Summary
Crowdsourcing websites let many online workers join contests, but sometimes too few workers enter important contests or don’t put in much effort. The authors study how these platforms can better suggest contests to workers so more people join and compete fairly. They developed a method called GRAF that arranges workers by scores to guide who joins which contest, making sure workers don’t regret their choices and the platform benefits. Because figuring out these scores is tricky, they created LLMScore, a tool powered by large language models that learns to pick good scores, even for new kinds of contests. Together, these tools help both platforms and workers by improving contest participation and satisfaction.
crowdsourcingTullock contestsself-selectionworker regretplatform utilitygreedy algorithmlarge language modelsevolutionary frameworkscore ordering
Authors
Nguyen Thach, Hau Chan, David Parkes, Karim Lakhani
Abstract
Crowdsourcing platforms coordinate large pools of online workers who strategically choose which contests to enter and how much effort to invest. This self-selection can leave important contests with too few participants or too little effort, while workers may regret entering contests that leave them worse off than available alternatives. We study how platforms can recommend contests to workers using self-selection in Tullock contests (SSTC), a two-stage model in which workers first choose contests and then compete within them. We introduce GRAF, a greedy polynomial-time framework that constructs self-selection outcomes by ordering workers according to a score vector, with guarantees of zero worker regret and platform optimality in special cases of SSTC. Because effective orderings are difficult to design under worker heterogeneity, we propose LLMScore, an LLM-driven evolutionary framework that automatically designs GRAF's scoring algorithm. LLMScore addresses two challenges: jointly optimizing platform utility and worker satisfaction, and evaluating worker regret when exact computation is intractable. Trained only on small instances of one setting, it transfers to larger and structurally different settings; moreover, its output is human-readable code that platform operators can inspect and modify. Across 1,000 synthetic instances spanning four settings, GRAF with LLMScore consistently achieves high-quality, often near-optimal, outcomes with low worker regret, benefiting both platforms and workers.