[okfn-discuss] Collaborative Development of Data

[Rufus Pollock]

John Bywater wrote:
> I thought wikis marked a movement away from support for (and I suspect 
> belief in) the opposition between structured and unstructured 
> information, and to support for semi-structured text. In other words, 
> thought moved on from so-called thinking down vertical lines (the figure 
> of the data worker entering data into a database, the binary machine 
> that structures data by pushing it down, perhaps management as such), to 
> thinking around horizontal lines (the circulations of meaning between 
> brains, the rhizosphere as such).

Interesting, I hadn't seen it quite this way. I think one would want to 
draw a distinction between structure in the data and structure in the 
process of creating the data. I would very much agree that we would want 
to move away from rigidity in the creation/updating of information: 
that, IMO, is one of the thing versioning provides as it allows freer 
write access, easier/more robust renaming, updating etc. However I don't 
think that means one does not want allow people to indicate structure in 
an efficient way *in the information they do enter*.

[snip]

> That's why wikis were aimed at supporting semi-structured data, so 
> people can work on renaming things over time (specifically for The 
> Portland Pattern Repository, to rename emerging patterns of software 
> design). On one hand, having no ability to record denotations prevents 
> production; on the other hand having everything structured in a 
> traditional way allows little room for development (and provides little 
> value before being finished). So the problem is how best to support the 
> passage collaborative works takes from the unrecorded, through the 
> poorly recorded, to the better recorded, and in such a way that the 
> emergence of the form of the recordings and of the recorded matter isn't 
> frustrated.

I would completely agree that one would not want everything structured 
and in most systems you'll always have your 'notes/comments' box where 
you can enter anything. I also agree that the more flexible you can make 
the system the better (including the ability to rename the data 
structures themselves ...). However there is always some trade-off 
between that flexibility and performance -- that's why trying to get the 
domain model right and keeping the focus on that in development is so 
important.

We have problems already thinking how one would efficiently/effectively 
support a domain model containing version objects let alone a domain 
model which itself was versioned -- just think what a pain data 
migration across different database schemas is.

> The new conception was clearly to place the structuring data (meta data 
> for some) inside the consideration regarding data, rather than outside 
> it. This directly leads to the innovative wiki feature of citation 
> before page creation.

That's a really good point i hadn't thought of: the ability to name 
before actually doing anything (to sketch out things easily).

[snip]

> The thing would seem to be to support flows rather than positions, to 
> follow emergence rather than existing structures. Ward didn't put 
> versioning in WardsWiki.

Interesting -- so wikis did not originally have versioning?

~rufus