In Empirical Algorithmics, researchers aim to understand the performance of algorithms through analyzing their empirical performance. This is quite common in machine learning and optimization. Right now, we would like to know something about the relative performance of quantum algorithms and their classical counterparts based on preliminary data from quantum computer emulators. My concern is that we might see encouraging empirical data that shows quantum computers with better scaling using simulators up to about 35 qubits, but then this advantage will disappear in the future once we have more data. What are the best practices for analyzing relative performance of classical and quantum algorithms in a robust way that gives insight about scaling? Or is this simply not yet possible?