[TrianguloY]

Ideally ads should be like a recommendation system. If you want to buy a hard drive, the 'best' ads should be of hard drives that suits you, either because of their price, their features, or whatever you personally want. In order to build this profile, the recommendation system should know you, your data. Lots of recommendation systems exists, and those who's first objective are users seems the best ones (preferably if they are offline so your data is not shared).

But what about an ad? By definition it's something someone pays to be shown. When Google shows you targeted ads, they aren't showing you the best ads for you, they are showing you the best ads of those companies who paid for it (and probably the price influences its ranking). The user is not the first objective.

In an ideal world, personalized ads wouldn't be called ads, and would be like having an assistant. You want a hard drive? Here is the best hard drive for you, or a list if you prefer, and then you buy it and the company earns your money.

In the current world, that's far from the reality.

[srg0]

Hard drive is a great example. And it shows that advertising is completely backwards.

It starts with a seller, who wants to sell this hard drive, no matter what, and then pays to find a buyer. Targeted advertising is a false promise that if the date of birth, sexual orientation, political preference, and the favourite color of everyone are known, the advertiser can somehow find buyers for a smaller fee. Indeed, most of the hard drive buyers are older than 20 and like women. There's a correlation to exploit.

This business model is completely backwards in the digital age. What is possible now, instead of profiling people, Google/Facebook/Amazon could have been profiling products, sellers, their supply chains, organize information, build knowledge graph, and rather than profit from a seller pushing their message, they could have took a commission for finding just the right hard drive for everyone, no matter what color they like.

The point is profiling sellers and their products is a harder problem than scamming and nudging unsuspecting users. We all know how great Amazon search is. Targeted advertising on the other hand is just the same as print era advertising but cheaper.

[jka]

If someone knows that they want a hard drive with at least 100GB of storage, is it really necessary to construct advertising system infrastructure that can (attempt to) learn the users' requirements and needs (the same way an assistant might do)?

Isn't it simpler, cheaper and more respectful of people's own decision-making faculties to provide them with a way to retrieve a list of hard drives that match their requirements, and make their own choice?

My guess is that in 80%+ of situations, simple measurements and properties of products (with price, reliability, and perhaps other ranking factors to tie-break) are all that consumers want.

[hilbert42]

"In an ideal world, personalized ads wouldn't be called ads, and would be like having an assistant."

Right, I'm truly stretched to the limit to think of any ad that I've seen on the web that's actually been truly informative to me. For that I have to go back to ads in specialized technical magazines that used to list detailed specifications for the products they were selling, today web advertisers simply don't do that - all we get is hyped-up uninformative and often misleading crap that interferes with our access to the site's web pages.

Similarly, many, many websites that are selling stuff are cluttered, disorganized and otherwise hard to navigate (as well as being very, very slow). Of course, this is a deliberate marketing ploy but the effect on me is that I'm straight out of there as fast as possible.

Internet advertising has always been an unmitigated mess and unlike those old magazine ads, almost without exception, it's gotten in the way of the main presentation to the utter annoyance of website visitors. It's little wonder so many use ad blockers.

For years, my solution has been to block ads as well as all the accompanying JavaScript of which much is devoted to spying on the user.

Eliminating both ads and JS not only speeds up websites enormously but also it eliminates all those hesitant pauses and other jerky/delay-like responses that make browsing such a damn pain.

I'm now so used to 'clean' browsing that I cannot ever imagine myself returning to the standard defaults - ads and JS. For those who ask 'how do you do this or that without JS?' I'd just say this, there are millions of sites on the web, if one blocks me then I instantly move on (one good point about the web is that most of the information on it is paralleled across multiple sites and there are many sites that are more user friendly than others). That said, there are rare situations where I still have to use JS. When necessary I just toggle it on and off, the default being off.

As far as I'm concerned the web has been ruined by invasive advertising and we need a new paradigm to fix it. There are many options for that but it's too big a subject to discuss here.

[godelski]

Interestingly a problem we as nerds often have is thinking that if we have the best product or best skills that it will win out. In reality the best often doesn't win because there are many other factors (and we can see a history of superior technologies being beat out). That assistant that acts as an actual recommendation system and removes the human element could bring us closer to that meritocracy based product world, which seems beneficial to everyone. But determining what is best is a very difficult problem to begin with.

[mattlutze]

Good long-term relationship building can be as compelling a feature as what the product actually does. It's short-sighted to discount the value of enjoying having bought what you buy.

[andreygrehov]

> But what about an ad? By definition it's something someone pays to be shown. When Google shows you targeted ads, they aren't showing you the best ads for you, they are showing you the best ads of those companies who paid for it (and probably the price influences its ranking). The user is not the first objective.

I mean, that is an assumption, unless you work at Google and is currently sharing the algorithm with us. Most likely, they show you the best ad they can based on targeting criterias and the information they know about you. It could very well be possible that Google not always picks the highest bidder. The winning score is most likely calculated based on a combination of signals, eg bidding price, user's preferences, user's buying history, etc.

[TrianguloY]

Of course is an assumption (should have used the word 'probably', my mistake). My point was that once the bidding price enters the equation it no longer can be considered a recommendation system, even if it's only one factor of multiple (as I suppose).