Here we are using the distinction between ordinary language and scientific language to establish a clear difference between the experimental and theoretical: experimental (objective) facts can be stated in ordinary language, theoretical assertions cannot. Most scientific entities can't be placed in either camp, but are a mixture. This means we don't have to follow the positivist's line -- that there is a division, but since anything can be p-doubted it can't be objective, so there is no fact, just theories. We also don't have to follow the realist and pretend that there is no distinction at all, so everything in science is real. Another variant is possible here: someone could claim with the realist that there is no clear distinction between theory and data, but infer from this that everything (including all common-sense beliefs) is just a theory. You might want to call this idealism/positivism, but it has been defended as scientific realism. (P. M. Churchland, Scientific Realism and the Plasticity of Mind , CUP, 1979)
It is interesting to note that one can see the two
ends of the spectrum in action when one watches scientists'
reactions to a great new experimental result. There are
two typical reactions.
(i) This is a discovery about the nature of the entities of our theory.
(ii) This is a discovery that our theory uses the wrong entities.
Case (i) shows scientists thinking in realistic terms -- eg the discovery that isolated quarks were never observed. The response was: sure they exist, they are just ``confined'' so we can't see them! Case (ii) shows them thinking positivistically -- eg the discovery of interference effects in light led to the abandonment of the corpuscular hypothesis.
 Wittgenstein is not saying that ordinary people's beliefs are what constitute reality, but that ordinary-language criteria of realness (meaning of the word ``reality'') must be used in deciding what is real. If another criteron is used then another word than ``real'' should be used. The same sort of reasoning applies to words like ``mind'': the mind-body problem can then be seen as the problem of how mental concepts and physical concepts fit together in ordinary use. If we wanted to discover the essence of mind then we would find that anything fitting the concept: ``essence of mind'' could not meet ordinary language criteria for realness.
 Such claims are often made by fields whose own lack of predictive or technological success leads them to claim philosophical significance, eg suggestions that artificial intelligence might illuminate the structure of thought, or that formal linguistics might solve the empiricism-rationalism debate (see, eg, N. Chomsky, Language and Responsibility, Harvester Press, Sussex, 1979, pp 81--99).