The value of the Cyc project can be assessed on the basis of its theoretical and
applied uses.  To be practically useful--that is, to have applied value--Cyc should
be able to supply correct answers to a wide range of queries; further, Cyc should
take inputs and generate outputs using a natural language.  At minimum, Cyc should
have useful applications to real problems.  A cursory look at cyc.com inspires
pessimism on each of these points.  It's unclear that *anyone* finds Cyc to be
practically useful.
  Cyc could also be worthwhile in terms of theoretical interest.  Cyc's goal is
lofty enough that its failure to be practically useful could be mitigated by
incremental theoretical advances.  And the Cyc website does list dozens of
publications, most released in the past decade.  However, the Cyc website also
shows a major theoretical shortcoming of the project--even though Cyc addresses
potentially interesting theoretical problems, other problems are probably more
interesting.  The website asks, "As statistics-based search engines become
increasingly powerful, is there still a role for common sense knowledge and
reasoning?"  Doug Lenat apparently believes that there is.  Still, the promise of
statistically-motivated text translators (Google Translate) and expert systems
probably outweighs the (non-negligible) promise of Cyc.
  Another point pertaining to Cyc's theoretical merit:  it's not clear that
domain-general encyclopedias are actually tractable.  An influential thread in
evolutionary psychology argues that the human mind is highly modularized (or--a
weaker argument--that the human mind has many domain-specific modules).  Why stray
from the path beaten by human evolution?