If one knows enough to make a prediction of a discovery with any confidence, then one has essentially already made the discovery; only routine checking remains. Equivalently, IF one hasn't made the discovery or, what's more common, doesn't even know that there is some indeterminate discovery to be made, then one is not going to be predicting discoveries with any accuracy.
The mistake historicists make in predicting scientific discoveries reminds me of the joke about the college student who wrote his mother that he had taken a speed-reading course. In the middle of her long letter of reply, she wrote, "Since you've taken that course, you've probably already finished reading this letter." The letter isn't finished until it's finished, and the discovery not made until it's made.
Guglielmo Marconi, who invented the radio, thought it would be used merely as a substitute when a telephone-wire connection was impossible-at sea, for example. In the late 1940s, IBM was not impressed with the commercial possibilities of computers and thought a dozen or so of them would fully satisfy the demand. Bell Labs was initially reluctant to seek a patent on lasers, seeing no use for them in communication, much less in any of the other fields where they've become indispensable.