Risk in science
All really innovative work in science is extremely risky, as one never knows how the unknown far beyond the current research frontier turns out to look like.
Without risk are only small extrapolations of current knowledge, the stuff that feeds doctoral theses, research grant applications, and the large research programs that consolidate frontiers that have already be opened for detailed surveillance.
Risky work is penalized by the current research institutions, (A number of my research proposals were turned down as far too risky and/or ambitions. To be respected in my scientific profession I therefore must work on less risky topics, and do the risky stuff in my spare time, if I want to do it at all.)
Penalizing risky research is perhaps understandable, as high risk implies frequent failure and thus wasted research resources. Public programs that must account for how public money is spent cannot afford that; the taxpayer wouldn't understand.
But this means that really innovative work in science is left to nearly
pure chance - the small chance of stumbling upon something truly new
while working on established research topics.
A possible way out is suggested in the very interesting article Mode of work of the retired scientist by Jürgen Krüger, a retired professor of Neurobiophysics.
Arnold Neumaier (Arnold.Neumaier@univie.ac.at) A theoretical physics FAQ