I think you're being a bit facetious; but what if it only works when 50% of the grains in the metal are a certain size, the initial temperature is 80.523C, +/- 0.3C, and the electrode is machined to a very high tolerance?
This leads to an interesting question: What's the most "finicky" chemical reaction, for lack of a better expression? I'm thinking of a starting point as seed germination. When you get a seed packet, you routinely expect some percentage of seeds to not germinate. What if the experiment were like germinating a single seed? Of course seeds are more complex than physics experiments are supposed to be... but what if each setup is like a seed? What if the "germination" is 10%? It wouldn't be useful for power generation; but it would still be interesting science.