[Geowanking] geo microformat BOF session at Where 2.0

[Anselm Hook]

I do see reasons why people like microformats:

1) They can take advantage of existing transports without developers
having to extend or re-work their engines.  Typepad (still!) and
Delicious are two services that do not transport geo:long and geo:lat
as rss tags but which would benefit from such an ability.  This is
actually a pretty compelling argument for at least the existence of
some 'microformat' or 'abuse the pipeline' type of metadata.

2) It is seen as a simplification.  It is conceivably easier to embed
some enhanced metadata and hints in a page rather than having a
machine try to discover a feed associated with said page.  This
'unified page' is seductive and appealing to humans; we are
subconsciously drawn towards perceived simplicity even if it is
hellacious for machines to deal with.

3) It fits the hacker ethos of 'tools not rules'.  Developers often
like to exercise a whole system and then refine the ugly snippets
later. Of course there tends to be legacy effects as quick hacks
become institutionalized.  We have the now entrenched idea of
mime-types such as .jpg and .txt for example.

On the negative side I see a couple of things:

1) Microformats excel at nothing and do a number of things fairly
poorly.  As far as a machine scannable notation they are likely to be
filled chockful of human errors unless produced by machines in the
first place.  And as a human notation they are basically irrelevant;
even annoying.

2) The fact is that they straddle two ecological niches handled better
by other things:

  <----------------------------------------------------------------------------------->
  RDF/XML            Microformat              Human Consumption

Given no niche that they dominate it seems hard to succeed.  When a
developer has to code up a solution (as opposed to simply viewing or
parsing something that already exists) they lean towards the simplest
solutions for them to implement; and there already are simple
solutions.

3) In the context of geo applications specifically I see a microformat
approach eventually having the same kinds of problems as GML.  A
grammer with domain specific semantics that realizes that it has to
reach beyond its domain to embrace other domains...  It is hard to
imagine something simpler than RDF as a transport notation for
semantic data.  For example a Microformat can trivially represent
geo:long and geo:lat but what happens when you want to also represent
tracklogs or polygons or raster coverages?  How about next year when
we start to do rdf3d type things; transporting 3d models of buildings
or joint limits on a digital puppet or any of the kinds of future uses
we can imagine for RDF?  The microformat ideology will fall over and
we'll be back to the same criticisms that were levelled against RSS
and GML -> needless domain specific goop enshrouding a simple and
basic need to for a implement-once grammer that can transport
semantics.

4)  'Microformats' seems to have become a hydra-like phrase that is
pretty hard to define let alone kill.  It largely survives by
attempting to escape definition.  For example wasn't the original idea
of microformats to specifically introduce new kinds of HTML tags that
could be reserved for machine use?  Now it seems that 'microformats'
implies any semantic data that a document can provide as hints to a
machine scanner.  This includes any portions that are human visible
such as the description text of a blog entry.

All this said - I am a still fan of embedding geo:long and geo:lat
tags in the description area of a post.  Is that a microformat?  I
have no idea.  I just want to be able to move geo data around today
without waiting for people to adopt geo ideas.  As Mikel points out -
we have this beautiful RDF castle that remains largely unoccupied -
microformats could be a bridge.

In that light it's probably reasonable to push a specific microformat
idea and see if it flies.  Human's haven't proven to be really
'rational' in a machinistic platonic sense; the contours of our
applications seem to follow a subjective and human sensibility that is
hard to predict...  who knows - we might all be reading 'microformat
hacks' next year while busily retrofitting our apps to be compliant
with the latest W3C microformat standard.

- a


On 7/6/05, Mikel Maron <mikel_maron at yahoo.com> wrote:
> RDF is a worthy goal, but how to get there? To me, microformats are being
> thought out with an emergent, user focus. Most applications in the wild do
> not enable the creation of arbitrary structured data, while users want to
> contribute in different vocabularies and capabilities, however messy that
> can be. Here we're talking about mapping and LBS from the ground up, with
> just a touch of help from the 'experts'. The question is what size
> evolutionary steps can be bridged by individuals, to push the entire
> ecosystem of developers and publishers to some criticality of proper
> implementation (in RDF, OGC standards, or whatever)? Is it even a good
> strategy? 
>  
> My instinct says so, but I could probably be convinced otherwise briefly.
> GeoURL was pretty successful and pretty simple (but it's still surprising
> all the ways that could be botched up). Geotagging in flickr (yea guily,
> those syntactically depraved geo:lat=* geo:long=* tags) has exploded due to
> the simplicity and help from come clever greasemonkey/googlemaps
> integration. Nearly 20k geotagged flickr photos, and that's a good thing.
>  
> -Mikel
> 
> _______________________________________________
> Geowanking mailing list
> Geowanking at lists.burri.to
> http://lists.burri.to/mailman/listinfo/geowanking
> 
> 
>