From zeilberg at math dot rutgers dot edu Wed Mar 2 09:17:11 2011
To: Koutschan at risc dot uni-linz dot ac dot at, manuel at kauers dot de,
Cc: Christian.Krattenthaler@univie.ac.at, andrews@math.psu.edu,
george at netvision dot net dot il, justin at math dot ucsd dot edu,
rstan at math dot mit dot edu,
stavros at math dot gatech dot edu
Subject: George Szpiro's beautiful Article in NZZ
Dear Manuel and Christoph,
Thanks for telling me about :
George Szpiro's beautiful article reporting our recent PNAS article.
George Szpiro told me that he planed to write something, and indeed did a great job!, thanks George Szpiro.
You probably noticed that some of the details, at the end, were not quite right but the global truth, allowing journalistic-license, came out much better than if the local truth would have been presented more accurately.
PNAS gets an undeserved compliment for being more broad-minded than it actually is, and the poor editors of the "Zeitschrifft für symbolische Logik" (does he mean SLC?), implicitly meaning Krattenthaler et. al, looking even more reactionary than they are. Of course the PNAS (and George Andrews) would not have agreed to publish the previous, semi-rigorous, version, whatever the title would have been (even if the title would have been: "A Plan for a computer-assisted proof of the q-TSPP conjecture"), since it would not have been "important enough for the PNAS", a plan for a proof does not constitute a breakthrough according to the (Boolean) conventional wisdom of contemporary mathematicians, and the SLC would have gladly published the second, fully-rigorous version, and even Krattenthaler et. al. would consider the second, fully-rigorous, proof legitimate (although not-as-good-as-a-purely-human-proof), in spite of the fact that it was computer-assisted. But of course, we would not have submitted it there, even without getting mad about their narrow-minded insistence on the title of the previous, semi-rigorous, version, since it would have been too important for SLC, and hence not a good match.
But this is all peanuts nit-picking, and the global truth that beautifully came across in George Szpiro's NZZ article, that many humans are still hostile to computer proofs and conjectures, even if the big picture is much more interesting than most human proofs-from-the-book, turned out better in George Szpiro's simplified version of the local truth.
Another data point of human-centric machino-phobia, was the recent narrow-minded rejection, by Justin Roberts, editor of the journal "Algebraic and Geometric Topology" of the masterpiece by Stavros Garoufalidis and Christoph Koutsthcan. Stavros kindly shared with me what Justin Roberts wrote in his rejection Email:
"It seems like only partial results: it would be much nicer if the authors could prove their conjecture. Without knowing it is true, what they are doing is really just stating a conjecture, and one too horrible to print in full! I don't believe anyone reading the paper will be excited to continue this work (that is the right of the authors anyway) or learn any useful technique for understanding colored Jones polynomials. Therefore I don't recommend publication at this stage of the work."
My dear Dr. Roberts, the truth of this conjecture is at least as certain as any of the human-generated theorems that you (or Andrew Wiles, or any human, for that matter) ever did, and the METHODOLOGY of finding the generators to the ideal, is at least as interesting. Quite a few people would be excited (for example myself) to continue this work, because the methodology of computer-assisted and computer-generated research, for which this is a CASE-STUDY, is our only hope to obtain truly deep results (any purely-human-generated theorem is, a posteriori, relatively shallow, or a human would not have been able to prove, or even conjecture, it.)
Best wishes
Doron