[jcrites]

The article with Knuth's program and Doug's reply is: http://dl.acm.org/citation.cfm?id=315654&CFID=329016541&... (Don't have a non-paywall version.)

Regarding the claim about Knuth's program being long, it is worth considering that (1) his program was primarily a demonstration of literate programming style, which contains considerable embedded documentation (2) the program needed to build significant basic infrastructure to support itself, including IO, sorting, case equivalence (including a table with each upper-case and lower-case letter), etc.

Doug's criticism and comparison of Knuth's solution vs. his own is insightful worth reading.