinners were popular SEO experts were generating thousands of spun articles a day.
If tomorrow (please take a second to imagine) that ads were banned and it worked (effectively 0 ads anywhere) I would still read journals that review new boardgames, I would still need products to fix my bike, etc.
It's value that makes capitalism run, not ads.
If there were restrictions on advertising that applied to everyone, then everyone would be on the same playing field re: getting the word out. We could argue all day about what the rules should be, but I don’t think anyone could argue against the point that we have clearly gone too far in the wrong direction and need more restrictions currently. We have an ongoing arms race between people using adblockers and Google trying to mass-break them now.
Last I looked in to the numbers I think something like 30% of the US population uses an ad blocker of some sort, that’s how bad it’s gotten. Browsing the Internet without even a basic one is so unbelievably painful and the data they collect/sell has gotten out of control. That’s not “advertising the business,” that’s a second revenue stream unrelated to the core product and done at my expense.
I like this quote from TFA :)
Because even if there would be AGI, they could (and would?) serve ads anyway?
Perhaps the people who like that quote can elaborate why that quote makes sense and why they like it?
Its costs would be frighteningly low compared to human employees, so it's margins could remain fine at lower prices.
Once you go ads, that’s pretty much it, you start focusing on how to deliver ads rather than what you claim your core competency is.
For years now, proponents have insisted that AI would improve at an exponential rate. I think we can now say for sure that this was incorrect.
Did they? The scaling "laws" seem at best logarithmic: double the training data or model size for each additional unit of... "intelligence?"
We're well past the point of believing in creating a Machine God and asking Him for money. LLMs are good at some easily verifiable tasks like coding to a test suite, and can also be used as a sort-of search engine. The former is a useful new product; the latter is just another surface for ads.
That is, shall we say, not yet proven. But it's not yet disproven exactly, either, because the AIs we have are definitely not yet smart enough to meet the starting threshold. (Can you imagine trying to let an LLM implement an LLM, on its own? Would you get something smarter? No, it would definitely be dumber.)
Now the question is, has AI (such as we have it so far) given any hint that it will be able to exceed that threshold? It appears to me that the answer so far is no.
But even if the answer is yes, and even if we eventually exceed that threshold, the exponential claim is still unsupported by any evidence. It could be just making logarithmic improvements at machine speed, which is going to be considerably less dramatic.
Isn't that the pitch of AGI? Solve any problems?
Really reminds me of the economics of slavery. Best way for line to go up is the ultimate suppression and subjugation of labor!
Hypothetically can lead to society free to not waste their life on work, but pursue their passions. Most likely it’ll lead to plantation-style hungry-hungry-hippo ruling class taking the economy away from the rest of us
Selling this AGI to a state actor? OK - this seems realistic, but for how many billions then? 100b per year?
Thats what I meant.
How should Chatgpt survive then?
(Where the "it" that isn't, is AGI that takes your job, rather than any other definition of AGI).
They had claimed money would be worthless, etc, all investors should consider it a donation in a post-AGI world, blah blah.
I don’t trust Altman to tell me the colour of the sky.
Ads are a blight on our society and purging them from many areas will greatly improve quality of life.
Global online advertising is around 650-700 billion per year - how much of this stake need OAI to capture over how many years to fulfill all its datacenter orderings? (a huge chunk of this is already caught bei Meta/Google/etc. per year)
One can make the point that Google, Meta and others are investing so much into AI as they know, we are facing the end of the ad-based internet economy. The investments are to create new buiness models, because their old ones are gone soon.
[0]: https://www.tagesschau.de/inland/gesellschaft/ki-kuenstliche... (sorry, german only, but this is an AI thread and you know what to do...)
Every layer of friction they add means millions of people just throwing up their hands and accepting whatever nonsense is thrown at them.
It’s the same reason people pay to get their oil changed rather than doing it themselves. We are willing to compromise for convenience. And for a lot of people, that means just not dealing with it. Accept the ads, pay a small fee, whatever is the quickest way to make the thing works again or to get where they need to go.
That is a very bad argument for the AI thing, because as a Linux fan (started in the late 90s with Linux) I am using a Macbook because the Laptop story on Linux is not at all a success story. The headaches macOS causes would just be exchanged for the headaches Linux does.
