Mark Zuckerberg Speaks Down to Users and Misses the Point

I disagree, it's not a false dichotomy, it's a real one.

Most Internet content isn't nearly as "contextual" as a nature magazine. Probably 99% of the content I read won't help target ads much at all, because it's mostly NYT news articles and the like.

I'm probably in a tiny minority here on HN, but I accept that ads are currently necessary to fund companies (even though I wish it weren't true), and I genuinely prefer ads targeted towards my interests (I've gone so far as to customize my interests on Google to help it serve me more relevant ads -- most of them were accurate but there were a handful that weren't).

Targeting obviously isn't always perfect -- they still don't have the kink ironed out where I keep seeing ads for something on Amazon even after I bought the thing or a similar thing -- but it really does benefit users, if they've got to put up with ads in the first place.

I find it comical that Mark thinks targeted ads are even the controversial part. We have children gambling their parents money away, data landing in the hands of Russian psyops campaigns, political violence in the developing world being organised etc..

If targeted advertisement was the most concerning thing coming out of Facebook that'd be a great improvement. And yet they're not even able to get rid of 'friendly fraud'.

But what would they suggest Facebook do instead of tracking? What ads are they supposed to show me if they're not allowed to remember things about me? There's nothing analogous to "outdoor gear in a nature magazine" that they could do instead.

I just looked through my Facebook Ad Preferences as the article suggested, and there's nothing there that would shock me. Yes, they know that I have an older model iPhone because I've used it to access their website. They know I'm a frequent traveler because they remember what IPs I've accessed their website from. They know that I'm recently married, and which zip code I live in because I told them those things. Using my zip code and some public records that anyone could Google, they can make a good guess about my race and income level. None of this is very scary because they only know what my friends and I told them about me. They can't read my email or listen to my phone calls.

Speaking of which, does anyone remember Edward Snowden? Remember when we were (rightfully) angry about NSA surveillance, the thing that does read your email and listen to your phone calls whether you consent or not, and the thing that you can't opt out of by deleting an app on your phone? They never stopped spying on us when the outrage machine moved on, and all this Facebook stuff seems petty by comparison.

Wow, I actually find this to be much more impressive on Facebook's side than what I had imagined. So that means that Facebook has personalized ads categories so accurate that 3 out of 4 users themselves rate them as very accurately representing them 3. That's close to 2 billion users being precisely targeted, right? I find this to be extremely disconcerting.

After breaking up Facebook and punishing credit agencies whose job it was to secure data, the only path forward I see is to introduce new data formats for all the things we took for granted and push hard for the adoption of those replacements. (Facebook has all your pictures; well, do they have them in $NEW_KEYED_IMAGE_FORMAT that has your new revocable authorized keys injected by modern cameras? Facebook has your phone number; well, do they have a new 2nd-factor token that makes the phone number useless without the token? Facebook has your full name; too bad, the government has moved on to a new system that requires several new government-issued identity tokens that Facebook can’t have, making them unable to correlate your name with other aspects of your modern identity. And so forth.) Ruin it all. Make every single spec of data useless.

But the scary thought is that in this unlikely scenario (or in bankruptcy), what would happen to all the data they hoard? Sold to highest bidder maybe? Published under Creative Commons license?

#toomuchdatatofail?

"If". This is a conditional sentence. What if the condition is not met?

What if they want to avoid ads? What do they tell FB when asked "Ads or no ads, free or small fee, which would you prefer?"

I suspect the participants in Facebook's questioning were not asked that question. Or, if they were, FB is not going to use the results in their PR and lobbying efforts.

There is a slogan one might see on the FB homepage, something like:

"Facebook is free and always will be."

As if the user's primary concern would be that they would sometime have to pay for using FB. (Not privacy or the security of their personal information.)

The slogan raises more questions than it answers.

If FB is "free" forever, then who covers the operating costs of FB?

Where does Zuckerberg's fortune come from?

Who finances the next FB acquisition of a perceived competitor or their risky bets chasing "the next big thing"?

If it isn't you the user who is about to sign up or sign in, then how much does your opinion matter to FB?

Wouldn't it stand to reason that FB would be catering to whomever is paying the bills for FB? (Not you.)

FB is not free for them.

Of course investors and advertisers paying FB expect something in return for their payments. They are not paying into FB's accounts just to keep FB afloat. They are not paying FB simply to sustain a website and apps for people to share photos, video, messages. (And that is probably all users want or need. They certainly never told FB they wanted data collection, monitoring and manipulation of their online behaviour, nor advertisements.)

Assuming FB has covered its operating costs, then whatever it is that FB must then deliver -- i.e. not the webpage and apps to help people to communicate, but the data collection, monitoring, manipulation, etc. -- comes at a further cost, not necessarily a monetary one. The question is, "Who bears it?"

I think it is becoming clear that this "cost" is more than simply "attention" or "time". There are serious impacts FB's work is having on an enormous number of people. The jury is still out on whether this is a net positive or a net negative.

The other truth is that a nature magazine that relies on outdoor gear ads to operate is a conflict of interest.

I remember talking with a chief-editor of a 90's video game magazine: "It's better for us and our readers to see ads for shampoo, cars or food than video game ads. Less pressure.".

So you turn around to some web advertising platform and buy an ad, and you spend something like $30 dollars to buy however many website visits it takes to drive one sale, and after adding a generous fudge factor to take into account all the uncertainty and mis-attribution you expect you wind up with a profit. There's both cost-per-click and cost-per-thousand-impressions pricing, but both end up feeding into "buy $X of advertising, which flows through our sale funnel resulting in $Y of profit, and $X is much less than $Y" at the end of the day.

Why do these advertising platforms collect data? So that instead of a generic "show this ad to whoever visits the website", you can break viewers into different categories and sell those instead. If it takes 100 website visits from a generic user to drive a sale, maybe it takes 50 instead if they just performed a search for a mattress. Or if it takes 200 impressions to get a click-through, maybe it takes 100 if it's done from a specific geographic area. Or if you end up selling to a specific job title, the SaaS contract tends to be more expensive. Whatever it is, certain slices of traffic are more valuable to the ad buyer, so this ends up flowing through to the advertiser's return-on-ad-spend math, which flows into higher ad bids. And so the advertising platforms collect a lot of data, so that when a user can be shown an ad, the platform can figure out what targeting options allow them to make the most money.

If you follow the math again back the other way through, it implies that there's a pretty powerful constraint on how relevant advertising can be to you. Specifically, the pure relevancy metric will get skewed by the producer surplus for converting a sale to you: doubling how likely you are to convert is just as relevant to the business math as doubling how much money is made off your conversion. A system that maximizes relevance of ads will not maximize advertising platform income, since it would fail to prefer a five-figure programming job placement fee over a $10 tee-shirt, and the difference in conversion value flows through to a difference in advertising bids.

From a public policy perspective, I'm pretty pessimistic about preventing these privacy issues. There's simply too much of an incentive to section out users into specific cohorts and sell access to them to the highest bidder. In other words, privacy-violating targeting is such a big deal because targeting works - it reliably convinces people to take courses of action that result in more producer surplus in the hands of the producer of the good advertised. The only saving grace is that for certain products, it'd only change which company gets the producer surplus, rather than convincing you to push more producer surplus into corporate pockets.