Advanced de-faking: using public sources to trace the true age of a suspected propaganda video


#1

Originally published at: http://boingboing.net/2017/03/06/fake-news-vs-real-research.html


#2

Propagating propaganda via propaganda.


#3

Interesting. I do hope they have more proof of the building being from 2016 than that one grainy satellite photo with scaffoldings. To me it did not conclusively look unfinished, it could also be scaffoldings for maintenance for example.

The analysis is probably right, but ‘probably’ doesn’t cut it in this situation IMHO.

But the build date of the building should be easy to verify by other means than the grainy sat image so still good work.

(badly written article though, but the author’s first language is probably german (or ,looking at the name, maybe dutch) so that’s just nitpicking. Though anti-dated makes me cringe, and thats also wrong in dutch or german.)


#4

OK, so he pushed the date of the video claimed to be from 2011 to 2016 at least, and not 2017.

However, whats the propagandistic value of a video finding a discarded AT weapon in a trash heap being from 2011 and not 2017?


#5

I didn’t want to mention that, but it annoyed me, and I didn’t want to sound petty and dismissive because of a few typos, no matter how awful they were.


#6

Much more at https://www.bellingcat.com/


#7

Showing that their video-ists claim of cover-up from 2011 is not exactly credible?


#8

It felt a bit odd that people were doing this searching. It’s a bit like the Futility Closet article on fingerprinting, where most fingerprint cases stayed on file but were never solved until computers came in on the act. Here we have a video, and we want to see whether it, and everything that touched it corresponded with the assumption of a unique source at an original site. The image bit is somewhat like the ‘Building Rome in a Day’ project, rebuilding a 3D model of the centre of Rome from random uncalibrated pictures from the internet, and rejecting the bits that don’t fit. Of course, we want humans to do it too, so we have a yardstick of how well it can be done, but I feel this is something that Google may take on.

Supposing they do? Most of the time, any new tool aids the fakers as much as it aids the investigators. The video faker could try their video out on the new tools, and modify them until they are convincing. However, if the tools only work if they are connected to a gigantic database, and making an enquiry also leaves a data trail…


#9

I believe a certain three letter agency has this fully automated.


#10

This topic was automatically closed after 5 days. New replies are no longer allowed.