An hypothesis or conjecture introduced to solve an hitherto insoluble problem.
One wonders if some hypotheses very far away from common sense and not verifiable, introduced for example in some scientific theories, are not simply a way to "get out of the trouble". Changing the basic conditions of some phenomena and/or theories does not seem a very satisfactory option. (See John BARROW: "Is nothing sacred?", 1999).
These doubts were already aired by the french philosopher and epistemologist G. BACHELARD (1884-1962) ("La Philosophie du Non", 1949).
It is however doubtless that the very different abduction (PEIRCE)(1839-1914) is highly valuable in terms of creativity. Without the resulting "educated guesses", knowledge in general, and science in particular, would remain finally bogged down into static conservatism.
The difficulty is to differentiate abduction at its incipient state from useless imagination.
An example of genial abduction was WEGENER's (1880-1930) idea of continental drift, whose value was only recognized various decades later.