3 ms·

I know but I have never heard any details especially how they want to keep the backdoor secret. Cracking the backdoor would be such a high value target that a

I know but I have never heard any details especially how they want to keep the backdoor secret. Cracking the backdoor would be such a high value target that a lot of people would spend insane amounts of money and energy on it.

JoshTriplett 7y ago
Do not under any circumstances let the discussion move from "whether" to "how". Reject the premise of the question. Such a system cannot be built; the requirements are broken. Treat it as though someone asked you to build a system that solves the halting problem, or factors products of large primes in linear time: the correct direct response is patient explanation of impossibility, and the correct indirect response is good PR from real security experts who understand how to do good PR.
- Ididntdothis 7y ago
  I don’t think it’s right to reject something outright. You should always give people the opportunity to show what they have. Then take a look at it and decide whether it’s good. A strong indicator for a scam is that they won’t show anything when asked for details. Then things should get rejected.
  - JoshTriplett 7y ago
    I'm not suggesting a refusal to listen to the question. I'm observing that it's dangerous to accept the premise of a question that may move the goalposts, without first noticing the implicit premise or differing definition, and explaining why the premise or definition is faulty. Accepting the faulty premise moves the line to "so you just need to come up with the right technology", which then moves to "what's the most secure system you can build while still having a backdoor". The right answer to that is not "here's what we could hypothetically build", it's "by definition we cannot build a secure system with a backdoor, and also here are all the additional problems that a backdoor would introduce". By all means, fully listen to proposals, in order to understand the misconceptions and better address them. Or, in some cases, it's useful to listen to proposals to anticipate upcoming security threats and build a more robust security model. For instance, the potential threat of compromising CAs led to the invention of Certificate Transparency, which almost completely eliminates that possibility. We also need to get a lot better at rhetoric and PR.
    - Ididntdothis 7y ago
      "so you just need to come up with the right technology" That line is not OK. If somebody wants something and you say you are not able to do it then it should be on them to prove you wrong. Reminds me of work BS. You say “it can’t be done in the timeframe”. VP response: “you are just not working hard enough”. It’s a nice way to manipulate the conversation. Technologists definitely should get better at messaging and arguing with bad faith actors.
SpicyLemonZest 7y ago
The obvious explanation for why we've heard no details is that they don't plan to work especially hard at it. Some DOJ messaging sounds very much like they don't believe it's a real problem[1] - companies can keep their centrally-managed signing keys secret, they say, so why couldn't we keep centrally-managed backdoor keys secret? [1] https://www.justice.gov/opa/speech/attorney-general-william-p-barr-delivers-keynote-address-international-conference-cyber https://www.justice.gov/opa/speech/attorney-general-william-...

JoshTriplett 7y ago

Do not under any circumstances let the discussion move from "whether" to "how". Reject the premise of the question. Such a system cannot be built; the requirements are broken. Treat it as though someone asked you to build a system that solves the halting problem, or factors products of large primes in linear time: the correct direct response is patient explanation of impossibility, and the correct indirect response is good PR from real security experts who understand how to do good PR.

Ididntdothis 7y ago
I don’t think it’s right to reject something outright. You should always give people the opportunity to show what they have. Then take a look at it and decide whether it’s good. A strong indicator for a scam is that they won’t show anything when asked for details. Then things should get rejected.
- JoshTriplett 7y ago
  I'm not suggesting a refusal to listen to the question. I'm observing that it's dangerous to accept the premise of a question that may move the goalposts, without first noticing the implicit premise or differing definition, and explaining why the premise or definition is faulty. Accepting the faulty premise moves the line to "so you just need to come up with the right technology", which then moves to "what's the most secure system you can build while still having a backdoor". The right answer to that is not "here's what we could hypothetically build", it's "by definition we cannot build a secure system with a backdoor, and also here are all the additional problems that a backdoor would introduce". By all means, fully listen to proposals, in order to understand the misconceptions and better address them. Or, in some cases, it's useful to listen to proposals to anticipate upcoming security threats and build a more robust security model. For instance, the potential threat of compromising CAs led to the invention of Certificate Transparency, which almost completely eliminates that possibility. We also need to get a lot better at rhetoric and PR.
  - Ididntdothis 7y ago
    "so you just need to come up with the right technology" That line is not OK. If somebody wants something and you say you are not able to do it then it should be on them to prove you wrong. Reminds me of work BS. You say “it can’t be done in the timeframe”. VP response: “you are just not working hard enough”. It’s a nice way to manipulate the conversation. Technologists definitely should get better at messaging and arguing with bad faith actors.

SpicyLemonZest 7y ago

The obvious explanation for why we've heard no details is that they don't plan to work especially hard at it. Some DOJ messaging sounds very much like they don't believe it's a real problem[1] - companies can keep their centrally-managed signing keys secret, they say, so why couldn't we keep centrally-managed backdoor keys secret? [1] https://www.justice.gov/opa/speech/attorney-general-william-p-barr-delivers-keynote-address-international-conference-cyber https://www.justice.gov/opa/speech/attorney-general-william-...