5 ms·

I have a different proposal to solving the privacy vs enforceability issue: require a license for strong encryption. Any Tom Dick or Harry can have encryption

I have a different proposal to solving the privacy vs enforceability issue: require a license for strong encryption. Any Tom Dick or Harry can have encryption of a certain strength, chosen so that the cost of breaking it is around some dollar cost over the next N years. Low enough that the NSA will do it for terrorist suspects, but high enough that any mass surveillance is infeasible. Also, require court warrants for accessing this data, or attempting to crack the cypher. There will be applications where this is insufficient, say e2e communication with banks. These can be licensed and supervised. I think this provides a decent tradeoff to a bunch of problems: - people can still enjoy fairly strong encryption - states can't spy on people willy-nilly - but will be able to do it, with a lot of effort, where they are really motivated - sure, "bad guys" can still download GPG and encrypt something with a very hard cipher - but that in itself will be an offense...

dmitrygr 3y ago
Poe's law vibes...
simonw 3y ago
This is a terrible idea. Today's "fairly strong encryption" is tomorrow's broken encryption.
- rich_sasha 3y ago
  Well, assume some Moore's law increases.
  - simonw 3y ago
    It's also assuming that only government entities could afford to build a cluster that could crack this kind of encryption. Criminal enterprises don't need to run mass surveillance, but they may well find it worth paying to crack encryption for very specific targets.
    - btilly 3y ago
      The criminals don't even need to build it. They can just repurpose a DDoS network...
betaby 3y ago
Not sure if OP is serious or all above is a sarcasm.
redeeman 3y ago
or how about this: government stays the hell away from me, and can come knocking on my door once they have some evidence of wrong doing
- cooper_ganglia 3y ago
  Everyday we slip closer and closer to the abyss of authoritarianism propped up under the guise of "security" by the thunderous applause of the populace. We didn't need a whole Star Wars trilogy to tell us why that's a bad idea...
omginternets 3y ago
You may want to read up about the so-called "crypto wars" of the 90's. You'll find it illuminating.
kmeisthax 3y ago
Encryption is either strong or it isn't - there's no room for a "breakable by governments but not by anyone else" tier of encryption. We learned this lesson the hard way with "export grade crypto" - i.e. the reason why DVDs were so hilariously easy to decrypt. The safety margin of how many bits there are between "breakable with a smartphone" and "unbreakable by every computer on earth" is very thin, and it shifts a lot, so the NSA has no advantage over 'bad guys'. The moment the NSA can brute-force a key so can anyone with a botnet of compromised AWS instances. Alternative proposals to force messaging apps to provide decryption keys for otherwise securely encrypted material aren't any better, because computers and encryption algorithms cannot read and understand a court order[0]. So if the keys are held by the government then they can decrypt everything without actually needing to ask someone else and prove that all the constitutionally-mandated procedures were followed. If the keys are held by the communications service, then they can at least theoretically refuse an illegal order[1], but then they run the risk of insiders stealing people's secrets. Remember that the push for E2E encryption started specifically because of two related trends: - Intelligence agencies doing incredibly sketchy shit to bypass encryption at major tech companies (e.g. that NSA diagram explaining how you can get implants into Google datacenters and bypass the encryption) - Companies wanting to protect the privacy of their users from themselves, because they hire a lot of people with potentially too much access to the production equipment. Banning E2E, going back to "export grade encryption", or demanding decryption capability rolls both of these back significantly. You can engineer a key escrow / backdoor system to resist one particular kind of attack at the expense of making another easier, but you can't eliminate both while maintaining nonconsentual decryption capability. [0] This is a corrolary of that old IBM saying: "a computer can never be held accountable, therefore a computer must never make a management decision". [1] Let's put aside the whole "no expectation of privacy on metadata" nonsense for now. While that is already a privacy nightmare, opening up communications content to unlimited decryption is a bigger can of worms.

dmitrygr 3y ago

Poe's law vibes...

simonw 3y ago

This is a terrible idea. Today's "fairly strong encryption" is tomorrow's broken encryption.

rich_sasha 3y ago
Well, assume some Moore's law increases.
- simonw 3y ago
  It's also assuming that only government entities could afford to build a cluster that could crack this kind of encryption. Criminal enterprises don't need to run mass surveillance, but they may well find it worth paying to crack encryption for very specific targets.
  - btilly 3y ago
    The criminals don't even need to build it. They can just repurpose a DDoS network...

betaby 3y ago

Not sure if OP is serious or all above is a sarcasm.

redeeman 3y ago

or how about this: government stays the hell away from me, and can come knocking on my door once they have some evidence of wrong doing

cooper_ganglia 3y ago
Everyday we slip closer and closer to the abyss of authoritarianism propped up under the guise of "security" by the thunderous applause of the populace. We didn't need a whole Star Wars trilogy to tell us why that's a bad idea...

omginternets 3y ago

You may want to read up about the so-called "crypto wars" of the 90's. You'll find it illuminating.

kmeisthax 3y ago

Encryption is either strong or it isn't - there's no room for a "breakable by governments but not by anyone else" tier of encryption. We learned this lesson the hard way with "export grade crypto" - i.e. the reason why DVDs were so hilariously easy to decrypt. The safety margin of how many bits there are between "breakable with a smartphone" and "unbreakable by every computer on earth" is very thin, and it shifts a lot, so the NSA has no advantage over 'bad guys'. The moment the NSA can brute-force a key so can anyone with a botnet of compromised AWS instances. Alternative proposals to force messaging apps to provide decryption keys for otherwise securely encrypted material aren't any better, because computers and encryption algorithms cannot read and understand a court order[0]. So if the keys are held by the government then they can decrypt everything without actually needing to ask someone else and prove that all the constitutionally-mandated procedures were followed. If the keys are held by the communications service, then they can at least theoretically refuse an illegal order[1], but then they run the risk of insiders stealing people's secrets. Remember that the push for E2E encryption started specifically because of two related trends: - Intelligence agencies doing incredibly sketchy shit to bypass encryption at major tech companies (e.g. that NSA diagram explaining how you can get implants into Google datacenters and bypass the encryption) - Companies wanting to protect the privacy of their users from themselves, because they hire a lot of people with potentially too much access to the production equipment. Banning E2E, going back to "export grade encryption", or demanding decryption capability rolls both of these back significantly. You can engineer a key escrow / backdoor system to resist one particular kind of attack at the expense of making another easier, but you can't eliminate both while maintaining nonconsentual decryption capability. [0] This is a corrolary of that old IBM saying: "a computer can never be held accountable, therefore a computer must never make a management decision". [1] Let's put aside the whole "no expectation of privacy on metadata" nonsense for now. While that is already a privacy nightmare, opening up communications content to unlimited decryption is a bigger can of worms.