The content economy on X

Every day is repost day

After Elon Musk acquired Twitter/X, he started monetizing it. There are two types of monetization schemes: 1) subscribers, which work the same way as every other platform (monthly sub, some subscriber only content), and 2) ad revenue, which works in mysterious ways, but some kind of function of engagement with your content. The latter is hard to get right because such systems are open to being gamed in one way or another. 9mmSMG has a good thread on how Indians game the system. Every day they post lame content like this and engage with each other:

Just keep doing this, and then wait for payout day, which is every 2 weeks. When this happens, it's another change to do more gaming:

It seems the algorithm mainly cares about blue checkmarked (verified) users' activity, hence why they all have this. Actual Indian news confirm this is a real thing.

Jonatan Pallesen proposed a revision to the algorithm to limit the effectiveness of this obvious gaming:

Indians figured out how to game the X payout system, and now it's ruined. Hard to see what a good solution is. Perhaps the payout system could have a number of separate areas. For example so that the total payout to Indian accounts is a proportion of the total revenue generated in India.

This modification would make X payouts siloed by country, which would mean that Indian farmers would only get paid out of the pool of whatever money X makes in India, and the same for every other country. This prevents Indians from earning money from, say, American ad revenue. However, this undermines the idea of the open content market. A good English language account based in a small country, say, based in Iceland (population ~350k), can no longer compete on the global market, but is confined to revenue generated in Iceland. This would result in a stark decline in revenue for such accounts. Perhaps for this purpose, one could group countries into Western, East Asian, etc., and allow money to move between the countries in each group. However, the obvious next step for the farmers would just be to move their accounts into the Western group. I am not sure what verification is in place that can prevent this, because it is fairly easy to set up a legal entity in some other country, so the system would probably have to be based on passports. If it is, maybe they will use fake passports. I don't know how feasible this is.

Viral content copying

Aside from this blatant gaming of the system, the more common approach is just to copy other people's content and post it again. Of course, this is the foundation of social media memes in general and is not generally a bad thing. If I see an interesting figure/table in some study or on a blogpost, I may post it on X and maybe this would make me 5 USD if it goes viral. However, reading tons of blogposts and research papers is a lot of work, so the easier approach is just to scroll popular content on social media. When you see something popular (or something that might go viral), you simply download it and post it yourself without a reference to the original. This can be done on the same platform or across. So if you scroll TikTok/Youtube Shorts/Instagram reel and see a popular animal video, just download this and post it on X and hope for the best. This can probably be automated fairly easily. X already has a system in place intended to interact with other users' content, the retweet and quote tweet functions. However, these tend to not make content go viral as much. I also imagine that the algorithm gives some of the money to the original poster, not the re-/quote poster. So from an economic perspective, it is better to repost it as a standalone post.

To give an example. The other day I was reading random Substack articles, and their algorithm suggested me a post by Jean M. Twenge from September 2024 about the rise of transgender identity in the US. It had a stunning plot, so I posted it on X and it went viral (826k views currently). 2 hours later, user John LeFevre reposted my screenshot to his larger audience. He hit the jackpot as Elon Musk saw it and quote tweeted it, giving it 41M views (38M from Musk). Someone noticed and sent it to me, so I told LeFevre that he did a dick move. Rather than admit, he decided to lie about it.

Of course, I did credit Twenge in my thread, it has a link to her blogpost at the end. I had to put it at the end because X's algorithm throttles posts with links in them (another terrible feature). Thus, the overall situation was this:

After this, another big account Billboard Chris had the same idea, and probably copied it from LeFevre. Amusingly, UberSoy also posted the same figure earlier the same day as me, but using his own screenshot that left out the caption, thus making it possible to see which version people copied from whom. He didn't get lucky with the algorithm and his post only got 5k views.

The problem here isn't so much that I am unhappy LeFevre got a bit of money and followers that come from a 40M+ post, but that the system incentives this behavior. Given the ease of posting and reposting content on the internet, there is no simple way to exactly track content across platforms to try to give credit where it is due (in this case, mostly to Twenge who actually made the plot). Furthermore, there is value in content getting reposted in this fashion, as it helps it reach a larger audience. If LeFevre et al had not reposted the screenshot I made of Twenge's post, millions of people would probably have not seen this drastic result (as they did not see it when UberSoy posted it ~10 hours before me and it failed to go viral). Thus, the algorithm did correctly incentivize the sharing of this information, but gave most of the monetary award to the bigger accounts (Matthew effect). X's algorithm seems to have a strong bias towards large accounts unlike Tiktok's algorithm, where content from small accounts frequently goes globally viral. Under X's algorithm, the best way to run a large account is just to scroll the timeline and repost popular things as standalone posts. The invisible hand of the content sharer market will then ensure everybody gets to see the currently most attention grabbing content. If X tracked content on its site and rewarded the first account to post a given image/video, then this problem would be lessened. However, one could simply modify it in some small way to avoid this tracking feature (just as people modify songs/videos slightly on Youtube to avoid copyright claims). Thus it would become a battle of tracking vs. masking content. Would this be better? I don't know.