Share this page!

Enter URL or ID to Unroll

You can paste full URL like: https://x.com/threadreaderapp/status/1644127596119195649
or just the ID like: 1644127596119195649

How to get URL link on X (Twitter) App

  1. On the Twitter thread, click on or icon on the bottom
  2. Click again on or Share Via icon
  3. Click on Copy Link to Tweet
  4. Paste it above and click "Unroll Thread"!
  5. More info at Twitter Help
Mark Ledwich Profile picture
Mark Ledwich
@mark_ledwich
Coder and pivot table worker. Insatiable reader and podcast listener. Building a forecasting platform for the Swift Centre.
no Profile picture 1 subscribed
Dec 4, 2020 • 18 tweets • 5 min read
I am one of the authors of this "recently published research". The problem is not with our classifications, it's the way the labels were converted into far-right and far-left.

This paper is actually very good, but is let down by the classifications.

🧵 They classified channels using our old dataset (Ref 23) and ones from the Auditing Radicalization Pathways study (Ref 20).
arxiv.org/pdf/2011.12843…

They translated labels from these studies into their categories in a really strange way. Image
Dec 31, 2019 • 5 tweets • 3 min read
1/ I found 4 articles on TubeFilter about YouTube recommendations. Every 👏 single 👏 one 👏 is reporting research with the same "fatal flaw" - anonymous data.

let me show you 🧵 2/ Pew report they used anonymous data
Dec 29, 2019 • 8 tweets • 3 min read
1/ There are two main criticisms about the study that I should have anticipated better.

a) anon recommendations could be very different from personalized ones
b) the NYT and others were reporting on 2018 radicalization, yet we only analyzed late 2019 recommendations 2/ Anonymous recs (even when averaged out over all users) could have a different influence compared to personalized ones. This is a legit limitation, but one that applies to all of the studies on recs so far.
Dec 28, 2019 • 5 tweets • 4 min read
1. I worked with Anna Zaitsev (Berkely postdoc) to study YouTube recommendation radicalization. We painstakingly collected and grouped channels (768) and recommendations (23M) and found that the algo has a deradicalizing influence.

Pre-print:
arxiv.org/abs/1912.11211
🧵 2. It turns out the late 2019 algorithm
*DESTROYS* conspiracy theorists, provocateurs and white identitarians
*Helps* partisans
*Hurts* almost everyone else.

👇 compares an estimate of the recommendations presented (grey) to received (green) for each of the groups:
Aug 27, 2019 • 12 tweets • 3 min read
1/ The “Auditing Radicalization Pathways” study is worthwhile looking at. arxiv.org/abs/1908.08313

They use comments to correlate users pathways across time. This is clever and it's a great way to get at the true movement of viewers between channels. However... 2/ I read the paper and come away baffled about why their study is considered the evidence for supporting the rabbit-hole/alt-right-radicalization narrative. Let me show you the important parts
Jun 30, 2019 • 26 tweets • 9 min read
The context of carines tweet is an incident with Antifa and @MrAndyNgo today.

This is a video of it

And the photo is the aftermath.

Caroline no longer extends humanity to him, or she is not aware of what happened. Image @MrAndyNgo Nathan knows what happened, but doesn't think andy can be a victim of serious assault because he is "relentlessly baiting and harassing antifa"