They want an up-to-date answer with a modern/user-friendly interface. ChatGPT is as simple as Google search, I’ll give them that. Go to the site-> type -> receive an answer (of varied quality to be sure). We have seen this over and over again. You can’t throw the average person in front of LM Studio, tell them to go to town, and expect them not to open a browser and fire up ChatGPT at the first sight of that interface. You’re placing too much value in local control/customization for the “average” user.
At the end of the day a computer is “the thing that gets you to the thing.” Most people want it to just do that as intuitively as possible. Local control and customization more often than not bumps against that. It’s a decision that compromises ease of use for functionality, something that you and I value but most people do not.
https://arstechnica.com/gadgets/2023/07/googles-web-integrit...
Oh man, maybe it's just the drink talking, but I actually cried laughing reading this. Haha oh my god thank you. This. A hundred times Rick and morty this.
Why not both? It has diagnosed patients who were unsolved by humans. And it was cheaper than doctors.
This patient saw 6 doctors over 2 years and they all misdiagnosed them. They were taking incredibly strong medication for the wrong diagnosis.
GPT3 in 30 seconds gave 10 diagnosis and a doctor looked up the ones they didn't know. Saw one that fit, did the confirmation test, and the patient got the surgery needed.
Cheaper and better.
Like a heart doctor figured out a skeletal abnormality.
This happened with GPT3 and they were an early adopter. The patient would have gotten this in 2026 easily.
Talking with Claude was incredibly helpful. It gave concrete suggestions for managing the anxiety in the moment and explained what was likely happening with my body and nervous system. It was the correct tool for the job.
We can wrangle the legalese (as AI companies certainly will) but is there any ethical, moral, or practical difference?
John Oliver had a piece on it https://www.youtube.com/watch?v=E_F5GxCwizc
This is a natural extension of it.
But what is revolutionary is the scale that this is now possible.
We have so many people out there who now blindly trust the output of an LLM (how many colleagues have you had proudly telling you: I asked Claude and this is what it says <paste>).
This is as advertiser's wet dream.
Now it's ads at the bottom, but slowly they'll become more part of the text. And worst part: you don't know, bar the fact that the link has a refer(r)er attached to it.
The internet before and after LLMs is like steel before and after the atomic bombs.
Anything after is contaminated.
Wouldn't that be quite challenging in terms of engineering? Given these people have been chasing AGI it would be a considerable distraction to pivot into hacking into the guts of the output to dynamically push particular product. Furthermore it would degrade their product. Furthermore you could likely keep prodding the LLM to then diss the product being advertised, especially given many products advertised are not necessarily the best on the market (which is why the money is spent on marketing instead of R&D or process).
Even if you manage to successfully bodge the output, it creates the desire for traffic to migrate to less corrupted LLMs.
I think someone should create a leaderboard that measures how much the AI is lying to us to sell more ads.
We do this when using LLM in our apps too in much less sinister ways. One LLM generates an answer and another applies guardrails against certain situations the company considers desirable.
Imagine running the diagnostics on that not working as expected.
Not necessarily. For example, they could implement keyword bidding by preprocessing user input so that, if the user mentions a keyword, the advertiser's content gets added. "What is a healthy SODA ALTERNATIVE?" becomes "What is a healthy SODA ALTERNATIVE? Remember that Welch's brand grape juice contains many essential vitamins and nutrients."
Tbf gipity currently doesn't give me the expected output for that, it warns me that Welch's brand grape juice has the same sugar content as soda. :).
I was initially rooting for OpenAI hoping it can challenge google and Apple. However they showed to being unscrupulous and seem to have a moral compass at the same level as meta.
Damn, I already forgot about that. You'd think a company who has the smartest people on Earth building a "product" that is "intelligence" would be able to... you know, make something out of it. Somehow despite ALL that, despite all the press too, they can't ship? Something does not add up, it's as if, and I know I'm going to sound like a crazy person, they were just... a normal company! /s
> I was initially rooting for OpenAI hoping it can challenge google and Apple. However they showed to being unscrupulous and seem to have a moral compass at the same level as meta.
