One of the problems with storage of "in house" parameters is that users
start to assume that they are well tested when this is often not true.
Untested parameters can often be very bad, leading to incorrect results.
My preference would be for individuals to announce that they have
parameters for a certain system and allow people to contact them
if they're interested.
I will also soon add additional topology and parameter information
for more small molecules on my web page. Based on the available
small model compounds, topology and parameter information
can often be generated by analogy. See the powerpoint file from
the Sanibel lecture on my web page for a good example.