You Are the Product (2017)

Social media is very good at making us angry. A debate on starter pokemon can quickly turn into slurs being thrown around. I have a few social circles I remain engaged in via zoom, and pre Covid I was meeting people at alumni events, etc , left and right.

It's not particularly easy to re engage with the real world if your way deep in the rabbit whole of social media. Another way to look at it. Online you become the worst thing you've ever posted , once you identify with that negativity it feeds upon itself. You end up in a death spiral of your own self-hatred.

Why give others the right to rip you apart. Why rip yourself a part on a daily basis. I don't think I've ever convinced anyone to my point of view on social media.

I have been left feeling distraught over some of the toxic things said to me though. I'll even say people are outright meaner online. In real life your friends will just distance themselves if you aren't adding to them.

Online it's not uncommon to see users creating various accounts just to attack each other.

The only way to win is not to play.

I also suggest reading Lost Connections, it goes more into detail on the need for real community.

That said, once we start to talk about "manipulation", things get very relative. Human beings are social animals. I want to have contact with other people and so be influenced by them? At some point we can call some influence manipulation but there's a lot of gray areas. "Make my feed better" is a key grey area. Filtering crap from the Internet is one of the biggest jobs of both Facebook and Google. At some points that filtering becomes self-serving censorship but there are lots of gray areas. The biggest thing is that isn't a static problem - creators of viral manipulation products and ideologies get more and more sophisticated as methods evolve (articles on QAnon as gamification are worth reading). Not that Facebook and Google are blameless but it seems a ultra-networked worked, there's going to be multiple powerful forces and filters competing for influence and which one is "evil" is a difficult call.

You don't need personal data in order to provide targeted advertising. The proper context of where the ad will be embedded is mostly enough for good targeting.

Do you think youtube algorithms optimize for you learning and becoming a better human or to maximize ad revenue? Based on the answer the knowledge you gained is on average either deep and meaningful or shallow, bordering entertainment value.

That is not to say you can’t find deep structured knowledge on youtube, but the proper way to support its creation is to directly pay for it. Next best thing is paying for youtube premium.

Imagine two youtubes, one ran by ads only and one exclusively for paid subscribers. Which one of these is more likely to have funny cat videos and which one deep, meaningful knowledge and why? In which case the interest of the platform is aligned with the interest of the user?

When overdone, ad business models incentivize mass creation of low quality entertainment which is what most of internet including news has become. In no way ads help democratize the access to valuable information.

The flaw is the assumption that this is not possible without them. Consider what happens in the alternative.

The services would still exist, but they would charge a ridiculous amount of money, even more so because charging money means most people won't pay, but if only a minority of people are paying the prices have to be even higher.

So then a hobbyist starts their own free one, and it's terrible, because it's the effort of one person on a weekend instead of a full-time staff. But free-as-in-beer. So everyone who can't afford the pay service uses that one instead.

It ends up with many users, because it's free. And the users want it to be better, so they make improvements and submit patches, or donate money etc. You end up with Linux, BitTorrent, OpenStreetMap. Somebody tell me why you couldn't build a YouTube based on BitTorrent.

You could, over time, if you could get people to use it while it's still terrible and therefore improve it. Which you could do if it was the free one when all the others cost money, but you can't do when the existing YouTube is better funded and proprietary yet still doesn't require the user to pay money to use.

And then the dominant platforms become controlled by individual companies instead of being federated open standards, which suppresses innovation by creating a monoculture that only one entity has the ability to modify.

I am somewhat sensitive to data mining but I guess not as much as other people on HN, given I am almost anonymized on the internet.

The worst thing about ads is security, and that's why I block ads even if I am browsing with significantly anonymous profile. Performance is a nice bonus.

I don't think data-mining based ads work well in practice. User_analyze.py can't tell whether I have already purchased the product I searched an hour ago, or lost interest in it. I have yet to see a good recommendation system based on data collection. Blame the big data / machine learning hype.

I wonder if you did this across multiple devices, you could start noticing differences based on what you used for what. A/B test them back.

Don't want to be the product ? Pay for your services then - the money has to come from somewhere.

In a world where Facebook just makes up video impressions and neither they nor Google let you really audit their inventory, it is basically impossible to tell the difference between (1) seasonal effects, (2) data-enabled ad targeting, and (3) income-related effects. In other words, if you show an ad at the right time to a rich person, say by only targeting iPad users shortly after a new Apple product release, which is known from the user agent and does not require behavioral tracking at all, your ad will perform as well as a typical interest-targeted ad but with lower cost.

Which is to say, your data is valuable in the marketing sense that it gets people to spend more on targeted ads, but not necessarily in some secular sense that it actually increases returns. After fundamentals (like what you are selling) and the creative, most of the evidence points to trends in how rich (and poor) people use technology, like iPad users versus entry-level Android users, as being the greatest predictor of an ad's returns.

Ads customers don't really know if user profile information like their browsing behavior improves ROI because it is secularly important for the ad or because it is also correlated with greater income. It is unmeasurable for ad customers, so people claiming that they know are just lying.

At least among the ad tech companies I know, I'm told the user profile data gathering story is very valuable for ad buyers and investors, but fundamentally they do not use any behavioral data in ad targeting because they did not observe an increase in clicks - for their guaranteeably low income user channels.

To watch the videos, you can copy paste the url into VLC. No ads and very little tracking. Best of all no related videos so you won't waste tons of time there.

Good old society was a little less duplicit about how it operates. We live in a i-passive-aggresive land.

>Plenty of companies, indeed entire industries, base their business model on being evil. The insurance business, for instance, depends on the fact that insurers charge customers more than their insurance is worth; that’s fair enough, since if they didn’t do that they wouldn’t be viable as businesses. What isn’t fair is the panoply of cynical techniques that many insurers use to avoid, as far as possible, paying out when the insured-against event happens. Just ask anyone who has had a property suffer a major mishap.

The article isn't saying that insurance companies are 'evil' for making a profit but instead for doing everything they can to withhold payment, not every insurance company does this of course but many of them know that customers are more likely to accept 50% blame in a 0% blame car accident or accept a reduced payout over minor technicalities because the alternative is going months without a car and spending a significant amount of time and money in court.

This analogy is entirely applicable to governments too where an 'evil' government might, at great cost, make benefit programs difficult or humiliating to obtain to discourage legitimate claimants from doing so, never mind straight up corruption or cronyism when it comes to allocating public funds.