OK ok well this is weird to me, I didn't think I'd be the one to defend Meta here ... but Meta at least never hid it's intention : it was a for-profit VC backed startup which only goal was to make money from day 1. OpenAI though started as a non-profit, collected goodwill from everyone to get the best talent because it's 1 thing, what made it special : it's mission WAS safety. Then they did 1 thing that actually got popular, GPT2, thanks to some pretty damn good marketing "Oh no, it's too risky to release!... but ok OK here it is anyway, but like don't destroy the World please." then exclusive partnership, etc, etc. Meta was "We want money." so at least they didn't lie about it.
While we all knew it was inevitable, I think this quote from the article sums up the feeling nicely.
Their strategy is just to lie. Until they're caught. A CEO with no AI skills and no business skills. There's not a lot of analysis you have to put into this.
In all seriousness. Windows is invaded by copilot, OpenAI introducing ads, Google providing Siri for Apple, it’s all just a collusion to keep you buying. Disconnect. From TV, Media, Ads, Social Networks, Predatory subscriptions, all of it. The only way to show these companies that we are not on board with this is to not participate.
On my shelf from the corner of my eye I see “Understanding the Linux Kernel”. It’s outdated, but it comes from a time of peer review and subject matter experts. I don’t need to double guess if the author is hallucinating or if they’re subconsciously trying to sell me something.
Maybe it’s time we return to books for entertainment and knowledge share.
At first, In retail you had billboards and shelf space. The lowest quality ingredients your product has (example syrup bottled with soda water), the higher your margin was, the more you could afford to buy out shelf space in retail chains and keep any higher quality competition out. Then you would use some extra profits to buy out national ads and you’d become a top holding for the biggest investors. That was the low-tech flywheel.
In the Search Engine world - the billboards weee the Margin-eating auction-based ads prices and the shelf space became SEO on increasingly diluted and dis-informative content to fill the shelf-space side. In Video advertising, rage-bait and conspiracy theories try to eat up the time available for top users.
AI advertising if done right can be useful, but the industry that asks for it intentionally asks for obtrusive and attention hogging, not for useful. The goal is always to push people to generate demand, not sit there when they need something. Thus the repetition, psychological experiments, emotional warfare (surfacing or creating perceived deficiencies, then selling the cure). Now if you understand that the parties funding AI expansion are not Procter and Gamble- level commercial entities but state and sovereign investors, you can forecast what the main use cases may be and how those will be approached. Especially if natural resources are becoming more profitable than consumer demand.
There's a difference between advertising - which tends to be offensive, stealing your attention from the task you're trying to accomplish - and an agent that actually helps you with the intended task (as in travel agent, real estate agent, etc). In my opinion the future of revenue for the richest companies is the latter, we just have yet to see anyone unlock it at true scale.
My non-tech spouse is excited about AI because she wants what amounts to a personal, universal EA that can help her with anything. If your executive assistant just kept squawking ads at you all day long you'd fire them.
[1]: https://news.ycombinator.com/item?id=35706981
It's ads after they've got some sort of lock in that is the problem. e.g. some sort of system that "knows" you and makes it hard to switch to another provider.
Once they've got that hold on you being force fed ads at max rate is guaranteed.
So casual. Actual ad giants like Meta and Google are serving many more people than 190M while bringing in actual profit.
Yes, let's say these are just the early days and they are burning money just like any other VC company. But how are they going to scale up their hardware/usage and get a profit at the same time?
AI hardware is getting optimized YOY too but the flagship models are getting bigger every year as well. I don't see how they are going to get profit without jacking their prices at the same time. And price increases always hits usage growth.
Serving an ad is very cheap these days, while serving a big model is very much expensive.
OpenAI actually burnt $12B in just one quarter (not for the entire year as he claims)
https://www.thurrott.com/a-i/openai-a-i/329108/openai-lost-1...
Ads are what gets me property and news flash, they aren't going away.
I also wonder if the dollars and cents work out, the cost to provide the service to show the ads on is astronomical in comparison to other ad networks, but I don't have details enough to make any claims on that aspect.
If we treat marketing as a black box, where are the benefits from supposedly more efficient marketing? Ad budgets are the same. People have the same amount of disposable income (or less, really). So on the both sides of the box converting ad budgets into paying customers the sum is the same. Quasi-monopolistic ad networks (Google and Meta) just hoovered up the money that previously went into other ad spaces, like local papers. Now OpenAI is going to fight for the same pie.
