Hypotheses and Insights

The purpose of our study was to test our hypothesis, based on the information provided by the literature and illustrated among our hypotheses that, due to its affordances that favor long-term affiliation and high level of engagement, the Twitch platform would display relatively small levels of revenue inequalities.

Contrary to our initial expectation, we observed that the platform reproduces the typical structure of digital platforms, with power law-like distributions of both revenue and viewership. Even if we found some factors with a potential countervailing effect on inequalities, namely that streamers with bigger audiences face relatively less opportunities to convert their activity volume into revenue per activity unit through the platform, we also showed that top streamers generated more revenue stream while monetizing their content.

To explain these results, we proposed the hypothesis that top streamers had a better ability to retain their audience thanks to a better capacity to adapt to changes. This conjecture appears to be supported by the fact that top streamers face less variability, maintain their audience through the period of observation, have a fairly stable ranking dynamic and play more games from more different genres than others. 

Our results thus question the relevance of an 'algorithm-free' approach such as that of Twitch when it comes to inequality reduction. The real ability provided by Twitch for new streamers to make use of the platform affordances to derive monetary gain from their activity is more limited than one could expect. In the absence of an algorithmic discriminant capable of sharply altering user behavior, the mismatch measured by our analyses is explained by referring to the old concept of information overload. As pointed out by Simon in the pre-digital era, with an excessive supply of information what becomes scarce are time and cognitive resources: this inevitably leads people to choose on the basis of some heuristics that allow an optimization of aims (acquiring the information being sought) and means (saving time). Among these heuristics, one of the most used when facing a dilemma of over-abundance is that of selective trust: the information already socially validated (in our case, opting for Twitch top streamers) reduces the risks of a cognitively expensive and lengthy exploration. This paves the way to a rich-get-richer effect within a relational sphere of influence (namely, that of streamers and subscribers), that can radically alter the expected distribution of resources under less resource-constrained informational conditions.  

Finally, our results raise a more general question regarding inequality reduction on online platforms. As mentioned, Twitch aims at developing highly engaged communities, and therefore made a conscious choice not to leverage upon algorithmic recommendation systems. Such choice, however, leads to low discovery affordances and allows top streamers to capture much of the audience, trend after trend. If authors like Spilker & colleagues note that viewers can actively seek smaller streams with higher social connectivity, it appears that they face a known situation of information overload or, as Schwartz put it, a paradox of choice, where in a situation of abundance of options, the consumer can be paralyzed in front of it and even face detrimental consequences when trying to maximize their choice satisfaction. We therefore can only assume that many, dealing with such platform design, do not try to optimize but rather settle for a “good enough choice” by going for top streamers rather than indefinitely scroll for the perfect broadcast, which would imply giving up considerable entertainment time for less enjoyable search activity. How to balance, then, the inequalities caused by algorithms and those embedded in human choice behavior? Despite the egalitarian rhetoric that has been built around the platform, Twitch does not seem to provide a viable solution to this pressing issue.