Re: Video Game Series (was VideoGame Proposal)

No-one would be happier than I (well, I and about 1,000 of my
colleagues) if Google et al. started respecting JSON-LD wholesale, and
while I can't predict this either I think said experimentation suggests
that will eventually happen.

Which is why the solution suggested by Dan Scott/"parameterized type
mechanism" has a lot of appeal.  Results-focused webmasters who wouldn't
normally use hidden markup or JSON-LD because of the negative ROI on the
work would, I think, be prepared to make such sorts of qualified
declarations.  That is, in the example Google's still provided with
information about "a series" which has a lot of potential value, even if it
chooses to ignore "oh, and I'm talking about a video game series"
qualification.

A little analogous to the use of the sameAs property, which is rarely
employed in such a way that the target URL is visible, but nonetheless is
still probably useful from a search perspective:  if Google chooses to
respect the <link>- or <meta>-declared property to get a better
understanding of the resource content, great; if not, the fact that Google
might ignore the sameAs value isn't a constraint on declaring other
properties about the resource that's visible to humans.


On Thu, Oct 16, 2014 at 1:28 PM, Dan Brickley <danbri@google.com> wrote:

> On 16 October 2014 20:53, Aaron Bradley <aaranged@gmail.com> wrote:
>
> > <meta itemprop="seriesType" content="VideoGame">
> >
> > ... then I think it's potentially problematic.  Anything requiring
> > non-visible markup is should be a red flag in schema.org, since the
> sponsors
> > are so endlessly keen to limit this to situations where the declaration
> > requires a particular data format for machine consumption, a la dates in
> > ISO-8601 format.  What you suggest with the "seriesType" itemprop
> certainly
> > works fabulously, but raises that red flag.
>
> I quite like seriesType, although simply finding a Series and seeing
> that it consists of several parts that are all of type VideoGame might
> be adequate for many applications, without stating an explicit
> Series-level type. There are probably a few mixed-media / format
> crossover cases where a series might mix different parts, even if a
> Series is most typically composed of the same kinds of thing.
>
> We have occasionally at schema.org discussed exploring a kind of
> parameterized type mechanism, possibly even with some syntax
> conventions. e.g. https://schema.org/SportsActivityLocation plus
> (somehow...) Sport=Bowling, Sport=Golf instead of a custom named
> BowlingAlley (or GolfCourse, TennisComplex etc.). This is heading in a
> similar direction.
>
> Regarding the sponsor's "endless" preference to avoid non-visible
> markup, ... there are a few statements in the schema.org FAQ
> discouraging too much hidden markup. But the world evolves, and the
> Web of today isn't quite as it was when
> http://www.well.com/~doctorow/metacrap.htm was written. I'm sure
> you've noticed a few Google products have been experimenting with
> using JSON-LD. Whether any broader schema.org and industry consensus
> around "non-visible" structured will emerge, I can't predict. Search
> engines can serve their users better when they have lots of high
> quality structured data - and sometimes different aspects of 'quality'
> pull in different directions. Modeling vs publisher/developer
> usability vs mischief disincentives like visibility. At this stage I
> think the best thing for the VideoGame design discussions is to focus
> on getting a reasonably simple but informative model of games. The
> seriesType: VideoGame construction is very similar to modeling
> patterns used elsewhere across schema.org.
>
> cheers,
>
> Dan
>