4 ms·
I'm not disinclined to believe that TikTok promotes pro-China sentiment, but this report is far from academically rigorous and has gaping holes in its methodolo
by ramblenode 3y ago
I'm not disinclined to believe that TikTok promotes pro-China sentiment, but this report is far from academically rigorous and has gaping holes in its methodology that would not pass peer review.
* Most notable is that the baseline for determining whether a hashtag has been boosted or deboosted on TikTok comes from Instagram---hardly an unbiased source of ground truth, especially in light of the recent allegations that Meta deboosts pro-Palestinian content and other controversial political content.
* Worse is that there's no statistical analysis showing that generic political hashtags are significantly different from China-sensitive political hashtags between TikTok and Instagram. There is large variance within categories, and even some of the generic political hashtags show a bigger difference between platforms than the China-sensitive political hashtags. Statistics would help determine whether a real difference exists beyond sampling error, but none are to be found in this report.
* No explanation of how the hashtags were determined. They were apparently chosen a priori by the authors, but no criteria for inclusion or exclusion are mentioned.
* Other small details like discrepencies between Fig 2. and Fig 3. in describing TikTok as having half the hashtags as Instagram and Instagram has having double the hashtags as TikTok.
Overall, this report just seems like fodder for the anti-TikTok crowd, not a dispassionate inquiry into how TikTok trends relfect CCP narrative building.
- Morte42 3y agoThe fact that they are using Instagram as a control in the study is hilarious. The standards in these universities are laughable.
- Hnrobert42 3y agoOne does not necessarily follow from the other.
- nneonneo 3y agoAnd that’s why it’s being posted as a white paper, not an academic paper. No peer review required!
- rutierut 3y ago> Overall, this report just seems like fodder for the anti-TikTok crowd, Can confirm, I do belong to that crowd and I did share it after reading it. An hour later, it dawned on me that this report is maybe kinda cool, but really isn't enough to draw any conclusions from and I should not have shared it.
- rakkhi 3y agoTotally. But it still makes front page of HN. It will get shared. Most will read just the title. Confirmation bias of echo chamber. So much bs in western msm propoganda. Do yourself a favour in 2024. Open your eyes. Read widely. Apply logic. Don't believe everything that is force fed down your mouth.
- hsuduebc2 3y agoSure because there isn't any Chinese propaganda specially on western media because they aren't they rival. Literally no evidence. Please do yourself a favor in 2024. Open your eyes. Read widely. Apply logic. There is a geopolitical clash between various players and propaganda is one of the weapons. If you truly believe that there is a western propaganda but not the Chinese one or any other then you are just hypocrite, ignorant or bad actor. Your comment seemed little bit offensive so sorry if I didn't get it's sentiment correctly. But ignoring this fact and using one phony study as a counter argument that it's all bs and propaganda does not exist is bizzare.
- Thorrez 3y ago> No explanation of how the hashtags were determined. > Our presentation of data starts with Uyghurs and Tiananmen Square because those were two topics specifically called out at the March 2023 Congressional hearings, and TikTok’s CEO explicitly denied that in either case were posts on those topics suppressed or in any way demoted on the platform
- ramblenode 3y agoThe Chinese-sensitive political hashtags are somewhat straightforward (they are topics that one would expect the state to be sensitive to), but there should still be a systematic procedure for choosing them to avoid biased sampling and post-hoc comparisons. The problem is more apparent with the controls: where do they come from? How were they chosen? Without this information there is no way to determine whether the categories are truly representative and whether the sensitive topics are really unique.