Testing a hypothesis requires that we make various judgments of plausibility along the way. We could not hope to test every conceivable hypothesis. We have to decide which ones are plausible enough to be worth testing.
A further complication is that we can usually derive a consequence from a hypothesis only in conjunction with some additional, auxiliary assumptions. When a hypothesis fails a test, the logical structure allows us to save the hypothesis by blaming an auxiliary assumption. Whether it is reasonable to do so depends on a judgment of plausibility.
What standards should we use in deciding what's plausible? How do we decide whether a hypothesis is plausible enough to be considered and tested? If a hypothesis fails a test, how do we decide whether it is more reasonable to reject the hypothesis or one of the auxiliary assumptions?
Other things being equal, one hypothesis is more plausible than another if it is more consistent with the rest of our knowledge, and if it is simpler.
We should remember that when we introduce a hypothesis to explain something, we are not operating in a vacuum. We have a vast context of background knowledge--beliefs, principles, and theories for which we have accumulated a great deal of evidence. A new hypothesis therefore starts out with a certain degree of initial plausibility that depends on how consistent it is, how well it fits, with that background knowledge. A hypothesis that conflicts with established principles and theories must meet a higher standard of evidence than a hypothesis that does not conflict.
The rule is that, other things being equal, one hypothesis is more plausible than another if it involves fewer new assumptions.
We've seen that an explanation rarely involves a single explanation. A full explanation usually requires a complex hypothesis, involving a number of separate propositions. If these separate propositions are new and cannot be derived from knowledge we already possess, then the fewer the better. The reason is appropriately simple. Just as an adequate explanation must account for every significant aspect of the explanandum, the evidence for the truth of a hypothesis must cover each positive claim it makes. The fewer claims, the less evidence will be required.