[ahrens]

Projections based on pure guesses are useless as information, but useful as a clue to the insights of the entrepreneur. However, projections based on real hard data, beta test growth, virality, those things can be valuable. It all depends on how the entrepreneur arrived at the conclusions.

[eoghan]

Buffett's point is that a projection prepared by a person wishing to sell his business, no matter what data went into the projection, will be inherently biased and is to be treated with skepticism. Buffett would much rather see said data himself. If he can understand it and the business, he'll make his own projections. If he can't, he certainly won't rely on the person with an interest in making the projection look good.