One of the most dangerous narratives coming out of the pro-encryption camp right now is that metadata surveillance (by any other name) can be a solution to "online harms" in lieu of banning or restricting access to end-to-end encryption.
Let us not blatantly ignore the fact that governments and corporations already conduct vast, intrusive metadata surveillance campaigns - and have done for decades.
Metadata surveillance is the gross, default state of the modern world.
The real solution to online harms rests in more anonymity, more encryption, more consensual communication and more decentralization of power.
We @OpenPriv are building @cwtch_im, a truly decentralized and surveillance resistant application.
If you want to actually start proactively defending privacy instead of reacting to every fresh legislative hell, then come help us.
I'm not sure people really understand this, so I'm going to say this again:
The current argument by the anti-encryption side is that the threat of state violence should be used to prevent the offering of math-as-a-service.
I will grant that there are legitimate sounding arguments that may prompt someone to consider pointing weapons at anyone who dares to carry out the dark arts of arithmetic but I would hope that by now most would have outgrown such childish notions of how the world aught to work.
The attack improvements come from considering temporal relationships (the probability of receiving messages over a given threshold in a period of time) instead of just over the lifetime of the system.
This can be devastating if false positive rates are poorly selected.
I think the main takeaway is that there hasn't been enough push back and that this now seems depressingly inevitable.
I expect we will see more calls for surveillance like this in the coming months heavily remixed into the ongoing "online harms" narrative.
Without a strong stance from other tech companies, in particular device manufacturers and OS developers, we will look back on the last few weeks as the beginning of the end of generally available consumer devices that don't conduct constant algorithmic surveillance.
Someone asked me on a reddit thread the other day what value t would have to be if NeuralHash had a similar false acceptance rate to other perceptual hashes and I ball parked it at between 20-60...so yeah.
Some quick calculations with the new numbers:
3-4 photos/day: 1 match every 286 days.
50 photos/day: 1 match every 20 days.