3 ms·

Reposting my comment from the previous discussion that didn't get quite so active: https://news.ycombinator.com/item?id=38736001 https://news.ycombinator.com/it

by yorwba 3y ago

Reposting my comment from the previous discussion that didn't get quite so active: https://news.ycombinator.com/item?id=38736001 https://news.ycombinator.com/item?id=38736001 Their methodology (comparing hashtag counts) does not make sense for detecting whether content is systematically promoted and demoted to fit some agenda. I can make one video, have it artificially promoted to reach a million viewers, and that's a hashtag increment of 1. Or I can make ten videos that are all unfairly demoted and therefore barely get any views, and that's a hashtag increment of 10. The discrepancy for Kashmir-related hashtags makes me think there are a lot of Pakistanis on TikTok who never use Instagram...

croemer 3y ago
That's right, to a first degree, it just seems to be differences in user bases. There may well be nudging going on, but it's hard to discern and would require more complex econometric models. Of course it's possible that users self-select: those who are pro-Tibet won't post on TikTok as they know that content won't work there. Would be good to compare other western platforms: is there similar discrepancy between Twitter, Instagram, Facebook, Bluesky, Mastodon? I'm sure there will be some discrepancies.
aragonite 3y ago
> The discrepancy for Kashmir-related hashtags makes me think there are a lot of Pakistanis on TikTok who never use Instagram... Also, India banned TikTok, so there's no Indian TikTok accounts pushing alternative viewpoints/hashtags. It's not hard to imagine how that might have led to the observed results, and it's a little odd that ctrl+F-ing "India" produced no result in that report.
theteapot 3y ago
Article states: > ... we replicated TikTok’s own methodology used in its November 13, 2023 letter. So Tiktok's own methodology doesn't make sense? Goes on to say: > NCRI analysts examined the volume of posts on TikTok by hashtag, using the platform’s own ads manager portal. We performed the same analysis using Instagram’s explore feature. Are your 10 shitty videos aren't going to show up on the ad manager portal?
- yorwba 3y ago
  The referenced letter by TikTok https://newsroom.tiktok.com/en-us/the-truth-about-tiktok-hashtags-and-content-during-the-israel-hamas-war https://newsroom.tiktok.com/en-us/the-truth-about-tiktok-has... contained this caveat: > In addition, the number of videos associated with a hashtag, alone, do not provide sufficient context. For example, the hashtag #standwithIsrael may be associated with fewer videos than #freePalestine, but it has 68% more views per video in the US, which means more people are seeing the content. And, some hashtags are newer (e.g. #standwithIsrael) while others are more established (e.g. #freePalestine)–the vast majority (9 in 10) of videos tagged #standwithIsrael were posted in the last 30 days in the US. A difference in views and posts is expected. I would assume that my 10 shitty videos (fyi they're actually very good, just unfairly withheld from viewers) would nonetheless show up in the "number of posts with this hashtag" provided by TikTok's ad manager portal, but I haven't tested this.

by yorwba 3y ago

croemer 3y ago

That's right, to a first degree, it just seems to be differences in user bases. There may well be nudging going on, but it's hard to discern and would require more complex econometric models. Of course it's possible that users self-select: those who are pro-Tibet won't post on TikTok as they know that content won't work there. Would be good to compare other western platforms: is there similar discrepancy between Twitter, Instagram, Facebook, Bluesky, Mastodon? I'm sure there will be some discrepancies.

aragonite 3y ago

> The discrepancy for Kashmir-related hashtags makes me think there are a lot of Pakistanis on TikTok who never use Instagram... Also, India banned TikTok, so there's no Indian TikTok accounts pushing alternative viewpoints/hashtags. It's not hard to imagine how that might have led to the observed results, and it's a little odd that ctrl+F-ing "India" produced no result in that report.

theteapot 3y ago

Article states: > ... we replicated TikTok’s own methodology used in its November 13, 2023 letter. So Tiktok's own methodology doesn't make sense? Goes on to say: > NCRI analysts examined the volume of posts on TikTok by hashtag, using the platform’s own ads manager portal. We performed the same analysis using Instagram’s explore feature. Are your 10 shitty videos aren't going to show up on the ad manager portal?

yorwba 3y ago
The referenced letter by TikTok https://newsroom.tiktok.com/en-us/the-truth-about-tiktok-hashtags-and-content-during-the-israel-hamas-war https://newsroom.tiktok.com/en-us/the-truth-about-tiktok-has... contained this caveat: > In addition, the number of videos associated with a hashtag, alone, do not provide sufficient context. For example, the hashtag #standwithIsrael may be associated with fewer videos than #freePalestine, but it has 68% more views per video in the US, which means more people are seeing the content. And, some hashtags are newer (e.g. #standwithIsrael) while others are more established (e.g. #freePalestine)–the vast majority (9 in 10) of videos tagged #standwithIsrael were posted in the last 30 days in the US. A difference in views and posts is expected. I would assume that my 10 shitty videos (fyi they're actually very good, just unfairly withheld from viewers) would nonetheless show up in the "number of posts with this hashtag" provided by TikTok's ad manager portal, but I haven't tested this.