There can be no all-unifying theories. Scientists may strive for a uniform explanation of very different experiences, they may construct general theories covering all the other theories as special cases—but, some day, yet another experience is bound to come that won't fit in the seemingly comprehensive theorization. The world is qualitatively infinite, and no theory can describe any of the world's turns.
Science as such is essentially a partial vision of the world, and theoretical science is even narrower. It is only in principle, in the infinite limit, that the whole world could become entirely reflected in science and theoretically described; however, this would mean that science itself ceased to exist separately from the other modes of reflection.
Now, what is the use of the modern integrative initiatives, like the unified field theory? Yes, one can show that all the existing field theories can be derived as special cases from a single theory with enough spatial dimensions. What of that? In principle, this could be predicted from the very beginning, since all the field theories are based on the same logical scheme, and hence they a priori allow combining in a single theory. Obviously, there are different ways to construct such a unified theory, and one will have to seek for observable effects favoring one of the possible solutions. Let's admit that we can overcome the technical difficulties and complete this work. Does it give us a clue to understanding anything except a narrow class of physical processes? Even admitting that every material thing consists of particles and fields, we cannot reduce the whole world to these partial manifestations. Collective motion is qualitatively different from the motion of isolated bodies, and there is no way to compute higher level effects on the only basis of their lower level mechanisms.
Physical theories cannot be simply extrapolated to the whole world, like many scientists do, to impress silly journalists and gain cheap popularity, and thus collect money for serious research. All the talk about the Big Bang, expanding/collapsing Universe, dark matter etc. is nothing but a kind of joke, a mental play without too much pretence, just to see what happens if… Such over-extrapolations are useful in science to clarify the logic of theory and outline the limits of its applicability. They do not have to apply to any physical reality, save the reality of the human thought. When they are presented as absolute truth and the highest achievement of science, this is always an ideologically motivated act, stretching a formally obtained result to a political interpretation. Though such ideological overloading is outside science, former scientists often get dragged into the fraud by psychological manipulation.
In any science at all, a theory is always about something particular, and never a theory of the world as a whole. As soon as the generalizations are becoming that wide, we drift from the domain of science into the realm of philosophy. Since the very distinctions of different sciences originates from the structure of human activity, and hence the currently possible applications, any science can only develop models of limited generality applicable to a very specific range of phenomena, on a single level of hierarchy. As human activities change, sciences evolve into other sciences, and other phenomena need to be described. However general, a science is still about one of the infinity of the possible human relations to the world, and that is what gives it strength and makes science useful in productive activity.
The impossibility of all-unifying science does not mean that a result obtained within one particular science cannot be used in another. However, this is never achieved through mere extrapolation, but rather using a special mechanism of activity scheme transfer. People learn from each other doing different things in a similar manner, and one science can borrow certain tricks from another, adapting them to a different context. With all the superficial similarity, the same method has different sense in different sciences. A formal apparatus borrowed from elsewhere will anyway describe phenomena specific to this particular application area; as a rule, it will require modifications to better describe the new reality, the feasibility of exact transfer being rather limited.
Science cannot produce but approximate models of the world, and no such model can pretend to describe the whole universe, albeit in a single aspect. The existence of theories of different levels of generality makes science hierarchical, but, as in any hierarchy, each level retains its specificity and cannot be reduced to any other level. And, like in any hierarchy, a hierarchical structure ordering theories by the degree of generality can be unfolded in a quite different manner, with formerly "special" sciences appearing to be more general than their former unifications. Of course, such hierarchical conversion in science means scientific revolution—but who can be sure that the present picture of world is already complete and shaped forever?