Re: CfC: Request transition of HTML Microdata to Candidate Recommendation

Charles McCathie Nevile wrote:
> On Mon, 26 Nov 2012 00:57:03 +0400, Nathan <nathan@webr3.org> wrote:
> 
>> Charles McCathie Nevile wrote:
>>> On Sun, 25 Nov 2012 23:25:38 +0400, Nathan <nathan@webr3.org> wrote:
>>>
>>>> Charles McCathie Nevile wrote:
>>>>> On Sun, 25 Nov 2012 22:15:26 +0400, Marcos Caceres 
>>>>> <w3c@marcosc.com> wrote:
>>>>>>> The fact that RDFa already does what Microdata does has
>>>>>>> been elaborated upon before:
>>>>>  Yes. For what it is worth, I personally think RDFa is generally a 
>>>>> technically better solution. But as Marcos says, "so what"? Our job 
>>>>> at W3C is to make standards for the technology the market decides 
>>>>> to use.
>>>>
>>>> Standards, plural? not a standard way to do things, singular?
>>>  If the market wants more than one way, that's what we should be 
>>> offering. I think we all agree that it is *better* to have an agreed 
>>> single way. Of *course* we all agree that it is better to have the 
>>> technologically best specification as the standard.
>>
>> I'd be interested to see the data / info that says the market wants 
>> more than one offering here.
>>
>> Do you, does yandex?
> 
> We supported making it a Candidate Recommendation. That means we think 
> it is worth stabilising, to decide if it should become a Recommendation.

That makes sense :)

> We believe that schema.org presents a reasonable use case - multiple 
> very large search engines, massive numbers of pages, currently using 
> microdata.

This also makes sense, I'll note that I also used microdata on websites 
I control, but that was entirely because I personally was unconvinced by 
RDFa+XHTML 1.0 support, noted the lack of support via major search 
engines for the schema.org vocabularies in RDFa at that time, and needed 
+ wanted features from HTML 5 (rather than XHTML). Now that RDFa Lite is 
on the scene and support for RDFa is there and growing, it is my go to 
choice and migration to RDFa has begun. I control in the region of 500 
million pages around the web, pulling circa 4.5 million views per day.

Naturally I would of course presume this is the same case for many other 
metadata authors, particularly those concerned with "SEO".

> On figures I have seen from Bing, RDFa and Microdata are both "widely 
> used" (along with microformats) with RDFa more common in their index, 
> while in Russia we find more microdata.

Unsurprising given the ways things have worked out over the past few 
years, Microdata got fast support from google (still somewhat flaky!), 
it was promoted through webmaster tools and other engines fell in line 
and supported too, since that what people were publishing.

I hate to say it, but for common web usage, most of this can be (and is) 
driven by the major search engines, if they all swapped to pushing RDFa 
Lite as the first choice, Microdata usage would fade massively, likewise 
any other approach to metadata in HTML. Other use cases for metadata in 
HTML are minor in comparison.

Personally, I feel general RDFa usage figures are perhaps a little 
inflated by the inclusion of OpenGraph, however the enhanced expressive 
capabilities and open approach to utilizing machine readable 
vocabularies from around the web make RDFa a natural choice for more 
complex annotations and for ecommerce use cases.

> This suggests that right now there is a market which wants both. Note 
> that schema.org (where we participate) now supports the use of RDFa as 
> well as microdata - see 
> http://blog.schema.org/2012/06/semtech-rdfa-microdata-and-more.html
> 
> The market may well lean toward one or the other solution in the future. 

The market will lean to whatever is pushed as the first choice and 
supported better by search engines.

As noted above, RDFa will always have a foothold due to it's expressive 
capabilities in comparison to Microdata, and since it's technically 
"better".

Ultimately then, no matter how much Microdata is pushed, RDFa will 
always have a foothold.

So this is the issue I have, RDFa (+Lite) can handle simple and complex 
use cases, Microdata cannot handle them all. While this continues there 
will be two competing specifications, interoperability issues, confusion 
on the user side, and a doubling up implementation burden for those who 
which to extract metadata from HTML. Less of a burden when you have a 
town full of developers and engineers at your disposal, more of a burden 
when you're a single developer or small team trying to compete or 
innovate in the marketplace.

For this reason, I personally feel it is the responsibility of W3C and 
the major search engines to get behind one specification, and provide a 
standard to the authors and publishers on the web. I believe others, and 
the TAG, feel the same.

> In the meantime, we see no reason not to work on making microdata a 
> first-class specification. If it is still appropriate when the question 
> arises, we would support microdata becoming a W3C Recommendation but 
> right now it is too early to tell.

I understand what you are saying, but hopefully above I've explained why 
my opinion differs, and why I feel providing a standard way to do 
things, rather than multiple ways, is ultimately the best approach, 
whether that single standard is "perfect" or not.

Thus, I'd like to see Microdata published as a stable NOTE, for 
reference to the many already using it, and a concerted effort to nudge 
users towards the single standard which handles all their use cases, 
both the simple ones they have now, and the more complex ones they will 
encounter further down the road.

> Note that I am not claiming that people *cannot* use RDFa Lite - but 
> that they choose to use microdata in addition. This is what "voluntary 
> adoption", "success determined by the market" looks like in practice.

I hope you mean choose one or the other, and not both in the same 
document! That would be *very* ugly!

Your last sentence, I'm unsure if I can agree. If the major search 
engines had given a clear neutral choice between the two, or only pushed 
RDFa, we wouldn't be having this conversation. That's history, to do 
with timelines of publication, and no doubt some level of personal 
affiliation and networking. Thus, I can't agree with you that it's 
voluntary adoption, or success determined by the market, as in my own 
case, and many others I know, this isn't the case, I just used what I 
could when it was available, and as more support for "better" solutions 
has came across, I've got the luxury of migrating, many won't have this 
luxury, time or funds to migrate, or to learn a new-to-them spec (RDFa) 
when they are already familiar with an it'll do spec (Microdata).

I'll bail out of this conversation now, IMHO this is was an avoidable 
mess, but it's not been avoided and now whatever will happen will happen.

Thanks for taking the time to reply Chaals, much appreciated, and I hope 
I've explained my thoughts as clearly as you.

Nathan

Received on Monday, 26 November 2012 11:10:48 UTC