So, what about writing a paper which backport the the S-to-L deduction as suffix-sorting algorithm to the KA's algorithm and compare the reduction version of the KA's algorithm to sais to advertise that it's better? Maybe it would be accepted, then we were to have a new joke :)
I'm not motivated. Because I'm lazy and as a programmer, a joking paper won't promote me. There are also faster code such as libdivsufsort and Archon4r0. c.f.http://code.google.com/p/libdivsufsort/wiki/SACA_Benchmarks.