forgot to mention that the understanding of the problem and its
modelling
(allowing for extrapolation) is also the only way to reliably
determine
a) the severity of the problem (the real issue is that variation in
the
estimations result in huge difference over periods of X years)
b) by when the existing routing and addressing system is
potentially going
to result in serious problems (how much head-room is left and over
which
period can this evolution be sustainted vs routing system
capabilities and
their own evolution)
c) which causes are the most impacting and in which conditions they
occur
(noticing that some of these causes are constraints, "future" solution
would have to accommodate them)
the point is that (even a partial) an answer to these question can
lead to
different levels of investigations in terms of solution space:
algorithmic, new routing paradigms, etc. but also routing protocol
design/engineering, system engineering, etc.