Competitive diffusion of two technologies in a finite market is modele
d through birth-and-death processes. For pure birth processes, final e
quilibria cover the whole range of possible outcomes, and their distri
bution depends both on ex-ante parameters and on initial advantages. F
urthermore, the necessary lead to offset ex-ante disadvantages in diff
usion rates decreases in relative terms with market size. When deaths
are introduced, there is extinction with probability one given infinit
e time. The system may spend large amounts of time around an ''equilib
rium'' point having the same characteristics as final equilibria in pu
re birth processes, however. The implication is that ex-post observati
on of the process tells little or nothing concerning ex-ante parameter
s. If a renewal is introduced at extinction, limit probabilities may b
e obtained. In this case, distribution tends to concentrate on states
in which only one technology is present. Results are robust with respe
ct to assumptions on the functional specification of birth and death r
ates.