[rdfweb-dev] responding to a new wave of anti-RDF sentiment

Graham Klyne GK at ninebynine.org
Tue Jul 19 09:52:07 UTC 2005


Hi Jo,

Strangely, I take some comfort from all this.  It has been my perception 
that major new technologies seem to go through these phases; roughly:
  (a) conception
  (b) take-up by a small technical community
  (c) hype and over-hype
  (d) backlash
  (e) real use; understanding the true strengths and limitations

By my own unscientific view of where RDF and the Semantic Web are in their 
evolution, which is very little changed over the past 3-4 years, we are 
about ripe for the backlash phase.  (An earlier sign of this was Clay 
Shirkey's (sp?) piece about the weakness of syllogisms.)

Anyway, FWIW, I think that a key component that will help lead to (e) will 
be SPARQL.  And I think SPARQL can help combat some of the concerns about 
complexity, and help us to focus on the basic (simple) model.  (Ironically, 
in view of the criticisms, one thing that attracted and attracts me to RDF 
is that it's fundamentally simpler than XML;  much of the complexity is 
fluff and detail that can be cast aside.)

By way of example, I'm currently working with a team on a semantic-web 
based database for publishing biological research data.  Lately, I've been 
floating the idea that much of the software could (not should) be replaced 
by publishing a bunch of RDF files on a web server and using an external 
SPARQL server (both Asemantics and Dave Beckett have done this) to query 
the data.  All the rest is technical detail for optimizing and scaling the 
system (hard, but not fundamental).

I also think that SPARQL provides the tools we can use to integrate 
constructively with the XML world.  There's a lot of bioinformatics data 
published as XML, and I think SPARQL gives us a way to query and combine 
these various sources.

#g
--

At 15:56 16/07/05 -0700, Jo Walsh wrote:

>i'm forwarding the message below, as an example of a tendency i've
>seen on several domain-orthogonal mailing lists recently, with a
>common anti-RDF set of messages which run roughly like this:
>
>"RDF is too hard"
>"It is too complicated to parse, XML is much easier"
>"There are no application toolkits or simple frameworks accessible in
>a low-footprint install to php-type developers"
>"I don't need RDF for my application to work, why should anyone else?"
>"Microformats / OPML / $monthly_flavour is what the other cool kids
>are doing and will provide an emergent standard"
>
>I find it hard not to come over defensive when others sound
>dismissive, but i can't resist making at least one response to this
>row when it comes up in different domains. Meanwhile, the people
>working on ministandards for small domains, while doing some good
>thinking about modelling and vocabularies, are definitely pushing
>their work as an RDF alternative, making the above claims, and gaining
>a lot of blogosphere leverage by virtue of their position on the
>power-law curve of public notice.
>
>When FOAF broke over the social net apps, dan and others wrote some nice,
>positive-view stuff on "why the big picture matters". My sense is that
>we could really do with a visually appealing, simple,
>positive-refutation document (not slagging off the xml formats, just
>setting out the rdfweb-dev stall).
>
>have been making cartoons to use as talk slides and to explain some of
>the proof-of-concept data aggregation / syndication stuff we've been
>doing with wirelesslondon. They are knockoffs of the CC cartoons -
>saul christened them 'creative cum-ons'. They go down pretty well.
>http://map.wirelesslondon.info/docs/ http://smal.omweb.org/
>
>I'm very much into making a 'why-RDF' series, perhaps more in the
>style of the beautiful isometric-view figures that ephidrina did for
>the SWAD-E t-shirt, which i get asked about a lot when i wear it.
>I moot this here because i'd like help with a script, that makes clear
>network-effect data merging, that touches on now-trendy geo and
>calendaring and connects to the hard work done there in RDF, that will
>make programmers think twice and non-programmers look harder.
>
>( Incidentally i am longing for a toolkit that i could use to
>dynamically generate isometric or perspective views from RL geodata
>and decorate with RDF-derived bits of cartoon imagery. Is there
>anything near that out in the world - in the mmorpg domain? )
>
>http://lists.advocacydev.org/lists/arc/dev/2005-07/msg00017.html are
>my current set of counter-arguments which i would tone down for a more
>public, less friendly list than the above, for sure, in response to:
>
>----- Forwarded message -----
>
>To: "Joshua Tauberer / GovTrack.us" <tauberer at govtrack.us>
>Subject: Re: [dev] Open Databases & the Semantic Web
>Cc: Jo Walsh <jo at frot.org>, dev at lists.advocacydev.org
>
>I agree, the Semantic Web is quite compelling in theory. However for a
>system such as you are proposing to work, you will either need to
>build some nice, easy to use tools in order to enter data, or use a
>format other than RDF. I agree that the set of semantic web
>technologies is really nice, they are also hard to use, and simply too
>complicated. Technology can't be the stumbling block or getting open
>legislative information won't get done.
>
>Succesful technologies employee David Weinberger's idea of 'small
>pieces, loosely joined' HTML and RSS have been successful because they
>were easy to learn, easy to parse, and simple to understand. RDF does
>not meet that test.
>
>Last thursday night I attended David Winer's road show at the Berkman
>Center where he evangelized OPML (and his new OPML editor) as another
>simple XML format that is going to be everywhere soon.  OPML is a
>simple XML format for creating trees of hierarchical information. At
>its core, legislative data is hierarchical information, and could be
>stored as OPML. What makes OPML magic is that you can include pointers
>between hierarchies. So you could delegate responsibility for the
>Massachusetts state legislature, to a volunteer who knows about MA,
>and he/she in turn could delegate responsibility about a particular
>legislator. This allows small pieces of the nationwide database to be
>loosely joined without a centralized system keeping track of things.
>
>I don't think David understands his own technologies potential (he
>tried to show it off as a blog publishing tool) - but I think that it
>will be used for more than outlines. The podcasting community is using
>OPML to maintain their distributed show directory, and it is just
>starting to catch on else where. Someday, OPML might be another sort
>of universal glue, in the same way that RSS is today.
>
>RDF is the right solution from a purely technological perspective, but
>it still might not be the right real world solution. There are serious
>downsides to either approach.
>
>----- End forwarded message -----
>
>
>
>_______________________________________________
>rdfweb-dev mailing list
>rdfweb-dev at vapours.rdfweb.org
>wiki: http://rdfweb.org/topic/FoafProject
>http://rdfweb.org/mailman/listinfo/rdfweb-dev

------------
Graham Klyne
For email:
http://www.ninebynine.org/#Contact




More information about the foaf-dev mailing list