New @ADL report w/new behavioral data: Exposure to Alternative & Extremist Content on YouTube adl.org/resources/repo…
-9% watched video(s) from extremist channel
-Viewership very concentrated
-95% of views from people high in racial resentment
-Still being recommended to users
-9% watched video(s) from extremist channel
-Viewership very concentrated
-95% of views from people high in racial resentment
-Still being recommended to users
Findings raise serious concerns. However, not clearly consistent with simple narratives re: algorithmic radicalization:
-Most viewers already high in racial resentment
-Most recs to alt/extremist channels shown on similar videos
More consistent with demand finding supply (alas)
-Most viewers already high in racial resentment
-Most recs to alt/extremist channels shown on similar videos
More consistent with demand finding supply (alas)
Caveats:
-Sample not nationally representative
-We used codings of alternative and extremist channels from academic and subject matter experts
-YouTube has changed a lot; findings surely would have differed in the past
-Sample not nationally representative
-We used codings of alternative and extremist channels from academic and subject matter experts
-YouTube has changed a lot; findings surely would have differed in the past
This is the first report from a joint project with the great @annie_y_chen, @JasonReifler, @RERobertson, and @bowlinearl. More to come! It would not have been possible without financial support from @ADL and @RussellSageFdn but the conclusions and any errors are ours alone.
cc FYI (adl.org/resources/repo… or see thread above): @beccalew @MichaelEHayden @CraigSilverman @Mantzarlis @BillAdairDuke @alicetiara @katestarbird @BostonJoan @RebekahKTromble @BrandyZadrozny @asankin @TaliaStroud @j_a_tucker @zeynep @davidlazer @oneunderscore__ @elipariser
• • •
Missing some Tweet in this thread? You can try to
force a refresh
