There certainly is a problem you'll see on HN because they think every company is out to get them because they read Snow Crash.
But Cambridge Analytica was not real for instance, it was just marketing fooling people. Palantir's biggest cheerleaders are the paranoid.
Marketers have done it for 100 years. So, yes it's all fake, you're hiring someone to fool people, see the problem, but I'm not sure it's a bubble if it never pops.
Exactly. People have been incorrectly measuring online ad effectiveness since online ads first existed. The only reason they continue to spend on online ads is that they couldn't measure offline ad effectiveness at all. As long as this is true, and as long as the ad buyers remain unsophisticated, this bubble will not pop.
I wonder if part of the demand for the newer creepy ads-- the extensive profiling and retargeting, for example-- is more on the publisher side than the advertiser side.
There are plenty of sites where you could run 100% content-targeted ads and they'd be relevant and convert well without the cost of Big Data Deep AI Learning (tm) or the associated creepiness. Hell, you could run your own ad server or use on-page links most of the time.
But there are some huge markets where this doesn't work well. Generalized local news is the classic problem. It's difficult to come up with a highly relevant ad mix that can be placed against "Mayor Announces End to Glass Recycling" today and "Local Sportsball Team Wins Championship" tomorrow. Even if you had Google's range of inventory, you're dealing with content that may not be brand-friendly. (Who wants their ad next to a Coronavirus death counter or a beached whale?).
It almost feels like the entire creepy-ad ecosystem is built to service that sort of publisher. It offers an endless stream of ads which aren't lowest-bidder without having to care about content relevance, and it provides an "algorithm said so" deniability when the ads get placed embarrassingly.
It sort of works, allowing the tattered remnants of the local newspaper to keep their heads above water if they pile enough ads on the page, but it's then projected onto a broader market that probably doesn't need it, and where it only provides marginal benefits at high costs (financially, technically, and in privacy)
6 comments
[ 3.7 ms ] story [ 26.4 ms ] threadThere certainly is a problem you'll see on HN because they think every company is out to get them because they read Snow Crash.
But Cambridge Analytica was not real for instance, it was just marketing fooling people. Palantir's biggest cheerleaders are the paranoid.
Marketers have done it for 100 years. So, yes it's all fake, you're hiring someone to fool people, see the problem, but I'm not sure it's a bubble if it never pops.
There are plenty of sites where you could run 100% content-targeted ads and they'd be relevant and convert well without the cost of Big Data Deep AI Learning (tm) or the associated creepiness. Hell, you could run your own ad server or use on-page links most of the time.
But there are some huge markets where this doesn't work well. Generalized local news is the classic problem. It's difficult to come up with a highly relevant ad mix that can be placed against "Mayor Announces End to Glass Recycling" today and "Local Sportsball Team Wins Championship" tomorrow. Even if you had Google's range of inventory, you're dealing with content that may not be brand-friendly. (Who wants their ad next to a Coronavirus death counter or a beached whale?).
It almost feels like the entire creepy-ad ecosystem is built to service that sort of publisher. It offers an endless stream of ads which aren't lowest-bidder without having to care about content relevance, and it provides an "algorithm said so" deniability when the ads get placed embarrassingly.
It sort of works, allowing the tattered remnants of the local newspaper to keep their heads above water if they pile enough ads on the page, but it's then projected onto a broader market that probably doesn't need it, and where it only provides marginal benefits at high costs (financially, technically, and in privacy)
I'm sure there is quite a bit of money to be made shorting Google and Facebook.
There is no value without the 'When'