"They are very good at dancing with data." - Facebook whistleblower. We've discussed this regarding Zuckerberg's misleading testimony on hate speech. But this Markup report is super important to understand prevalence of news brands. Stay with me here. /1 themarkup.org/citizen-browse…
When Facebook finally, under public and government pressure, began releasing a quarterly content report (first in q2 after reportedly suppressing q1), it's used a metric simply showing the # of users who saw links to a website without analyzing frequency of those sites. /2 
any researcher with access to source data would avoid this pretty terrible metric. Since The Markup has developed its own source data from users volunteering their aggregated data for research, they are able to use a much better metric - taking into account frequency. /3
For those in the media business, what Facebook did was the equivalent of forcing an advertiser to only know the number of users (aka "reach") who saw their ad without letting them know the number of times they saw it (aka "frequency"). Put another way... /4
...if you were buying ads,
A) one million people saw your ad campaign one time
-or-
B) 800,000 people saw your ad campaign 100 times
These two things aren't remotely the same in terms of impact and any media researcher worth a high school degree understands this. /5
A) one million people saw your ad campaign one time
-or-
B) 800,000 people saw your ad campaign 100 times
These two things aren't remotely the same in terms of impact and any media researcher worth a high school degree understands this. /5
and a wonky point for those dialed in on the @FTC antitrust lawsuit to break up Facebook, the presentation of these data points showing users as relevant analysis undermines Facebook's argument in the lawsuit that user metrics don't measure intensity. /eof
https://twitter.com/jason_kint/status/1461141847863140357?s=20
ps very well covered, @corintxt @JuliaAngwin. Citizen Browser + technologist + reporter + journalism = wins again. Important work you're doing.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
