So what should you conclude from this?

• VOI is higher when the experiment shifts your beliefs a lot, lower when the expected change in belief is small. For example, praying is sufficiently unlikely to work that it's not worth my time to test it. There are other cases where my uncertainty is high, but I can't think of sufficiently good cheap experiments.

• VOI is higher when you would gain a lot if it told you to change your plans. For example, if you would have taken Adderall without an experiment, and Adderall is expensive, then finding out it doesn't work saves you a lot of money. This is less true for melatonin.