A fruit fly brain looks for criminology it can read

How far can a reader without a subscription get through the criminology literature? Percentages of open-access (OA) articles count what is open, but they don't capture how research actually happens: you start at one paper and follow its references outward. Using Perplexity, we modeled that process with a simulated fruit fly brain. It was built from a 20,000-neuron subnetwork (1,016,133 connections) of the recently published adult male Drosophila connectome (MaleCNS v1.0), modeled as leaky integrate-and-fire neurons. We placed the fly on 100 seed papers drawn at random from Web of Science's Criminology & Penology area (51 journals, 30,654 articles). It walked through reference lists in OpenAlex, choosing each next paper from the readable references in front of it. Diamond, gold, green, and hybrid OA papers were passable, while bronze and closed papers were walls. Of the 100 seeds, 75 were paywalled, so the fly couldn't take a single step. Of the 25 runs that could start, the median run read 158 papers, and 20 runs reached the 800-paper cap. Across all seeds, the fly read 4,022 papers and hit 69,102 paywalls, about 17 walls for every paper it could open. All code (MIT), data, and figures (CC-BY 4.0) are openly available.