Stephan,
[stuff about increasing Smalltalk's public profile snipped]
These are all excellent points. I would like to add one more. The visibility of Smalltalk would be hugely increased by an 'open source killer app' If Sendmail, Apache, or Netscape had been written using Smalltalk, a lot more people would have heard of the language and considered using it. Actions often speak louder than words, although I agree we need both.
At university it frustrates me that the lecturers all feel they have to include a section on Smalltalk, but normally in a historical context. "Smalltalk, well no-one uses it" was one notable quote.
With programming languages, mindshare is everything. It's difficult to explain this in a few short lines, but the principle of 'emergent properties' applies. Richard Gabriel has argued similarly in the context of c and Lisp. This basically says that a language is a bit like a virus. It doesn't matter how bad a language is, if you can get enough people using it, the language will 'improve itself'. This happens directly as a function of the number of users. The fundamental prerequisite is that the virus must propagate itself simply and easily, and this is where Smalltalk has fallen down in the past. With cheap and free Smalltalks now plentiful, things are better.
If Smalltalk is to become a more effective 'virus', programs must be able to be passed along and replicated without a 5-10 MB image. A virus is only as plentiful as the number of hosts it can infect. Practically, this will mean users should be able to hack together small utility programs and pass them along and run them without moving Megabytes across the network.
I suspect Smalltalk makes an excellent 'Web Glue' along the lines of Perl, but as you say, it is essential to increase the user base.
Peter
squeak-dev@lists.squeakfoundation.org