🚀 go-pugleaf

In article <1992Oct23.132146.20784@eua.ericsson.se>, joe@erix.ericsson.se (Joe Armstrong) writes:
|> How about:
|> 
|> 	language L1 is move expressive than L2 for solving the problem P iff:
|> 
|> 	#(min_prog(L1, P)) < #(min_prog(L2, P))
|> 
|> 	# is the number of characters in the program
|> 
|> 	min_prog(L, P) is the shortest algorithm in langage L which solves
|> problem P.
|> 
|> 	Then L1 is more expressive than L2 if the above criteria is met for
|> all problems which I am interested in solving!

Bad metric.  A better metric would be that L1 is more expressive
than L2 for problem P iff:

    time(L1,P) < time(L2,P)

where `time' is a measure of how long it takes a knowledgable programmer
to produce a debugged solution.  Actually, a weighted average of execute
time and programmer time (weights determined by cost of each) would be
a better measure still.

L1 is generally more expressive than L2 if the above criterion is met for
the majority of programs P.

The textual length of a program is irrelevant except to the extent 
that unnecessarily verbose or unnecessarily terse languages impeed
programming effectiveness.

-- 
J. Giles