News for hackers. Stuff that matters.
Re:Twitter now requires an account to view tweets by AnthonyMouse on Saturday July 01, 2023 @02:03PM (#36553043)
This is a legitimate concern, but whether the people doing the analysis get the data via scraping vs. a box of hard drives is pretty irrelevant to it. To actually solve it you would need the data to not be public.

One of the things you could do is reduce the granularity. So instead of showing that someone posted at 1:23:45 PM on Saturday, July 1, 2023, you show that they posted the week of June 25, 2023. Then you're not going to be doing much time of day or day of week analysis because you don't have that anymore.

Re:Twitter now requires an account to view tweets by dredmorbius on Saturday July 01, 2023 @02:44PM (#36553593)
Yes, once the data are out there ... it's difficult to do much.

Though I've thought for quite some time that making the trade and transaction of such data illegal might help a lot.

Otherwise ... what I see many people falling into the trap of is thinking of their discussions amongst friends online as equivalent, say, to a discussion in a public space such as a park or cafe --- possibly overheard by bystanders, but not broadcast to the world.

In fact there is both a recording and distribution modality attached to online discussions that's utterly different to such spoken conversations, and those also give rise to the capability to aggregate and correlate information from many sources.

Socially, legally, psychologically, legislatively, and even technically, we're ill-equipped to deal with this.

Fuzzing and randomising data can help, but has been shown to be stubbornly prone to de-fuzzing and de-randomising, especially where it can be correlated to other signals, either unfuzzed or differently-fuzzed.