[uf-discuss] Comments from IBM/Lotus rep about Microformats
scott at randomchaos.com
Wed Dec 6 07:24:42 PST 2006
On Dec 6, 2006, at 7:45 AM, Bruce D'Arcus wrote:
> On 12/5/06, Scott Reynen <scott at randomchaos.com> wrote:
>> In HTML or JSON, new formats need new parsers, which must be written
>> by someone.
> Exactly. The point is if you have a generic model you have a
> generic parser.
Right. HTML doesn't have a generic semantic model. JSON doesn't
either, nor does XML. These are all data models. But all can be
used to represent a generic semantic model, such as RDF. If there
were a generic semantic model established with JSON syntax (RDF/
JSON?), we could convert microformats to it just as easily as we can
convert microformats to RDF/XML, but I don't know of any such model,
and microformats themselves certainly aren't that model.
>> Elias is coming from an RDF background, and microformats
>> simply aren't RDF, and they never will be. And that's a good thing.
>> If what you want is RDF, just use RDF.
> The issue isn't really microformats vs. RDF (except as RDF provides a
> model), but microformats vs. RDFa.
I don't think the issue is "vs." at all. The two approaches solve
different problems, are interoperable, and collectively improve the
semantics of the web. It's all good. It's just not all the same.
And the differences are a good thing.
> Both microformats and RDFa are addressing the exact same use cases and
> requirements (augmenting visible content with structured data).
I don't think the use cases and requirements are the same at all, and
I hope they never are or we're just doing redundant work here.
RDFa's use cases include a generic semantic model. Microformats do
not. Microformats have a requirement of making publishing as easy as
possible to maximize adoption. RDFa does not share this
requirement. These are two different efforts that will lose
usefulness if merged into one.
> RDFa includes namespacing, the lack of which is already a problem in
> microformats (witness hCite and the serious awkwardness that title
> will be indicate using fn), and which will grow over time as more and
> more people want to mark up their content.
I don't think that's a problem. I think it's just a limited goal of
solving specific problems as simply as possible. If people want to
solve general problems without the constraints of keeping it simple
for publishers, I'd say they should do that somewhere else. The RDF
community seems like an obvious choice. I hope the various attempts
at marking up the RDF model in HTML syntax work out well, but I don't
think that should become a goal of this community.
> Moreover, the need to write dedicate code for each new microformat
> will also present serious scalability problems.
So then microformats won't scale quickly. That's okay. RDF can
scale quickly while microformats are more accessible to HTML
publishers. We can build inter-op tools and everyone can be happy.
> Finally, that there's no model at the heart of microformats with clear
> extension rules means that the vaunted social process here is a mess.
> It's all centralized, and people get frustrated when their pet
> property isn't included because they know what that means: the tools
> written for the blessed microformats won't see them.
Right, so if you want a semantic model at the heart of your HTML
markup, there's one already developed in RDFa. Or you could develop
another. But microformats can not have a semantic model beyond HTML
without becoming more cumbersome to HTML publishers, and that's
something we should avoid.
From my perspective, all of these attempts to make microformats more
generalizable are sort of like telling people who are doing math on
their fingers that they should stop because that won't scale. That's
true, but they don't want it to scale right now. They just want to
solve a simple problem using familiar tools. When they get to
calculus, they'll pull out the calculator. I don't want to see
microformats turned into a calculator while there are plenty of
finger-math problems left to be solved.
> So while it might be comforting to dismiss RDFa and "it's not our
> problem", I don't think it's good strategy.
A good strategy toward what end? I think Elias has a problem that
microformats are not intended to solve. What he wants to do is have
a generic semantic model that anyone can use with any type of data,
and put it in HTML. What microformats are intended to do is provide
specific semantic models, not just /in/ HTML, but using the familiar
tools of HTML as much as possible.
More information about the microformats-discuss