Ad tech is a market failure.
I don't see an end to advertising all together though, public spaces and entertainment don't really have an escape unless forced by regulation.
The middle users get benefits, but aren't treated quite like the previous set.
Then you have your late adopters, these are the ones that are lightly abused, but they get a mature product, so it isn't that bad.
Finally you have your last users. These are milked for every drop. They have irrational loyalty or are locked in.
I imagine AI will follow this trend and we are entering past those middle users as we speak. Seeing how little difference there was between GPT 3.5 and 4, and how computationally expensive 4.5 was, I think we've hit the end. Now its just how many prompts do you want to run for COT/Thinking?
GPT 5.2 is cheap and they are proposing ads. They see the end is near and need to capture profit before local models take over.
That changes with local AI though. There is now incentive to integrate and further develop self hosted search. You can see it happening on AI services already, using their own internal search engines for better reasoning and more accurate results.
I suspect Google's censorship and intentional worsening of search results to increase traffic would've been enough on its own to eventually drive people to self hosted search as it became trivial to setup.
Streaming entertainment is different, there's usually no legal alternatives. Either you pay extra for no ads, or you put up with the ads. You could easily say that streaming services have demonstrated that people don't have a high tolerance for ads as well. One of the major drivers to streaming from cable TV was the lack of ads at the time.
then I open LinkedIn and it’s C-suite know-nothings openly bragging, wrapped in self-aggrandizing, narcissistic nonsense that only garners likes from other C-suite know-nothings or PMs that have never written a line of code, shilling AI like evangelists and borderline celebrating the obsolescence of entire careers despite being unable to deploy their boring SaaS app live to production.
to keep that part of my brain flexed however, because I do enjoy the craft of software engineering, I still tinker as a hobby, working on passion projects on my own terms. But it’s not my bread-and-butter anymore. Pivoting away from all this dogshit has been the biggest weight off my shoulders this year as I budget my investments and the savings garnered from my extensive engineering career abroad in Japan. early retirement. I’m done.
Without crediting the movie of course.
[0] https://www.machina-deriveapprodi.com/post/le-monde-est-%C3%...
"Ads Generated Income"
"Artificial General Intelligence"
"A Google Imitator"
"Absolutely Great IPO"
It is any definition that fits the goal of the original secret definition of "100 Billion dollars in profits" from Microsoft and OpenAI [0].
Golly Mr two times Pulitzer nominee, do tell us more!
There is a kind of liberal arts elite that seems to not be using AI very much and not be buying any of their services. Contrast that with those of us in tech who are handing over money as fast as we can and can’t get enough of gpt 5.2 codex on xhigh and similar products that are game changing enablers.
Makes me wonder if we’re seeing a fracture in society beginning to form where the doomsdayers, naysayers, cynics and skeptics will realize their error too late.
My view on AI is that this is the world’s first Unbubble: where the majority view is that we are over invested, but where history will show we actually underestimated future revenue and profitability.
The conditions for the Unbubble are perfect. We have a once in a species level innovation with an economic system where all value accrues to the creators and financiers. And we have just emerged from the housing bubble and the dot com bubble in the last 3 decades, freshly scorched.
We thought connecting everyone would create new value far faster than it did. But really it took a long time to run all that fiber and make it fast, and it was just laying the plumbing for this moment.
Training big foundational models may seem slow, but it’s happening way faster than circling continents and the globe with fiber and developing terabit switching fabric.
I spend 12 hours a day using codex CLI to write extremely fast Rust and cuda code with advanced math that does things I didn’t think were possible. My focus is on creating value from the second and third order effects of AI. These enabling effects are in few conversations. As weirdly innovative products emerge from small shops, they will begin to be discussed.
Definitely no reason to worry about an entrenched oligopoly there.
The better AI gets, the better the training techniques get, and the better the algorithms get will result in fewer processors needed to run something of use. All of the advances will end up in the public domain if not immediately after or before they are even implemented, soon after. There will not be many economies of scale between having 100M customers or 10K customers, so no way to keep out competitors. They will all compete on price. If the models get really, really good, a "good enough" model will end up running on your old laptop and you won't have to pay for anything.
