Choosing the Right Weights: Balancing Value, Strategy, and Noise in Recommender Systems

Unfortunately, the chosen weights can often lead to unintended consequences. For example, when Facebook introduced emoji reactions, they gave all emoji reactions a weight five times that of the standard thumbs-up. However, after evidence that the high weight on the angry reaction led to more misinformation, toxicity, and low-quality content, its weight was eventually reduced from five to four, to one and a half, and then finally, to zero times that of a thumbs up [Merrill and Oremus, 2021]. Furthermore, the weights can have a significant impact on content producers who strategically adapt in response to them. For example, leaked documents from Facebook stated that, “Research conducted in the EU reveals that political parties feel strongly that the change to the algorithm has forced them to skew negative in their communications on Facebook, with the downstream effect of leading them into more extreme policy positions,” [Hagey and Horwitz, 2021, Morris, 2021]