> In a recommender engine, the best recommendation for your quality of life is at the top. In an ad engine… that list gets resorted
A recommender engine in principle could avoid recommending things that are unhealthy. Advertising has no such guardrails. It recommends things that reflect someone else's economic incentives, and not what's actually best for me.
Exactly - there are many ads for products which are unhealthy, dangerous, or just very low quality. All of us have probably bought some in each category and regretted it.
I'm not trying to defend ads in general, but one problem that a recommender engine certainly suffers from is how to deal with a new entity. How do you make a new entrant gets into that list even though, while they may provide value, there's currently no evidence of that are far as the recommender is concerned?
While often ignored, item-to-item similarity recommenders work really well for new items and have been the basis for Amazon's and YouTube's early recommendation algorithms
I think you may be misunderstanding the parent (I may also be misunderstanding though) - they're saying that with a new user (not a new product), you have no information for providing recommendations to them; you have no idea what they're interested in.
I was talking about new products/movies/whatever, not users. Sorry, "new entity" was too vague in retrospect.
My point was around the utility of marketing, at least in theory. If I release a new brand of baked beans, how would a recommender based on past sales (and combinations of sales) know to recommend it to anyone. Marketing allows it to go up the list (unfairly, according to the post I was replying to).
Ah, okay. Then yeah, the original reply is valid. I assumed you meant a new user to a site, since you can't recommend anything then no matter what you've built (without user tracking that can maybe have built up some kind of profile from elsewhere).
One (partial) way of looking at it is that modern advertising is a recommendation engine, but just reworked to be adversarial.
Because consumers know that advertising is an adversarial recommendation engine, restricting ads to a single place (like Google searches) isn't really feasible for the entire industry -- comparatively fewer people would visit that recommendation list because a large portion of them know that the recommendation list is adversarial.
So, modern advertisers are left with two options:
1. Insert their recommendation engine into other activities that users want to engage with (ads inside of or alongside other content).
2. Try to convince users that a portal for doing searches on their recommendation engine is actually a non-adversarial recommendation engine (recommendation engines where it's not always immediately obvious that advertisers can pay to influence results).
Closely related to 2 is in pairing their adversarial recommendation engine with a non-adversarial engine making it somewhat adversarial, but not so adversarial that users are willing to leave. For example, Google Maps does legitimately tell you about businesses nearby. It also combines ads into that system that are adversarial and that try to get you to buy things against your own best interest. But it's not so adversarial that the average user views it as useless, it still sort of works as a search engine. It's still food, it's just a little bit more poisonous than normal: enough to harm you a little bit, but not enough to kill you or to prompt you to stop eating it.
In a perfect market economy, those two lists are equal.
Thats because the company who produces the product you are most interested in can afford to spend the most on ads. Products you aren't very interested in, either because they cost too much, or aren't up to your standards, you are unlikely to buy, and therefore those manufacturers won't be able to afford to get to the top ad spot.
In reality, this doesn't work because the market is far from perfect, and the amount of margin a company has on a product to spend on advertising widely varies.
"Perfect market" is a technical term meaning one where "buyers and sellers are so numerous and well informed that monopoly is absent and market prices cannot be manipulated". A perfect market would not help this problem.
The notion that perfect realizations of Econ 101 concepts would produce outcomes everybody likes it both wrong and quasi-religious. It's like saying, "The perfect knife would never cut a human." Nope.
If anything, the whole purpose of advertising is to create market distortions. Look at Coke vs Pepsi, for example. Billions are spent yearly to manipulate consumers to make choices that differ from whatever the natural marketplace outcome would be.
What's more, even in a perfect market as you describe, advertising will only ever represent the subset of the solution space for any given problem that involves spending money. With how powerful a voice advertising has, we end up training people to deal with problems by buying rather than doing/learning/making.
In a perfect market, the top ad would be for the best treadmill. You'd never see an ad for just going outside for a run.
In a totally perfect market, the park charges runners a few cents to go for a run, and spends some of that revenue on advertising the park to runners, and then they would outcompete treadmill ads (for some users).
The reason you don't see this is because free parks are themselves a market distortion.
In a recommender engine, the best recommendation for your quality of life is at the top.
In an ad engine… that list gets resorted
Advertising’s true cost is opportunity cost. It’s the lost opportunity of knowing the actual best option for your quality of life.