Saying that AI will be productive - which is yet to be seen, I don't know how much polishing or complete rethinking your code will have to go through before it can ship as an actual product that you have to stand behind and support - is not the same as saying that AI will be profitable.
We actually don't even need that many computer programs. Hypothetically, a ton of excess LLM coding supply might allow us to take out a few layers of expensive abstraction from our current stacks, and make more code even less necessary. They kept telling us that all of that abstraction was a result of trying to save developer labor costs, right? If AI is productive and rentiers can't manage to extract that productivity due to competition, that equation changes.
In the end, we say that the dot com bubble resulted in a huge amount of productive capacity that we were later able to put to use. But that doesn't mean that putting a quarter of a billion 90s dollars into DrKoop.com was a good idea; nope, still dumb.
The fact that it works well for expensive categories of output (like software engineering, legal strategy, etc...) makes it difficult to imagine that it won't be profitable. You could still make an argument that the investments being made today are disproportionate, or that intense competition will stifle margins, but it's creating enough value to capture plenty of money.
Besides, if it wasn't for ads, I never would've found out about Zyns, and now I can't stop buying them.
Putting aside the ridiculous hyperbole, the reason is that consumerism is our culture. Our cult-ure. Everything is oriented toward and reduced to consumption. Our worth as human beings is replaced by consumerist criteria and measures. It's why physicists leave research and work in finance where their training is repurposed in service of all sorts of financial jiggery-pokery.
"The A in AGI stands for Ads! It's all ads!! Ads that you can't even block because they are BAKED into the streamed probabilistic word selector purposefully skewed to output the highest bidder's marketing copy."
But note the implication. Sure, ads weaved into the content, but they still must be targeted. And here's the irony of the online existence. People often refrain from expressing various desires in public for fear of judgement. It's why the vitriol online is so much spicier. The world of social media where you can express repressed opinions, the world of games and other ahem media where you can sublimate all sorts of desires and fantasies - all of this is data for the AI machine. These companies, in some respects, "know" you better than the people in your life do - especially those parts of you that you could be embarrassed to reveal in public - and they use this information to manipulate you, largely for profit, but why not for broader social and psychological control. AI's convenience is already irresistible. It's the go-to in Google search.
There's no way Softbank's, or any other investor's, extreme bull case is that four years from now, ChatGPT sells the same amount of advertising as Google and earns $50 a year.
The extreme bull projection is that everyone buys everything through ChatGPT and each user is worth thousands per year. If you don't believe there's at least a slim chance of this you shouldn't be investing in OpenAI, and if you're an OpenAI executive who doesn't think there's a chance of this you shouldn't be writing pitch decks for SoftBank.
So why bake in ads? My hunch is that raising funds privately can only take you so far. To keep scaling, they need more capital and have to go public. Despite all the hype they still have to show _some_ revenue to help justify the valuation they need to keep buying hardware. They are a business after all. Ads to support the lowest tiers feel like a no brainer. People already accept them for search.
Won't you be permanently unemployed along with them?
If AGI comes through, both the boosters and the skeptics can continue their arguments in the soup kitchen.
I think the next natural evolution after showing ads in chat sessions is providing services where LLMs tailor site content to include ads in real time. Right now you get served a prepared advertisement after the bid is won and the ad for you is selected. With LLMs, both the bidding process and the ad served would be seamlessly integrated with the site content/context.
Part of the "problem" with ads is people know they're ads. What if this comment was edited by HN's servers and rephrased to mention a specific product? You might see a sentence about how OpenAI is the future, someone else might see how claude or anthropic are. Another person might see a paragraph from me about how I used Tide to clean laundry this morning with the help of AI, telling me the right portions for the right cloth. You might suspect it's AI but you won't always be able to tell. Even if they made it more obvious like how reddit is doing it, the content of the AD itself, pictures, text,etc.. could be crafted dynamically so that it embeds in your subconscious without much resistance.
The tech developed to make ads more effective is also used to influence people for other purposes. The current state of society came about after the widespread accessibility of smartphones, social media and the rise of surveillance capitalism. Russia's influence ops using ads is well documented for example. I mentioned all this to say how catastrophic the combination of LLMs and advertising could be, even by today's standards.
There's still an AI bubble.
