Epilogue Under the light of the various examples seen in this book, it seems reasonable to conjecture that, if a topological 4-manifold admits a smooth structure at all, then it might admit infinitely many. While gauge theory was how the door was opened on those vast unexplored realms, it might not be how these will be charted. We have seen that there are whole realms where the Seiberg-Witten invariants cannot help us. For example, the theory is blind on 4-manifolds that admit metrics of positive scalar curvature, on homology 4-spheres (which in particular leaves the smooth 4-dimensional Poincare conjecture with no solution in sight), on all manifolds with hi even, and in general on 4-manifolds that are far from complex. More, gauge theory offers only negative results (as in "two manifolds are not diffeomorphic"). Indeed, the field of 4-manifolds lacks enough techniques for obtaining affirmative results (as in "two manifolds are diffeomorphic"). Looking back, the only affirmative results we encountered came either from ad hoc constructions, from Kirby calculus, or from complex geometry. The field also lacks techniques for building enough examples, which might one day be organized into any sort of classification scheme. We are lost in an ever-growing jungle. Hence the final conclusion of this volume can only be that We know that we don't know. This only makes it all the more exciting ...
