Bert Freudenberg wrote:
> "Unfortunately, despite a lot of interest among individual faculty and a 
> few small programs, the widespread institutional adoption necessary to 
> form a viable Sophie 1.0 sustaining community was not happening - due in 
> large part, our inquiries suggested, to lack of interest in supporting 
> an enterprise software application written in Squeak. In the community 
> whose support was most essential to Sophie's survival, everyone wanted a 
> language that was more widely known and used; the largest single group 
> of potential adopters wanted Java [...] The Squeak contractors were 
> understandably unhappy about the move to Java, both because they lost 
> the contract and because they believe in Squeak and want to see it used 
> more widely. We have the greatest respect for their capabilities and 
> their enthusiasm for their community, but our responsibilities to our 
> own institutions, our community, and Mellon require us to give Sophie 
> the greatest possible chance of success."

Great summary for an issue that I've seen come up several times. A 
couple of years back for example NASA evaluated Croquet for a project of 
theirs and decided against it for basically the same reasons: They 
needed people (other NASA scientists and engineers) who would build 
add-ons and extensions and requiring those to learn Squeak was perceived 
a hopeless exercise (and that isn't even mentioning the modularity 
issues Squeak has).

When we faced the same issue again at Qwaq/Teleplace and this time we 
decided to "work around" it by providing the extension APIs in Python 
instead allowing direct access to Squeak. This has served us very well.

This entirely line of arguments is one of the better reasons why "being 
popular" isn't such a bad thing :-)

Cheers,
   - Andreas