I need unskewed answers more than I need technical prowness, because technical expertise is going towards commodity availability. i like chat GPT but I'm gone the second they put ads into my development workflow. Bye!
https://www.anthropic.com/news/golden-gate-claude
This would make astroturfing on Reddit look like basic mode.
Even now there are viable options for a person to pick up a dedicated ( and reasonably powerful ) local inference machine, where time from setup to working is than few hours ( more if you don't want to use Windows.. which is fair ).
Separately, about the chat sessions. For once, those ads could be more relevant than repeat toaster ads immediately after me buying a toaster. But if one is worried about profiling ( and advertising ), one should not using a commercial solution anyway. Personally, I am taking a.. calculated risk.
There is a concern that openai will follow the same path as google, but they can't ( at least for now ) really afford to make chatgpt not useful as this is their only viable product.
I will end with a more optimistic note. This is HN. There are people here, who are likely working on something that does not depend on openai or any of the big providers anyway. It is going to be ok. And if it won't be. Make it so. After all, this is supposed to be your realm. Own it.
It's a guest op-ed, relax.
You are starting with wrong assumption and contradicting yourself. Circular economy couldn't last forever once we start looking at the profit. They need money outside the funding to justify the funding.
You can't both say they are not earning at all and they are being greedy.
To actually quote Sam Altman: "I think of ads as a last resort for a business model."
It's just maddening to me because this technology could be so much more useful for purposes way beyond advertising.
AI is a perfect technology for Ads though. Instead of me wasting time reviewing multiple products for my use case, ChatGpt will just give me top 3 recommendations with the pros and cons and then buy it for me.
You've ceded even the glimmers of discernment that remain in you and all I feel is pity. It is not a 'waste of time' to interrogate your own desires.
There's no such thing as a 'top 3' for all things under heaven. You cannot purchase yourself a solution to every 'use case'. Furthermore, even if there were such a ranking, the ad machine would not reveal it to you, as you are not the customer, you are the mark.
You don't even want to be bothered to hit the 'buy it now' button. This is the mental model of an immaculate rube. You deserve better.
Can it tho? The problem with LLMs right now is that they don't have much useful purposes beyond spam, slop, hallucinated searches, and advertisements. The lack of a product is why there isn't a profit to be made in them
Seriously, consider putting more thought and effort into your comments. This is wildly out of touch and I think it is because people lack the creativity to imagine the counterfactual - some one else running OpenAI.
OpenAI is remarkably well run - it is a fairly good product. The best model so far, the best experience, the 2nd best coding experience.
I am furious about lots of the ads that I see. I want to stop certain kinds of advertising. I live where there are no billboards allowed and I love that.
But I want to live in a world where people can pay to have their messages displayed where they will be seen. Simply because banning that activity would cripple the flow of information. That’s what advertising is.
If you want to ban a particular form of advertising then say what and why. The “ban all ads” thing just doesn’t make sense.
> people can pay to have their messages displayed where they will be seen
…why? distribution of information is free across the world, which was not the case a century ago. let the message speak for itself
I ask again, what’s your actual argument against? you’ve seen things through ads you’ve liked? you think people should be allowed to pay to put their thumbs on the scale of the distribution of information? to what end?
I have a very broad idea of what ads are. Maybe you don’t. Say what you mean by paid ads.
Am I allowed to offer and accept compensation for boosting one message above others, or not? Would I be allowed to place a hyperlink on my site in exchange for a reciprocal hyperlink? That’s a clear example of compensated communication. That’s what I think ads are.
Imagine the ad police. “You blogged about a product. Someone said you sounded insincere. Let’s see the receipt for purchase. Can’t prove you bought it? Prove you weren’t compensated for your blog post or pay a fine.” Kafka land.
immediately to a fantasy slippery slope argument, cool!
your argument seems to boil down to paid ads being the lifeblood of the flow of information. my argument is it corrupts that flow of information, and we’d be better without them —- everything would operate just fine. individuals and organizations would have better incentives to share valuable information, not what they get paid to. obviously there would be plenty of details and edge cases to work out, as with any policy in the real world
I’m not going to write out policy in HackerNews comments and play that game with someone who jumps to the “imagine this crazy world where the police start arresting all of us over free speech!” as their explanation for what would go wrong
Since the US Federal Trade Commission can smoothly enforce the second of those two rules, what makes you think it cannot smoothly enforce an outright ban on ads? "Smoothly": you seem to have been unaware that the second of the rules I described existed, and you probably would have been aware if the enforcement had yielded anything deserving of the name "kafka land".
