Conversation

Replying to
there's a counting argument here there are far more worthless ads for things that people don't want than ads for things people do and they gotta show them to *someone* because how else are they gonna launder the money
4
Replying to
You need both good ML, and products that can make more money by advertising to those who are very likely to buy. In reality, product A makes $1/sale, and you're 20% likely to buy, but product B makes $100/sale, and you're 0.5% likely to buy - so ads will be for B.
Quote Tweet
i can’t imagine how different the world would feel if even 20% of the ads i saw offered me anything i wanted. right now i’m at less than 1% pretty sure even when the ads are targeted. massively disappointing. motherfuckers should be good enough at ML to do better by now
Show this thread
1
Show replies
Replying to
You think that the people who are solving protein folding haven’t been able to do the math on whether or not high price ads with high conversion are more profitable than low cost ads with bad conversion?
1
1
Show replies
Replying to
Yeah, I don't understand why this happens at all. Are there not *huge* incentives to get this right? This is the kind of thing I would expect civilization to be good at.
Quote Tweet
I imagine a substantial population of people essentially never click on ads (and Google must know this), so one would think they're very low-cost targets for weird experiments like this. What have the advertisers got to lose? twitter.com/ESYudkowsky/st…
Show this thread