Resource pooling improves system efficiency drastically in large stochastic systems, but its effective implementation in decentralized systems remains relatively underexplored. In this paper, we study the problem of incentivizing resourcing pooling when agents are self-interested, and their states are private information. Specifically, we study a standard multi-server queueing model in which each server is associated with an M/M/1 queue and aims to minimize its time-average job holding and processing costs. Our primary motivation is applications in the design of decentralized computing markets (potentially implemented on blockchains), among others.

Experimentation is prevalent in online marketplaces and social networks to assess the effectiveness of new market intervention. In an experiment, the platform exposes a (randomized) group of targeted users to the new feature or, equivalently, assigns each user to either the treatment or the control group. The platform then uses the resulting outcomes to estimate the new feature's total market effect, i.e., the difference in total user outcomes if the feature is introduced to the entire market.