Louis:

I have been careful to use the qualifier of common practice which one can quantify as the aggregate of system type (email, document editors, spreadsheets, visual analytics (maps)  etc )clusters on desktops.

Of the kinds of information types you list, how many can not be done using relational dbs (SQL) and common three tier systems and how many users need to do those things?

Of those that are left, how many are best done with XBRL for some given metric of better?

Of those, how many implementations are there in open source and how many are serviced?

len

-----Original Message-----
From: Louis Matherne [mailto:matherne@o...]
Sent: Friday, August 28, 2009 9:19 AM
To: xml-dev@l...
Subject: RE: XPointer is dead. What about XLink?

Len,

It’s certainly true that “humans read the stuff” and largely expect to receive it in a manner that’s easy to use – many still prefer the print version.

Much of this is dependent on the nature of the information. Information that is largely narrative in nature is less dependent on links to other resources to be complete (although I can spend hours on hours reading something online that takes me down many other paths through links).

Contrast this with financial information that is highly structured and metric based and you might come to a different conclusion.  There may be a lot of words with financial information but those words are usually carefully selected to report a required disclosure.  As such, there is a vibrant global industry that creates, aggregates, disaggregates, normalizes, massages, analyzes, and redistributes this information.  At the end, the ‘human still needs to be able to read this stuff” but they aren’t reading a book.  They are reading comparative metrics and other analyzes in an application (think spreadsheet here) that is far more dependent on the ability to pull together multiple sources of information.  In this scenario, XBRL is just the plumbing that the end user cares little about, but this plumbing ties it into the city utilities as opposed to a well and septic system.

Louis Matherne

Those technologies don’t get traction I think because for most common practice uses of XML they aren’t that useful.   Most useful related office data is stored relationally.

As Tim Bray observes, XML thrives as bits on the wires.  As a representation most programmers can schlep in and out of RAM as documents, it’s ok.  Most common uses can use microformats for the longer lived semantically loaded bits because microformats track the average information density of tables.  Wise URI management does the rest.   If you need a doc of links, you usually have a table of contents or the reverse index which is fine because those are where some of the ideas in XPointer and XLInk originate as well as glosses/annotations.

Very complex abstractions of semantically loaded data can be fascinating to think about, but what practical desktop uses are made of them?  I’m not saying they don’t exist, but where they exist in the information ecosystem, what other systems are their dependent neighbors?

Humans read the stuff.   Documents work.   As a result, the most frequent user of the system doesn’t use XPointer or XLink.  They don’t care.   No care: no market.