4 ms·

> Posteo could probably still be coerced to push out a fraudulent client update that still breaks their encrpytion, but that is a very hard problem to deal with

> Posteo could probably still be coerced to push out a fraudulent client update that still breaks their encrpytion, but that is a very hard problem to deal with. They are based in Germany and would need to be coerced in accordance to german law. I don't think that there is something like National Security Letters in Germany, so doing such a thing without (eventual?) public disclosure seems unlikely. Also posteo regularly releases as much information as they are allowed to regarding their interaction with law enforcement, e.g.: https://posteo.de/en/site/transparency_report https://posteo.de/en/site/transparency_report

rocqua 9y ago
There are other methods of coercion than courts, but I agree that the German courts are probably much safer than the American courts. In any case, the fraudulent client update is a very hard hole to patch. The only solution for this I know of is local hosting. At the moment, defending against this in web-apps is simply not possible.
- crankylinuxuser 9y ago
  There's one way I know to stop that. Ipfs. Its immutable, for a given key. And its easy to see what an IPNS link points to. It may not be a way to verify, but others could do that hard work. But it strictly shows proof that codebase for a web app hasn't changed.
  - dvdkhlng 9y ago
    The only real antidote against unauthorized access is using end-to-end encryption between email's sender and recipient(s), via PGP/GPG or S/MIME. Unfortunately this requires the sender's cooperation. As long as the sender does not cooperate, there will always be some leg of the communication path that is vulnerable.
    - crankylinuxuser 9y ago
      Indeed. But that problem has been to "Show clients how to install the relevant plugins, make their certs, and then appropriately use certs." That's been error-prone since the beginning with PGPMail and PGPphone. That's why this was switched to doing it in a webapp, to streamline the process and remove user error out of the equation. There's one problem, and that the owner of the script can change it to a bad one that does X. With an immutable data structure, like what IPFS uses, can provide that chain of custody with a script they make that simplifies PGP usage, while still maintaining "We didn't change anything" - and you can prove that.
_Codemonkeyism 9y ago
German intelligence services in the past just listened to all traffic at internet exchange points. No need for national security letters.
- dvdkhlng 9y ago
  And not only German intelligence services. This is why most websites nowadays protect against pervasive monitoring by using TLS and ephemeral keys (including Posteo). Also the mail transport layer of most mail servers uses TLS-encrypted SMTP which is safe against pervasive monitoring (but may be affected by MITM attacks). Some mail servers are set up or can be configured to not trannsmit or receive mails unless TLS protocol can be negotiated (I think mailbox.org allows users to enforce SMTP via TLS by using the domain secure.mailbox.org, not sure about Posteo). (See also https://tools.ietf.org/html/rfc7258 https://tools.ietf.org/html/rfc7258 https://support-en.mailbox.org/knowledge-base/article/before-you-send-check-ssl https://support-en.mailbox.org/knowledge-base/article/before...) [edit] replaced SSL with TLS

rocqua 9y ago

There are other methods of coercion than courts, but I agree that the German courts are probably much safer than the American courts. In any case, the fraudulent client update is a very hard hole to patch. The only solution for this I know of is local hosting. At the moment, defending against this in web-apps is simply not possible.

crankylinuxuser 9y ago
There's one way I know to stop that. Ipfs. Its immutable, for a given key. And its easy to see what an IPNS link points to. It may not be a way to verify, but others could do that hard work. But it strictly shows proof that codebase for a web app hasn't changed.
- dvdkhlng 9y ago
  The only real antidote against unauthorized access is using end-to-end encryption between email's sender and recipient(s), via PGP/GPG or S/MIME. Unfortunately this requires the sender's cooperation. As long as the sender does not cooperate, there will always be some leg of the communication path that is vulnerable.
  - crankylinuxuser 9y ago
    Indeed. But that problem has been to "Show clients how to install the relevant plugins, make their certs, and then appropriately use certs." That's been error-prone since the beginning with PGPMail and PGPphone. That's why this was switched to doing it in a webapp, to streamline the process and remove user error out of the equation. There's one problem, and that the owner of the script can change it to a bad one that does X. With an immutable data structure, like what IPFS uses, can provide that chain of custody with a script they make that simplifies PGP usage, while still maintaining "We didn't change anything" - and you can prove that.

_Codemonkeyism 9y ago

German intelligence services in the past just listened to all traffic at internet exchange points. No need for national security letters.

dvdkhlng 9y ago
And not only German intelligence services. This is why most websites nowadays protect against pervasive monitoring by using TLS and ephemeral keys (including Posteo). Also the mail transport layer of most mail servers uses TLS-encrypted SMTP which is safe against pervasive monitoring (but may be affected by MITM attacks). Some mail servers are set up or can be configured to not trannsmit or receive mails unless TLS protocol can be negotiated (I think mailbox.org allows users to enforce SMTP via TLS by using the domain secure.mailbox.org, not sure about Posteo). (See also https://tools.ietf.org/html/rfc7258 https://tools.ietf.org/html/rfc7258 https://support-en.mailbox.org/knowledge-base/article/before-you-send-check-ssl https://support-en.mailbox.org/knowledge-base/article/before...) [edit] replaced SSL with TLS