Here is a thing I think about a lot:

People like ads.

You know "brand Twitter" and how people like talking about the cheeky thing that Burger King said? That's an ad.

The Super Bowl? People watch for the ads.

Magazines: people collect pretty ad series, like the old Absolute Vodka ones.

Podcasts? We all loved saying "mail kimp" like idiots for a couple months because of a popular ad.

People like ads! …Except on the web.

No one has ever liked a banner ad. Search ads people sometimes like, but no one has ever liked a banner ad. So, they kept adding more banner ads to try to make it up in volume and then they added a shit ton of tracking because they could, but guess what, banner ads still suck and no one likes them.

So, maybe stop trying to make fetch happen? If banner ads haven't been successful for 20 years, maybe they aren't going to suddenly become successful and it's time to stop trying.

Facebook ads are basically extremely successful banner ads, both for FB and advertisers. There's a reason 90% of new ad spend is split between Goog and FB.

Honest question: How do shareholders know FB doesn't lie about their metrics? Maybe if it leaked it would be fraud, but what stops them from saying some psuedo-b.s? Like "our ad score went from 0.4 schmeckle to 8.6 schmeckle increasing ad yield by over 78% this quarter, here is a 500 page document about it" and if I were an investor I wouldn't have time at all to figure out what a schmeckle means, and I certainly wouldn't care what a schmeckle means or even if it's all an elaborate ruse if my position is improving.

FB shareholders do not care about metrics, they care about ad revenue. The shareholders of companies spending ad money do not care about metrics, they care about revenue. People who buy those ads care about metrics, but while they cannot see inside the FB machine and need to trust them somewhat regarding views and reach they have absolute visibility on the results of that ad spend and where those visitors are coming from. These last two are not things that Facebook could fake for very long...

> FB shareholders do not care about metrics, they care about ad revenue.

Agree. And I guess they have reason to be happy for now.

> The shareholders of companies spending ad money do not care about metrics, they care about revenue.

Agree. And they have reason to be sceptical: a lot of Facebooks growth in ad sales seems to have come from companies who suddenly had to pay to reach their fan base.

> People who buy those ads care about metrics, but while they cannot see inside the FB machine and need to trust them somewhat regarding views and reach they have absolute visibility on the results of that ad spend and where those visitors are coming from.

Another explanation: pay per view ads are massively overvalued but for some reason or another management isn't aware of it or doesn't care yet.

Most FB advertisers measure sales from Facebook directly using 1st or 3rd party cookies. So you can see that you spent 100k and made 200k.

While there's plenty of grey area in measuring "engagement", eventually ad spend either delivers ROI or it doesn't. That's not something you can fake.

Successful advertising campaigns generate revenue. Can't fake that.

Also there are companies that will track your FB ads and tell you if FB is lying.

You can't talk about the past and just forget all the context. The scale in people, services, and speed of connectivity make a massive difference. It's an entirely different world.

Also classifieds were a major source of revenue for many publications and it was destroyed by listing sites like Craigslist. Nobody talks about it but that was the first big blow to the old model.

Every article about newspaper sales talks about it. That was ~15 years ago though.

Surely you could use NLP to extract context tokens and ad labels for ad submitters to create an ad network for this scale.

I thought that was mainly paid subscriptions and advertising?

Ad-only publications were always very low quality.

The big moneymaker for papers was and is advertisements:

https://www.journalism.org/2014/03/26/revenue-sources-a-heav...

This does not work as well. If you take any machine learning model for ads, "content" related features are way less predictive than "user" related features. ie who you are matters more than what you're currently reading.

[Disclosure: I've built a lot of these models in the past]

Do you mean that advertising won't generate the maximum amount of revenue possible unless it violates users' privacy?

Correct. The crux of the issue in the economics of the web is that users privacy value is not explicitly part of the advertiser-publisher transaction. Advertisers want more data because it works. Publishers are ok with that as long as it doesn't bother their users too much. But it's not explicitly priced in the transaction, and users have no say in the transaction at all.

I cannot verify your models.

But I can verify that I think Google gets my interests wrong more than 9 out of 10 times to the point of repeatingly insulting me and my family, and they are supposed to be very smart so I guess they to have equally smart models.

If they had just stuck to something relevant to the pages I visit they'd probably hit more than 50% of the time. If I visit a video about fixing stuff I might be interested in tools, paint or parts. If I visit a local news source around lunch I'd probably enjoy ideas for dinner from local shops.

