Algorithmic Curation Impacts Media Exposure in Twitter Timelines

I'll go for "duh!" for $100, Alex. Any curation affects media exposure (algorithmic or not). By definition, curation means there's a filter, and a filter implies that some things won't get through. A bit of a shame to see my tax dollars funding do-nothing studies like this one.

Study also seems weirdly partisan: for example, "During the COVID-19 pandemic, informative tweets about fatality rates and preventative measures are often more important than tweets related to political events"—important to whom? I have a feeling Republicans were more interested in the latter, while Democrats were more interested in the former.

"Adding tweets to user timelines is an especially opportune mechanism to mitigate the potential negative impacts of social media platforms. Added content could be curated using more traditional editorial values, in service to both journalists and Twitter users"—authors seem to be confused as to what Twitter is. It's NOT a newspaper, and it's absolutely not supposed to have the same editorial standards as a newspaper. Twitter has echo chambers by design. From a product standpoint, people happen to like them. So the algorithm reinforces this.

Is this healthy? I don't know, probably not, but neither is alcohol or tobacco. Maybe all social media apps should have a "surgeon general's" warning. Which seems like a way more actionable course than making social media apps use journalistic-style editorial standards (how do you even enforce that on something like TikTok?).

It's a wonderful article, I'm glad someone is finally quantifying what has been evident to many people for a long time. Namely that Twitter is a harmful tool that encourages extremism and widens the political divide in America.

One can only hope it will finally be regulated, and not with attacks on free-speech like "misinformation boards" but through transparency and fairness in the way the algorithmic feeds are implemented.

One of the articles referenced in this paper looks pretty interesting:

https://sci-hub.se/10.1080/1369118x.2019.1637447

> Bruns, A. (2019). "After the “APIcalypse”: social media platforms and their fight against critical scholarly research."

Here's a sample:

> "In early 2019, Facebook also took explicit steps to disable the functionality of a browser plugin promoted by the investigative journalism initiative ProPublica that would capture the Facebook ads encountered by plugin users, as well as a number of similar other tools. Using direct Web scraping methods that did not rely on API access, the plugin sought to generate independent, publicly accessible observational data especially on political advertising practices in the United States and elsewhere; citing a need ‘to keep people’s information safe’, Facebook responded by deliberately obfuscating the HTML code served by its Website in such a way that ads could no longer be immediately recognised as distinct elements of a Facebook page (Merrill & Tobin, 2019)."

The social media business model is still based on selling people's information to third parties, but they don't want independent scrutiny of their propaganda practices (aka content curation, labeling and promotion of 'trusted sources', etc.)

It's "duh" to HN users, not so much to the normies. So studies like this are still needed, unfortunately

At least for Twitter, I think this still happens if you move everyone into a list and only view the list.