Myopic Optimization with Non-myopic Approval (MONA) mitigates multi-step reward hacking by restricting the agent's planning horizon while supplying far-sighted approval as a training signal~\cite{farquhar2025mona}. The original paper identifies a critical open question: how the method of constructing approval -- particularly the degree to which approval depends on achieved outcomes -- affects whether MONA's safety guarantees hold. We present a reproduction-first extension of the public MONA Camera Dropbox environment that (i)~repackages the released codebase as a standard Python project with scripted PPO training, (ii)~confirms the published contrast between ordinary RL (91.5\% reward-hacking rate) and oracle MONA (0.0\% hacking rate) using the released reference arrays, and (iii)~introduces a modular learned-approval suite spanning oracle, noisy, misspecified, learned, and calibrated approval mechanisms. In reduced-budget pilot sweeps across approval methods, horizons, dataset sizes, and calibration strategies, the best calibrated learned-overseer run achieves zero observed reward hacking but substantially lower intended-behavior rates than oracle MONA (11.9\% vs.\ 99.9\%), consistent with under-optimization rather than re-emergent hacking. These results operationalize the MONA paper's approval-spectrum conjecture as a runnable experimental object and suggest that the central engineering challenge shifts from proving MONA's concept to building learned approval models that preserve sufficient foresight without reopening reward-hacking channels. Code, configurations, and reproduction commands are publicly available. https://github.com/codernate92/mona-camera-dropbox-repro

翻译：短视优化与非短视审批（MONA）通过限制智能体的规划视野，同时提供长视审批作为训练信号，缓解了多步奖励黑客行为~\cite{farquhar2025mona}。原文提出了一个关键开放问题：审批构建方法（特别是审批结果对实际达成结果的依赖程度）如何影响MONA安全保障的有效性。我们呈现了一个以复现优先的MONA Camera Dropbox环境扩展，它(i)将已发布代码库重构为标准Python项目并包含脚本化PPO训练，(ii)使用已发布的参考数组确认了普通RL（91.5%奖励黑客率）与神谕MONA（0.0%黑客率）之间的显著差异，以及(iii)引入了模块化学习型审批工具包，涵盖神谕、噪声、错误指定、学习与校准审批机制。在跨审批方法、视野、数据集规模及校准策略的缩减预算试点搜索中，最佳校准学习监管器运行实现了零观测奖励黑客，但预期行为率显著低于神谕MONA（11.9% vs. 99.9%），这与欠优化而非黑客行为复现现象一致。这些结果将MONA论文的审批谱系猜想转化为可运行的实验对象，并表明核心工程挑战已从验证MONA概念转向构建能够保留足够预判能力且不重新打开奖励黑客通道的学习型审批模型。代码、配置及复现命令已公开。https://github.com/codernate92/mona-camera-dropbox-repro