Because it would not make sense otherwise, considering how much it must cost them to run something like WhatsApp with millions of users and not get anything in return.
I mean, what's the point of all the data they have from me plus the armies of well paid ML engineers they have on their payroll, when if I order a laptop from Amazon, I'll get bombarded with ads for laptops for the next month. Do advertisers imagine I go through laptops like toilet rolls or something? If this is peak ML, then those who fear an AI uprising can sleep soundly.
They target something like "people interested in computer hardware" and they get them, including those who don't need a laptop. It's bad but must be somewhat better than completely non-targeted ads, as it excludes people who have never expressed any interest (so, while having false positives at least it tries to save on false negatives). And, of course, there is opportunity for Facebook to tell how their superior ad platform is smarter than last year.
The rest (but maybe even more important than actual value advertisers are getting) is advertising of advertising - all those stories how this machine learning agile blockchain big data is our only future. And many people just believe it, and non-ironically talk how Facebook is that omniscient deus ex machina that knows us better than we do ourselves. As this meme lives, those who have a product to advertise are inclined to believe it as well (and this makes them spend money on Facebook Ads).
So, in the end, the fact you're getting ads you're not interested in is less relevant and is kind of shadowed over by the fact advertisers get somewhat better value or maybe even just believe they do.
There are shouts that the king is naked and memes that modern machine learning has two objectives: classifying data and making predictions^W^W^W^W^W bullshitting investors and raising money... but so far, such voices are in minority.
Is there another channel that will accept the money of a small business and provide the same results? Even with a compelling event we don’t have the relationships or budget to achieve the same effect through other channels.
If I had the money I would definitely have both.
Bought a Mac recently? How about a new case? Or some discounted software or games? Or how about a new mouse?
Bought a house recently? How about some gardening services? Or some home insurance?
I'm not convinced that if you went and did an actual scientific study on the efficacy of "big data" teams and ad-targeting and went through every business it would do any better than just regular boring contextual ads.
The lie is telling users that targeted ads are good for them
The ML experts in my company keep telling me maintaining these regexps is too costly dev wise, and that they can use ML to adapt to changes in the real world, and get a better result faster.
After nearly a year of work, buying some fancy new computers with giant GPUs, and countless meetings to describe the problem space, they have something that's around 60-70% as accurate as the current state of the art, and say they're close to closing the gap.
Meanwhile I add a new regexp or tweak an existing one and cause them to go back to the drawing board because I've "changed the entire problem description".
Sometimes I feel like I want the job where the only criteria is that I'm using the fancy new toys. It seems like more fun.
Not believe there is totalitarian is the greatest myth one can have in today workd.
"I've got nothing to hide"
YouTube knows my UI is in English, and I've never watched a Cantonese video, and yet it constantly shows me ads in Cantonese. There are a lot of people in Hong Kong who don't speak Cantonese, or at least who are more fluent in English than Cantonese.
All that said, Signal is making steady progress, and I've had zero resistance getting people to use it. Got 12 people on it this past week. Starting a few group threads helped it stick.
Edit: and signal is down...
Is it though? What's the actual benefit for WhatsApp?
They used to charge about $1/yr before acquisition and had over 500mm users. And I think they had fewer than 30 employees.
So they were bringing in about $500mm/yr and their infrastructure costs were unlikely to have been in the 100s of millions.
I'd be very curious to know. I don't remember specifics, but they were running a shockingly low number of servers early on in large part because WhatsApp is built on top of Erlang/OTP.
Or justify the costs of buying it. Supposedly the infrastructure and tech-stack is pretty lean.