[Bigdatasur] unique experiment on the 25th of March, YT and Covid conspiracy in the internet (large) peripheries
Claudio
claudio at tacticaltech.org
Mon Mar 23 21:17:34 CET 2020
Dear list!
YouTube claims to take down conspiracy theories with algorithms. meh.
We argue this works, not entirely well, in English, but can't perform
well in other languages, especially the more peripheric.
Is it a form of digital colonialism?
Might happen, perhaps, by implying in a language, you can find more
accurate information on a language different from your own?
Think of the Rohingya crisis on Facebook, with content moderator unable
to read a foreign language. How do you expect they can discern a
scientifically-accurate-information from a scientifically-looking-scam?
There is a lot of complexity in analyzing the YouTube algorithm and
attribute meaning to their biases.
A complexity even for researchers, who understand they need a dedicated
technology, but they can't make it (
https://mashable.com/article/youtube-flawed-radicalization-study/ )
well, luckly:
1) we did it! part of ALEX (is https://youtube.tracking.exposed )
2) we pick videos in Chinese, Arabic, Portuguese, Spanish (and English),
and we'll ask, on the same day, to watch the videos.
3) before watching the video, you've to install a browser extension for
Firefox and Chrome, which collects the 'recommended videos'
4) we'll release the data as anonymized open data, and discuss findings
Here you can find our call: https://youtube.tracking.exposed/wetest/1/ .
Portuguese and Spanish tweets:
https://twitter.com/trackingexposed/status/1237908761710342144
https://twitter.com/narriral/status/1242122974259527684
and Portuguese call do partecipation page:
https://youtube.tracking.exposed/wetest/1-pt/
I'll love to get your comments, and if you think it is worthy, we need
to share it in networks different from the classic 'digital rights
activism'.
.. We know algorithms personalize based on behavior. We should try to
outreach to who has a more diverse behavioral pattern on youtube.
kind regards,
Claudio
More information about the Bigdatasur
mailing list