3 ms·

I'm just saying that if you go out and verify all those things today (even looking at archive.org etc), then as long as you're using just one (type of) browser

by 0x0 10y ago

I'm just saying that if you go out and verify all those things today (even looking at archive.org etc), then as long as you're using just one (type of) browser configuration, compromising a single extension could be enough to "blur your vision" when you decide to sign that key. A couple of regexes replacing a few variants of the real fingerprint with the fake fingerprint in all HTML documents - then every forum post you look at echoing the key might be a lie. Those scenarios are "OR", not "AND"... :) Or, you know, for paranoia level 1000, your posts are the sockpuppet ;) I'll agree this is all super unlikely and I'm not accusing you or anyone here of shenanigans. I'm just saying that signing keys shouldn't be taken lightly, because if people start getting sloppy then a lot of the trust in the web of trust is no longer water tight.

web007 10y ago
I don't run that kind of extension [0], but that's the reason I included this line: > That same entity compromised my browser, curl, my Linux laptop, Windows desktop, Android phone and 3 different networks. At least 3 platforms, potentially different 3 browsers, plus command-line. If I want to do this "properly" it's a lot of work. I'd get 3 or more independent VPN providers and run the test while connected to each, independently or in series. I'd run everything in a VM that I created from a verified source 5+ years ago on hardware from 10+ years ago[1]. I would have to categorize and confirm that every vulnerability in all software was either fixable via configuration, or couldn't contribute to remote compromise of the box. I'd run the test with and without Tor, hard-resetting the VM to its initial state for every variation and comparing the results across every run to detect abnormalities. I'd confirm the content at Archive, CommonCrawl, Yahoo, Google and at least one foreign provider, Yandex or Baidu or whoever else I could track down. I'd look for that same string in all places I could find it (web, newsgroups, torrents, friends, friends-of-friends), and confirm that neither those pages nor any linked pages contain invalidations, or if any of them contain further evidence to support the identity of said key. By "linked" I mean both links from those pages, and anything I could discover via search engine pointing to those pages independently. I'd go find some old torrents or archives of older versions of the binaries and/or signatures, and confirm that their content matches the published content. I'd confirm that the old key was in use before 0.11, and that the new key is indeed the one that signed release 0.11+. I'm sure there are some further steps I could take to invalidate any known or theorized attacks from state-level agencies, but at this point it's probably more difficult to fake than a GPG signature. And obviously I'm a sockpuppet - nobody should listen to random strangers on the internet without verifying their identity as well! You could look at my keybase profile, but A) that's easy to fake and B) that's not the key I used to sign above - clearly something hinkey is going on. ;) [0] https://chrome.google.com/webstore/detail/ext/apmlngnhgbnjpajelfkmabhkfapgnoai?hl=en https://chrome.google.com/webstore/detail/ext/apmlngnhgbnjpa... [1] https://libreboot.org/faq/#intel https://libreboot.org/faq/#intel

by 0x0 10y ago

web007 10y ago

I don't run that kind of extension [0], but that's the reason I included this line: > That same entity compromised my browser, curl, my Linux laptop, Windows desktop, Android phone and 3 different networks. At least 3 platforms, potentially different 3 browsers, plus command-line. If I want to do this "properly" it's a lot of work. I'd get 3 or more independent VPN providers and run the test while connected to each, independently or in series. I'd run everything in a VM that I created from a verified source 5+ years ago on hardware from 10+ years ago[1]. I would have to categorize and confirm that every vulnerability in all software was either fixable via configuration, or couldn't contribute to remote compromise of the box. I'd run the test with and without Tor, hard-resetting the VM to its initial state for every variation and comparing the results across every run to detect abnormalities. I'd confirm the content at Archive, CommonCrawl, Yahoo, Google and at least one foreign provider, Yandex or Baidu or whoever else I could track down. I'd look for that same string in all places I could find it (web, newsgroups, torrents, friends, friends-of-friends), and confirm that neither those pages nor any linked pages contain invalidations, or if any of them contain further evidence to support the identity of said key. By "linked" I mean both links from those pages, and anything I could discover via search engine pointing to those pages independently. I'd go find some old torrents or archives of older versions of the binaries and/or signatures, and confirm that their content matches the published content. I'd confirm that the old key was in use before 0.11, and that the new key is indeed the one that signed release 0.11+. I'm sure there are some further steps I could take to invalidate any known or theorized attacks from state-level agencies, but at this point it's probably more difficult to fake than a GPG signature. And obviously I'm a sockpuppet - nobody should listen to random strangers on the internet without verifying their identity as well! You could look at my keybase profile, but A) that's easy to fake and B) that's not the key I used to sign above - clearly something hinkey is going on. ;) [0] https://chrome.google.com/webstore/detail/ext/apmlngnhgbnjpajelfkmabhkfapgnoai?hl=en https://chrome.google.com/webstore/detail/ext/apmlngnhgbnjpa... [1] https://libreboot.org/faq/#intel https://libreboot.org/faq/#intel