The lesson of this story (of a Catholic blog using commercially resold Grindr data to out a gay priest) is *either* that anonymised data can always be de-anonymised (pretty much the intuition of lots of experts I know), or, less generally, you can't expect an org that benefits from selling other people's data to calibrate how much they should spend on anonymising.
twitter.com/josephfcox/status/

@mala

The other lesson of this is that hookup apps should be under the control of the community, not private interests.

@emma or possibly different communities? I know that folks were finding out similar problems when hookup apps were leaking info, which was relatively (relatively!) harmless in say a US context, but were deadly in countries with harsh anti-gay laws. It's hard to generalise threat models even when you think there's a commonality.

One of the advantages of a more federated model, I guess.

Follow

@mala

Yes, federation would help, but there's no amount of tech that will fix a systemic problem like homophobia and it's underlying causes.

@emma @mala there is no combination of federation or community that addresses the basic issue that we are still talking about an app that collects and distributes data. it doesn’t matter who runs it or how it is run, you’re still generating the data.

@emma @mala worse is that grindr and other apps has eroded or outright destroyed the traditional communities that didn’t rely on data collection

@zens @mala

These apps take advantage of alienation under capital and try to replace the role of community.

@zens @emma @mala My understanding on how to properly anonymize data: keep only aggregates. Throw away all individual samples & (wherever you draw this line) outliers.

Adding uniformly random noise is also a promising technique, but it intentionally limits the analysis which can be performed. Looks very promising for opensource app analytics to assert it's not doing anything fishy, but I haven't seen it fully used that way.

yeah the outlier or small cohorts thing is where lazy deanonymization fails
Sign in to participate in the conversation
Magical Girl Party!

The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!