As Esther-Mirjam Sent aptly suggests, the question is what constitutes 
"proof" in empirical social (or for that matter natural) science. 
 
Rather than talk of "proof" we could perhaps talk of soundness of 
argument.  As so much of the rhetoric of economics literature has made 
clear, this is a complex subject.  Let me offer one take on it. 
 
One thing that many natural scientists do to try to establish the 
reliability of their empirical work is to make sure that it can be 
replicated.  Clearly specifying one's data, the kind of equipment one is 
using, the computer software used to analyze it etc. are all ways of 
letting other researchers retrace the scientist's steps to see whether 
the same result would occur. 
 
One could argue that the process of scholarly reference and citation in 
the history of economics is analogous to the replication process in 
the natural and social sciences.  What careful and accurate scholarship 
does is to say to one's reader "Go back and retrace my sources and look 
at what they said and did and try to understand why I would draw my 
conclusions from this."  Of course such a process can never be "proof" 
nor should we expect all reasonable readers of sources to come to the 
same conclusions.  Nonetheless, careful scholarship, in this sense, 
is necessary (although, again, not sufficient) to generate quality 
arguments and any degree of scholarly consensus. 
 
To hold ourselves to standards of "proof" beyond this is a sure path to 
permanent disappointment.  Vain hopes for extra-human standards of 
proof need to be replaced with realistic hopes for human standards of 
discourse, scholarship, and scholarly consensus. 
 
Steve Horwitz 
St. Lawrence University