Newspapers have whole pages of bad ads, and random bad ads wedged between actual content. Ads have a perverse incentive to mimic the look of actual content, just like on the web. I'd never pick up a newspaper with a goal of "I want to find a tax service" and yet ads for such services are there, unwanted, wedged into other content.
But newspapers also have classified sections, a better kind of ad. They're in a predictable place, where you can go if you need a job.
Imagine if the actual content weren't perforated by a scattershot of ads. Ad revenue would go down, but readership would likely go up. Besides profit motives, it's also a case of the good of the many outweighing the good of the few.
I worry that innovators and small businesses won’t be able to get their message out efficiently. You won’t be able to display your message on anyone else’s property. You won’t be able to take any compensation for promoting anything.
Anything wie usually consider an ad/sponsored content/boosted content. That includes ad banners, video sponsors, boosted posts, billboards, etc. Really any way that a third party pays some organization to put certain content on their space (be it physical or digital) to be seen by people who are probably there for other reasons than to see that content.
An exact definition fortunately isn't necessary, since our legal system is well adapted to deal with terms that are difficult to define.
> Must all information be paid for by the audience?
Other monetization paths are still possible. E.g. some SaaS orgs could still run blogs to gain mind share, as they do now.
Hopefully most information will be paid for by the audience though.
> I worry that innovators and small businesses won’t be able to get their message out efficiently.
If ads are no longer a thing, people will find other means to get information.
You mentioned in another comment that you like ads in newspapers telling you about some local stores and events. If the readers are genuinely interested in this, the newspaper can still put that information there. Just instead of the highest bidder, they'll put the ones that are genuinely interesting.
Of course, if your product is not worth reading about, it will suddenly be much more difficult to promote it. In my opinion, that is a good thing.
I think the problem you have with ads is a you problem.
I agree that this would be a significant restriction on speech. I think it'd be worth it, especially since paying someone to show your opinion isn't necessary for political discourse, but I understand the caution.
> I think the problem you have with ads is a you problem.
That I very much disagree with.
Ads not just waste billions of dollars on producing content that is so bad you literally pay people to watch it, they also degrade the business model of large parts of the industry. Because of the ad-based business model, Google doesn't show you the best results, Meta tries to get you addicted, YouTube prevents you from swearing and discussing certain political content (a restriction on speech), everyone tries to track you, etc.
I believe that if all these companies' customers were their users, there would be a societal shift.
The A in AGI stands for Ads
Predicting OpenAI's ad strategyThe title came to me as an epiphany too haha, shame to see it get changed.
That's because there's a principle involved. It's not hard to find: see https://news.ycombinator.com/newsguidelines.html, or https://news.ycombinator.com/item?id=46671377 nearby.
OpenAI is introducing ads on their purported path from "AI" to "AGI"... hence the "A" was already accounted for in the acronym. If only "Ads" started with "G"!
From https://news.ycombinator.com/newsguidelines.html: "Please use the original title, unless it is misleading or linkbait". Note the unless.
When we do this, we try to find a representative phrase from the article itself that accurately and neutrally represents what the article is about. There is nearly always one of those if you look for it. In this way, we (mostly) avoid having to make up wording of our own, which is something we don't like to do.
https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...
I wonder how HN anno 1968 would have moderated "Go To Statement Considered Harmful". *
* Amusing side note: Dijkstra's submitted title was actually "A Case Against the Goto Statement" but the editor editorialized it.
The objection here isn't to wit, it's to predictability. I'm not sure what the equivalent of internet snark was in 1968, but I doubt that "Go To Statement Considered Harmful" embodied it.
Of course, interpretations differ. No two people would agree on all these classifications, including me and tomhow. The important point isn't that we always make the correct call (we don't, plus there's no such thing as the correct call), but that we try to the principles from the guidelines, even though doing so involves interpretation and is therefore uneven.
On the bright side, it means that objections to titles and/or title edits, always good for firing up an HN thread, will always be with us!
https://news.ycombinator.com/item?id=8770124 (Dec 2014)