That said I probably wouldn't buy much more but at least they wouldn't insult me and my family.

I can throw around anecdotes as well. I can verify that I think the majority of ads on content websites that I see and find interesting often have nothing to do with the article that I'm reading.

But in the absence of data, I'm leaning towards trusting that a market-leading company incentivized and dedicated to optimizing targeting models has a stronger research on targeting the majority of users than vague internet claims of bad targeting. But what do I know.

Really? To me it seems that "user-predictive" modeling just mean "show ads for sites user visited recently". E.g. if I buy a blender on Amazon (and maybe check a few comparison sites or something before that) I'm going to get blender-related ads. If that's the best advertisers can do, it's pretty weak sauce... Why would I need 7 blenders?

It seems to have worked better than well for New York Times: https://digiday.com/media/gumgumtest-new-york-times-gdpr-cut...

Do you think it's not repeatable?

> except for privacy-minded users

...that's like saying "high standards of medicine are bad except for hygiene-minded users." The fact that the user has other concerns in their life doesn't change the fact that it's still bad for them.

It might be good for supply (publishers) short term, but it's worse for demand (advertisers) and for ads quality since it removes personalization opportunities. An open-market like RTB where offer meets demand is more efficient (except for privacy-minded users which aren't even part of the transaction). It's also worse for supply long-term as it ties them more to Google and FB, or to a non-scalable contract negotiation process mad-men style with a handful of advertisers.

Targeted ads long predate tech companies. Not every publication gets printed with the same ads. Some metro areas get different ads than others. Same with television. The content itself may be the same nation wide, but local stations run their own ads to better target certain populations. You're providing the publication with your name and address - do you really think they aren't using that to deliver more fine grained ads?

That's still a huge difference. Previously ads would be people who read x, people who go to y, here's a shoe that you might like because people who read wsj wear loafers, here's a hair product you like because people who read seventeen magazine are generally seventeen. The ad wouldn't know anything about you, just what the readers of where the ad is placed might be into based on past advertising decisions, and the content of whatever they are reading.

Now the game has been turned. Advertisers don't need to target sports illustrated to sell gatorade. They can see you checked into LA Fitness and most of the people who checked into this particular LA fitness also happened to spend 30s longer looking at a tester gatorade ad, so there is a slim possibility that you too will look at a gatorade ad for 30s and maybe subliminally engage with it next time you are presented with a gatorade in a cooler at the grocery store. The classic is you looking at a product on amazon and seeing it appear all over the web. It only gets more insidious the deeper your digital footprint goes, the more data they have on you, the more correlations they are free to make which would have been impossible back when advertisers were catering to audiences, not individuals. Sprinkle in exploiting modern psychology for profit, and it's disturbing.

Yes, and it's shitty for consumers and publishers as well. Before if I was interested in a niche product, there would be publications that were supported by ads from the niche industry. Now, they can figure out that I am interested in fountain pens from my browsing history, and instead of paying someone to write fountain pen content for me to read, they put the ads on any random website that I happen to visit. This means that now niche publishing is unviable because it doesn't have a high level of traffic and it also can't deliver more highly targeted ads than e.g. Facebook.

The clearest example of this is the collapse of local news.

Yes. Do you know how commercial printing works? The newspaper is not targeting individuals.

They could be if they wanted to. My address is printed on the cover of some of my magazines - so the capability of person-specific printing exists it's just a matter of extending this to the ads too.

Realistically, I wouldn't expect targeting more granular than a zip code. But that's still pretty decent as far as advertisement targeting goes. Definitely better than broadcast radio, as one example.

Lol, you got me, they can put a sticker on magazines, so there's no difference between print and online advertising.

You mean the model that is no longer viable and is forcing increasing numbers of publications into paywalls to remain solvent?

To be honest I prefer paywalls. If I only read a couple of articles randomly a month from the publication I don't see them at all. If I read any more, I don't mind paying $1 a week to support good content. In the old days you had to either pay a newspaper subscription or buy a paper from a newsstand if you wanted to walk home with it, so it's not all that much different.

To play devil's advocate - how do you determine what constitutes research to figure out what your "target audience" actually is, versus "privacy violation/GDPR no-no"?

