Hacker News Reader

Comment by highland586 | Hacker News Reader

> it indicates the current technical capabilities on using and abusing personal information is still limited

Does it really though? I would expect a successfully manipulative algorithm to be much more subtle than this; just because we don't notice it, doesn't mean it's not there.

View on HN

Since events like this famous case, where Walmart outed a teenage girl pregnancy, it's well-known they mix in some "weird" or irrelevant things to give a sense of plausible deniability. "oh, pregger testers, haha, how weird they can be <nervous laughter>".

https://www.forbes.com/sites/kashmirhill/2012/02/16/how-targ...

This was Target not Walmart and I say that not to nitpick but rather highlight that arguably Walmart is somewhat an upper tier tech company.

If this is what Target does in 2012, Amazon in 2020 not only inserts plausible deniability to optimize for lack-of-creepiness factor. They're probably running AI bots on forums and social media injecting seamless product placement in commentary you see in your browser.

If we observe that the targeted advertising doesn't seem all that good, it's a bit of a leap to suspect there's actually a great algorithm that's so subtly manipulative we don't notice it. By Occam's razor, the algorithm is probably just showing too many vacuums because it's not that good.

I don't deny its existence (in conference papers, in labs, in black projects by government and corporations, or even in a tiny startup company). I'm just saying so far it's not mature enough to be widely deployed on large platforms, like Amazon's banner ads.

Reply on news.ycombinator.com