However, the scientific method relies on the formulation of hypotheses that are at least halfway coherent and to do that...
No, it does not. If it did, we wouldn't be where we are today.
A hypothesis needs to
1) Describe a mechanism in which observed phenomena function.
2) Be testable.
Those are the only real requirements for a hypothesis to be subjected to the Scientific Method. Coherency or distance from fringe are not required. In fact, most of the more revolutionary ideas are entirely on the Fringe and do throw out EVERYTHING. You know, like the fact the sun doesn't revolve around the earth after all? We take it for granted, but at one time that was throwing everything everyone knew and based information on out the window.
You're missing the point. When something new is discovered, some piece of information disagreeing with a given theory, it is not automatically branded "wrong", rejected, and forgotten in its entirety. After all it IS a theory - it explains some observed phenomena and can produce valid predictions. Unless you can refine the current theory, or find a completely new one that would explain the new bit of information as well as everything the previous theory did, then the conflicting information is going to be treated as an exception to the current theory.
Take for example light - up until late 19th century, the theory of wave-like properties of light was used to explain all of the observed phenomena. Then as more detailed information came through, and light was shown to sometimes behave as a bunch of particles, the previous theory(light is a wave) HAD NOT been rejected. It still described e.g.interference or diffraction.
Also, even when you do make an effort to refine the incomplete theory to take account for the problematic observation, you do not reject all of the postulates of the previous one. It's as if the last century's disovery of quantum and relativistic effects would have been enough to call the entirety of Newtonian laws of motion and gravity false.
Finally, this being a minor nitpick, the change from geocentricsm to heliocentrism was only revolutionary in the sense that it contradicted religious dogma, not in a sense of it being an original scientific idea. It simply changed the frame of reference, to one in which calculations of the motion of celestial bodies is easier. Also, the geocentric approach is still a valid one, and produces testable predictions.
People who scream "down with the wrong theory" seem to think that there can be no incomplete theories in science. They appear to have taken the realisation that science by definition can be wrong to the extreme, ready to tear down and cast into oblivion any theory that is shown to be lacking.
Note that photons are affected by gravity, which is only possible if they have mass.
That is a bold statement. Actually, Einstein's GR theory proposes a different explanation for gravitational lensing - the space-time is distorted around mass.