Citizens' assemblies—a democratic paradigm where representatives are randomly-chosen citizens—are becoming increasingly mainstream. As these assemblies are used to make higher-stakes decisions, concerns emerge about volunteers manipulating the process of selecting participants. In particular, because selection algorithms must select volunteers based on their self-reported features, a volunteer could misreport their features to increase their chance of being chosen, decrease someone else's chance, and/or increase the expected number of seats given to their own group. While several selection algorithms have been introduced, their manipulability has never been considered. In this talk, we examine what aspects of the selection process, including the selection algorithm, can be changed to limit such incentives. Strikingly, we show that Leximin — an algorithm that is widely used for its fairness—is highly manipulable. We then introduce a new class of selection algorithms that use $\ell_p$ norms as objective functions. We show that the manipulability of the $\ell_p$-based algorithm decreases in $O(1/n^{1-1/p})$ as the number of volunteers $n$ grows, approaching the optimal rate of $O(1/n)$ as $p \to \infty$. Our theoretical results are confirmed via experiments in eight real-world datasets.