Martin said:
...but there's a known solution for RFCs - the RFC search engine
http://www.rfc-editor.org/rfcsearch.html
This takes an RFC number or a word/phrase in the RFC title and digs out
the referenced RFC plus a chain linked forward and backward to show all
the RFCs your taget obsoleted as well as those that make it obsolete.
Martin disproves this.
Google disproves this.
Sure, but using the search engine is easier for both author and
researchers.
... I now return you to the scheduled program.
Good search tools simulate intelligence better than cognitive
algorithms do, under many circumstances.
http://xkcd.com/903/
Suppose you write seventeen articles a year online, "you" being
any arbitrary entity such as you personally, a committee, a reporting
system or whatnot. Say the articles have a half-life of interval until
a link needs updating, along the lines that sun.com and RFC articles do.
You would need to review your own documents over years to keep them
up to date. This is what Oracle has done. Given finite resources,
there is a limit to how many articles you can update per year, and
still maintain your output of new articles. It stands to reason that
some articles will not be updated at any given time.
Regardless of the actual numbers, it's clear that link maintenance
will require increasing energy over time, as the mass of articles grows.
Consider instead a search solution along the lines Martin mentioned
for RFCs. Links can still be maintained, according to a triage system
of need and benefit. But they need not be, given a search system that
elicits the same connections on demand.
The search system complexity and deficiencies are completely independent
of the article base, and presumably bounded. Instead of an increasing
energy investment in link maintenance, lacking opportunity for innovation,
you have steady energy investment in search enhancement, with ample
opportunity for innovation and increased value.