I am probably biased having run both BIONET and GOS, two of the larger
molecular biology computing central facilities, so you might want to
stop reading now.
Assuming that you are still here, my position on this has always been
one of compromise. I believe that there will always be a place for
central facilities because the less wealthy institutions do not always
have the resources to buy their own equipment and, ***even more
importantly***, hire the scarce expertise needed to provide proper
service. Central facilities can be a cost effective way to assist
them, e.g., on BIONET and GOS a small number of expert systems people
were able to service literally **thousands** of biologists.
For those places that can afford their own facilities, more power to
them. They needn't wait for their jobs to go through the queues, etc.
As usual there is no need to debate only the extremes, i.e.,
completely centralized versus completely decentralized options. The
"market" usually finds a compromise solution.
kristoff at genbank.bio.net