[jarcane]

I have at this point lost count of the number of times I or someone else I know on Discord has described getting Youtube recs or ads directly referencing shit they'd just talked about.

I started getting blasted with YT recs for weeks for a channel I had never heard of, never watched, had no real common connections with ... until someone referenced them in a conversation on Discord.

Now, you can shit on that as anecdotal or "intuition" or whatever, but come on. This is HN. No one here is ignorant about the level of data sharing that goes on in the adtech space, and everyone claims they don't until Gmail's selling you ads based on your emails.

It's not like someone's just argued that the president is a space alien. We're talking about suggesting a for-profit internet company is doing something that literally damn near all the for-profit internet companies do and have done now for decades. The bigger question is to how anyone could be so credulous as to assume they are not.

[d3nj4l]

This is the same thing people accuse Facebook of and nobody has convincingly proved Facebook does it. I despise Facebook as much as the next HNer but I am loath to make it seem more capable than it is. Likewise with Discord, I sincerely doubt they can handle combing through voice + text data to build profiles of users, and I'll keep that stance until someone convincingly proves that they do.

[jarcane]

Natural Language Processing is literally a major wing of AI development and data science, but even that isn't really necessary. Matching text on ad-based keywords has been going on since fucking About.com.

[yosito]

I'm not shitting on it, I'm just asking what level of information we have about which data Discord is actually collecting/storing and what they're doing with it.