Hacker News Reader

Comment by oliwarner | Hacker News Reader

> If an agency overloads […] its ex ante regulator, they simply won’t get permission to do things.

The point the author makes about agencies deliberately overloading the authorisation framework is fair but it cuts both ways. The authorisation becomes a bottleneck.

I don't have the answer but what's worse? Missing threat intel because one of three people couldn't review your warrant, or over-collecting?

Bert clearly phrases this all like a regulator, as if agencies should be stopped, and I guess they've seen enough bad warrants to know what's needed, but I don't see the appreciation for the cost of overregulation.

View on HN

> Missing threat intel because one of three people couldn't review your warrant, or over-collecting?

Over-collecting, obviously. If you prioritize not ever missing a threat, there's can be no limit to your expansion of collecting. Because you're always going to miss things. The warnings could be a guy getting angry on a messageboard, and the next morning taking a different bus than usual. Or someone buying a particular progression of books on Amazon, then showing an interest in travel. Or somebody spending an inordinate amount of time chatting rather than playing in some MMO. Or an angry email between father and son.

If you prioritize never missing anything, there's no reason not to constantly collect and run everything through an algorithm.

In the Dutch version (not sure about the English one), he mentions that there's always a fast-track if needed.

Futhermore he says, agencies are incentivices to not overload 'permission' regulators. For 'forgiveness' regulators they have either no such incentive, or worse they might burry those regulators in work so they can't do their work properly.

I don't think he means to say agencies shoud be stopped, more as they need to be kept in line.

> I don't have the answer but what's worse? Missing threat intel because one of three people couldn't review your warrant, or over-collecting?

My money's on over collecting being way worse. I strongly suspect that the genuine benefits of unsupervised deep state activities are greatly blown out of proportion.

The pragmatic (and widely applied) solution here is simply the regulator rubberstamping stuff, then you can optimize the system further by stripping out the checks on the applying side as well.

Reply on news.ycombinator.com