" three researchers have crafted a long-sought version of private information retrieval and extended it to build a more general privacy strategy. The work, which received a Best Paper Award in June 2023 at the annual Symposium on Theory of Computing, topples a major theoretical barrier on the way to a truly private search."
Just in time for all the searchable information to be completely drowned out by low quality AI content.
Thank God I did all my searching before this happened.
did you also make hard copies of everything you found?
Dang it! No! brb
It’s already drowned out by low quality SEO spam, so no difference
This is my nightmare
Let’s take a moment and acknowledge that it was never hard to make searches private.
It’s just that doing that requires trusting a company not to fuck with you behind the scenes and sell you out, and ensuring that doesn’t happen is fucking hard
requires trusting a company not to fuck with you behind the scenes
The point of this cryptography is that you don’t have to trust the company implementing it not to do that, as long as you trust the software doing the retrieval.
Unless you do it all yourself, it will never be trustless.
Just like your favorite phone app, in the end you will have to trust the actual code, and you will have to trust that the actual app on your actual phone is from the actual source they claim. Do you feel lucky?
Actually, to make it with cryptographic guarantees is pretty hard… I know of at least one university professor in the PET (Privacy Enhancing Technologies)/cryptography space who spent quite some time on his startup to develop such a search engine. In the end it all fell apart because of one the mathematical assumptions being unprovable. This is just one example but I guess it illustrates pretty well why we’ve yet to see a cryptographically secure/private search engine as a product!
Most Linux package managers sync the list of all packages they can download. I wonder if some sort of system like that can be used to federate web searching.
So we all just have a copy of the internet and then we can grep for things we are interested in … I’m actually not super against this.
The storage requirements might be ever so slightly prohibitive.
Well perhaps a selection? You don’t need the data set for crotcheting or perl coding if that’s not up your alley.
Once you do that though you end up telling people what you are interested in which goes against the whole purpose in the first place.
Ah ok so we’re talking perfect safety here
