3 ms·
> Well, let's just say that there aren't many journalists I trust to validate a DKIM signature Uh... why would the journalist "validate" the DKIM signature? P
by newacct583 6y ago
> Well, let's just say that there aren't many journalists I trust to validate a DKIM signature
Uh... why would the journalist "validate" the DKIM signature? Publish the RFC822 content of the email and let everyone do it themselves. That's the whole point of public key encryption. And the fact that the Post skipped that very obvious and easy step tells me that this is almost certainly faked data.
> I don't just believe any random person who claims XYZ
With all respect: you clearly seem inclined to believe this nonsense about the emails with only the barest of evidence. It's only the attempts to refute it that have you worried about "rigor".
- Natsu 6y agoEh, sometimes the signatures fail to validate because mail clients do non-substantive modifications (spacing, etc.). We went through that before, too. You're right that one should always validate it themselves and I should've said that I don't think many journalists know how to work it, because honestly, I haven't seen anyone but Wikileaks actually do that validation, ever. > With all respect: you clearly seem inclined to believe this nonsense about the emails with only the barest of evidence. I haven't said anything about the email content, though. I investigated a photo of Hunter sleeping with a meth pipe using hackerfactor.com's tools to look for manipulation and did not find any. That doesn't mean they're real, but it gives them some level of credence, given that there are old reports about his rental car substantiated by police reports. You're right that it doesn't validate the email content. I plan to withhold judgement until we have more data, but it looks like they do have some files of his. It doesn't help that the last time we went through this sort of thing, there were spurious claims of manipulated documents which were actively disproved and some of the denials I've read parse very narrowly, which isn't right.