[pixl97]

>Predicting the probability of water spontaneously combusting, while interesting, provides much less food.

Unless you can make a prediction that water will combust in low energy conditions, in which you can use this combusting water to generate excess power. Then use that power to compress nitrogen into ammonia, and then use that product as fertilizer.

The British show Connections went over how completely different things sometimes 'connect' and bring fruitful new ideas.

The problem space of reality has emergent behaviors that are not (easily?) predictable. Sometimes you have to iterate large portions of it.