Maarten: and the great advantage of having our own source code and compiler is that we can tweak, extend and alter it as desired - given sufficient health, that is. (But programmatically, Cort, what you specified is no rocket science... and much of what is needed, such as code for graphs, charts (bar and pie a.s.o.), plotting and statistics simply exists on the net, for many open source programming efforts, gratis, only in need of tweaking for one's own ends...
Cort: I agree that there is a great advantage to being able to tweak, extend and alter as we wish - a great advantage.
How about the statistics end - if say we were going to do rigorous statistical analyses (patterns recognition, anova) - is that kind of stuff available on the web?
I'll pass your ideas on.
I am looking around - and there is a lot, in various forms and programming languages.
Also, there are many alternatives not yet mentioned. Here are two:
1) writing a shell for an existing good stats program of a university (there are such, tested and free)
As to (1): The advantage is that you have a tested statistics package you don't write, with an environment of your own for it. (Take a look e.g. at MicrOsiris, that's shareware at the Univ. of Miichigan. Just found it, but that sort of thing is possible.) Writing a shell can be done in virtually any programming language (and may help much keeping it together and combining things) while specifically, and in general
As to (2): Lots of things are possible with JS and indeed have been programmed in it and are available. Also, it is not a difficult language. Restriction: browser bound. (I have programmed some in it, but not much, basically because of ME and the restriction.)
Anyway... one concern is that I think you should not lay out money until the possibilities have been considerably better discussed on the PR Forums. There is really much that is possible and available for free and/or in open source, and personally I much prefer that, for various reasons, such as tweaking and extending.
Finally, in fact a whole lot can be done also in spreadsheets, often more easily than programmatically, both database-wise and statistically.