4 ms·

I don't really see the contribution from this study. The paper looks fine: quality writing, ample research, statistics are probably fine. What I mean is I don't

I don't really see the contribution from this study. The paper looks fine: quality writing, ample research, statistics are probably fine. What I mean is I don't see any non-obvious insight here. Also, I think this study would have been much more interesting on more generic video sites like Instagram or Tiktok, which (I assume) have more balanced creator and viewer populations. I don't personally take Twitch seriously, and so it's hard for me to embrace a social study that samples from Twitch.

cdrini 3y ago
I hear this feedback a lot when people think the results of a research study are "obvious", but anecdotally 'obvious' vs. evidence supported conclusion hold an intensity different weight! But firstly, this can't really be obvious. There's no real way one individual would be able to reach this conclusion without falling prey to their own biases. Especially online where most platforms create user-tailored echo chambers which distort an individual's perception of the entire platform. Having this conclusion be evidence supported also means it can now be built on top of. If I was a twitch employee, this research would be convincing enough to make me put in place interventions on the platform if I was inclined to prevent this kind of behaviour. It's incredibly unlikely someone would put into play an expensive project with potential risks to the platform based solely on "well some people think it's obvious". But with actually numbers and a method to measure progress, that's way more feasible. It also means people who want to do research in this space can use the methods of this study in other platforms or on different slices of the twitch audience, and then they have a baseline to compare against! So overall seems like a decent contribution.
geraldhh 3y ago
couldn't agree more. maybe be starter for more elaborative studies, given the citation festivities within, or just a jab at amazon.
pavon 3y ago
If they had broken out results by category it would have been much more interesting. Instead it amounts to "Pools, Hot Tubs & Beaches" is mostly sexualized women, while "Video Games" is mostly nonsexualized men, which is not informative.
joshspankit 3y ago
TikTok can’t really be researched as a generic site: Everything about the user interactions (scroll/not scroll, how long before scroll, whether a comment was left, whether comments were open, etc, etc, etc) informs the content that is pushed to that user. Simply investigating a sexualized video will make it much more likely to get sexualized videos throughout the use of the app.
- wjholden 3y ago
  I hadn't thought about that, but I think you're right. In fact, any recommendations engine could be a difficult confounding factor in any study of social media. I'm thinking of how the machine might even be aware of the user's IP address or User-Agent, which could bias recommendations.

cdrini 3y ago

I hear this feedback a lot when people think the results of a research study are "obvious", but anecdotally 'obvious' vs. evidence supported conclusion hold an intensity different weight! But firstly, this can't really be obvious. There's no real way one individual would be able to reach this conclusion without falling prey to their own biases. Especially online where most platforms create user-tailored echo chambers which distort an individual's perception of the entire platform. Having this conclusion be evidence supported also means it can now be built on top of. If I was a twitch employee, this research would be convincing enough to make me put in place interventions on the platform if I was inclined to prevent this kind of behaviour. It's incredibly unlikely someone would put into play an expensive project with potential risks to the platform based solely on "well some people think it's obvious". But with actually numbers and a method to measure progress, that's way more feasible. It also means people who want to do research in this space can use the methods of this study in other platforms or on different slices of the twitch audience, and then they have a baseline to compare against! So overall seems like a decent contribution.

geraldhh 3y ago

couldn't agree more. maybe be starter for more elaborative studies, given the citation festivities within, or just a jab at amazon.

pavon 3y ago

If they had broken out results by category it would have been much more interesting. Instead it amounts to "Pools, Hot Tubs & Beaches" is mostly sexualized women, while "Video Games" is mostly nonsexualized men, which is not informative.

joshspankit 3y ago

TikTok can’t really be researched as a generic site: Everything about the user interactions (scroll/not scroll, how long before scroll, whether a comment was left, whether comments were open, etc, etc, etc) informs the content that is pushed to that user. Simply investigating a sexualized video will make it much more likely to get sexualized videos throughout the use of the app.

wjholden 3y ago
I hadn't thought about that, but I think you're right. In fact, any recommendations engine could be a difficult confounding factor in any study of social media. I'm thinking of how the machine might even be aware of the user's IP address or User-Agent, which could bias recommendations.