I can imagine say, readers of Golf Digest are probably interested in golf. That's about the extent of the research that can be done by a traditional magazine. But there's an entire ad industry (for better or worse) to minimize 'wasted' impressions, and there's a large chasm between "male 25-35" and "specifically John Doe who looked at #9 clubs yesterday but hesitated to buy them, maybe we can win him over". Clearly the later example is prohibited in the spirit of the GDPR, but it's broad enough that it kills the former as well (and puts the extent of information available to the industry back to "golf digest readers play golf").

Not that I am particularly sympathetic to advertising agencies and the machinations of what online ad-tech has become, but I am sensitive to legislation that puts entire industries (I.e. real people's jobs) under a coal fire of "maybe we as a company literally have no viable path forward".

Well, say that I want to know if you are interested in golf. Do you prefer me asking if you can fill in a questionnaire, or should I just pop where you live and take pictures from inside your house?

You really don't have to play devil's advocate, one of them is illegal.

If a company has no viable way forward, and has to resort to what the rest of us deemed illegal, it should stop existing. Not all companies should survive at any cost, that's not utilitiatian or desirable.

Certainly, spying on someone's domicile is illegal. But it wasn't really the analogy I was going for.

I'm going to be purposefully obtuse and set up a contrived hypothetical, but to me it's really more like: you came to my store today, but I saw you at the golf shop yesterday. Maybe I watched you closely to learn exactly what you were thinking about buying. I happen to also sell golf things and maybe I want to peddle them to you. So I really go all in on my sales pitch and single you out while you're shopping. That's legal in meatspace. What makes "the same thing, but do it on a computer" illegal and unethical? I think it's a real important question for legislators to answer. Is it the scale? The fact that it can be automated? That we're really not actually comfortable with the meatspace example but didn't have a good way to ban it before?

Just food for thought. I'm mostly neutral on the GDPR, I appreciate the intention but I think we as a society need to define consistent expectations about privacy.

Following me around all over town like you did yesterday is considered stalking in my books.

Google isn't following you, they're just in every store you walk into no matter what.

Imagine almost every store in your city installed something facial recognition (a la Amazon Go). This system, given to owners literally for free, was created by one company - Google - and allowed them to automatically see what products most people were looking at, what time they were doing so, and their demographic. As payment, Google also gets this data and is able to use it for their means.

Note: i'm just making this analogy. Google Analytics (probably) isn't being used for ad targeting.

I was careful to say only that I saw you at the other store rather than implying I followed everywhere, since that's a fair line to draw and I wanted to avoid that particular connotation. But even still, hiring a private investigator is not illegal - it's just observational.

What you describe is not what actually happens.

What actually happens is that your behavior (in store or not) is permanently observed all the time and ads are served whether you want it or not.

Imagine if in all shops, always you'd get jumped on by attendant trying to sell you things. And sometimes shop attendant would appear out of the blue.

What you're describing you can have today - just convince user to make account in your shop and give you permission to send offers. You can chase him any time you want.

> Imagine if in all shops, always you'd get jumped on by attendant trying to sell you things. And sometimes shop attendant would appear out of the blue.

As opposed to the half of them where it currently happens?

50% better than on internet and they do not have my data.

> Maybe I watched you closely to learn exactly what you were thinking about buying.

If creepy shopkeepers following you round was a thing I suspect it would be illegal already.

That's the precipice of the argument though, right? It's not illegal to observe interactions that you witness, or take notes about customers you saw shopping your competition.

You could even hire someone to wander around your competing stores and take lots of notes about the people they saw. "Red shirt guy shopping around for slim fit pants". Unlikely scenario, but legal to do and to use that information.

My whole thing is - what makes the scenarios different enough that one is prohibited and one is allowed? Certainly if we told folks they must forget their observations about others upon request, that would face some baseline rebuttal about their own autonomy.

The difference is that one is an actual problem and the other isn't. That's it. There isn't some deep legal theory driving lawmaking.

It wasn't necessary to make this behavior illegal before because it was impractical, but now that computers have made mass surveillance practical, it should be illegal.

> I am sensitive to legislation that puts entire industries (I.e. real people's jobs) under a coal fire of "maybe we as a company literally have no viable path forward".

Good that you're sensitive, but some jobs simply deserve to die. It happened in the past; behind leaded gasoline, or CFCs, or poisons sold as miracle drugs, were real people with real jobs. Advertising as an industry proved itself to have cancerous nature and grew to do incredible damage to people; it deserves to be burned down to the ground.