Suppose an application has been filed (pre-AIA) and a prototype was made a few years before filing.
Suppose the claims are rejected in view of 102(b) prior art -- 2 references, both of which came after the invention was made.
35 U.S.C. 103 Conditions for patentability; non-obvious subject matter.
(a) A patent may not be obtained though the invention is not identically disclosed or described as set forth in section 102 of this title, if the differences between the subject matter sought to be patented and the prior art are such that the subject matter as a whole would have been obvious at the time the invention was made to a person having ordinary skill in the art to which said subject matter pertains. Patentability shall not be negatived by the manner in which the invention was made.
How can it be obvious, at the time the invention was made, to combine references that didn't exist yet? It seems that I could swear behind the 102(b) references to remove them from 103(a) analysis.
Certainly seems so. But "whether or not a thing is prior art" seems to trump the simple statement in 103. As NJP1 points out, whether or not a thing is useful as prior art in 103, is found in 102(b), and relates only to filing date.
The typical construction is certainly no worse, in my mind, than using 102(e) art in a 103 rejection as (pretty much only) the US permits. Seems this falls under the same question you asked, "How can it be obvious,
at the time the invention was made, to combine references that didn't exist yet?", if you consider that "did not exist" in the obviousness sense means "was not published/available to OSIA".
The funny thing is, I used to groan about some of these apparent contradictions in (old) 102-103. It only seems the number of them has been increased with AIA. Humbug.