If I quickly glance at the "
Computer networks" and "
Computer networking" categories on Wikipedia I can think of a lot of functionality that would fall under the scope of this collaborative library. Making it easier to run sophisticated simulations is a key personal motivator for this project. We could use the articles about specific network equipment (historical and modern wires, routers, etc) as scaffolding and inclusion guidelines (regarding significance) to expand the Modelica library to higher-level objects in this domain. Many computations will remain outside the scope of Wolfram Alpha due to the requirements to limit freely available server-side computation times, so there is definitely the need for efforts to make significantly more advanced client-side libraries. I also think to match the scope of Wikipedia it will take a collaborative, open-source effort.
There is also a lot of data in those articles that should be made computable for purposes outside of simulations. The advantage of this project compared to DBpedia and Wikidata is that you have a lot more flexibility in how to represent data associated with articles when you are using a full programming language (especially a homoiconic one as expressive as Mathematica) as compared to just tables.
Leonid Shifrin on the Stack Exchange site has started writing some code to load packages that people post on GitHub. He provides links to it in his comments in the discussion linked in the first post. I'm not convinced GitHub is our ideal host because you either have to pre-approve contributors to a repository or every new editor causes a new project fork when using Gists. If we use Wikipedia articles as a design scaffold it helps to resolve a lot of potential naming, discoverability, modularity, encapsulation, and significance for inclusion issues. I can see a spectrum of collaboration approaches that might work for this project. Maybe we could just start with a few small utility functions to periodically run all of the tests on the wiki pages and then automatically revert the articles to the last known good state if a break occurs. Maybe we could use a proof-of-work system similar to Bitcoin that shows you have already verified that your check-in works on a recent revision of all of the test cases before a check-in is allowed. I think for documentation a traditional wiki would work. I kind of like the idea of extending the functionality of something like $Path to support URL's for web-based package repositories. Then we could just use Needs with a URL to declare dependencies like we already do for ordinary packages.
I was just at a beach a month ago, but I'm willingly being dragged to another one for a few days tomorrow. However, you mentioned astronomy, so I think I will add a fun, small update this week to support approximate solar irradiance computations for various planets at different times using the inverse-square law and AstronomicalData.
