Re: Group merge?

(remaining off-topic, because it's interesting :) )


On Sunday, November 27, 2011 at 10:47 PM, Manu Sporny wrote:

> WARNING: mildly off-topic.
>  
> On 11/20/2011 11:22 AM, Marcos Caceres wrote:So, it's not simply a matter of "democracy" as both are going
> > > through the standards process. The problem is that when a large
> > > company backs one of the standards with a "study" and then doesn't
> > > release the data or methodology.
> >  
> >  
> >  
> > Google was under no obligation to release anything, so I would be
> > thankful that they did release what little they did.
>  
>  
>  
> I disagree - if you are going to do a study and say that you have data
> to back up your position - you should be prepared to publish that data.


Lets view this a little differently. Imagine Google was a journalist and it could not reveal its source (most would agree protecting sources is a good thing, right?). However, Google could still serve as a valuable whistle blower putting forward allegations that an issue has been uncovered. The fact that they blew the whistle has already had significant impact: causing you to actually go and verify the facts (and it's a great motivator behind establishing the data driven standards group) . So, it's galvanized and mobilized you and others to apply a scientific method* to verify the claims put forward by Google.  
  
* we are yet to establish if your own approach is "scientific", particularly any data analysis and the statistic modeling behind it.
> Anything else is not science and when we go away from data and science,
> we delve into politics.

Science is politics. It just has a method. You unfairly seem to cast science is something clean and pure, and politics as something dirty and impure. I don't want to sound like a Google apologist, but I don't think Google ever claim to be applying the scientific method (even if they used word from the scientific vernacular, like "study"). It's kinda like if you open a carton of milk, smell it, and it don't smell right: you know it's not right, but you don't need to go to a lab to prove it. You still warn others "hey, I think the milk might have gone off. Have a smell."  

Also, it's immediately suspect to take a purely scientific approach, particularly from a business entity. Your own website claims to have a stake and vested interest in Semantic Web technologies - which may be in direct competition to other technologies and businesses. It's not disingenuous to want to make money from this, nor does it need to be stated. My point is, anything can be cast in an "impure" light even if the motivations are pure.     

> So, while nobody is under any obligation to
> release anything, it's bad form to say that you have a study that proves
> something and then refuse to share the methodology and the data behind
> the study.

Again, it's not fair to put this argument "in your own terms". Ontologically, it's important to see the world from multiple perspectives, where truths are established not just through the scientific method. As George Costanza famously said, "It's not a lie… if you believe it". http://www.youtube.com/watch?v=vn_PSJsl0LQ  
> To put it another way, you wouldn't accept the argument from a small
> team doing research... so why should you accept the argument from Google?
>  
> Trust, but verify.
You are rhetorically projecting your own bias onto me and other ("you wouldn't accept").     
> > Also, Google (and any other company) is free to back any technology
> > they want. Wether its a W3C specification is irrelevant (IPR of RDF
> > WG VS HTMLWG aside): there is no law that any company needs to use a
> > W3C spec. It's the reason the W3C is a "consortium" and not a real
> > standards body (like ISO, which does have legal implications).
>  
>  
> I don't think I was expressing anything to the contrary?
Right, I was just drumming my own drum.   
> > > In that case, you have a large company (with skin in the game)
> > > saying "trust us, we have the data"... and a loose community (with
> > > skin in the game) saying "we'd like to see the data, because it
> > > goes counter to the data that we have".
> > >  
> > > Ultimately, this is about making the right decision in a
> > > scientific way.
> >  
> > Science won't win here, I'm afraid… only the market and rough
> > consensus/running code. One decision can't be seen as more valid
> > than another, there are just technological choices.
>  
> I don't understand why Science is cordoned off into it's own area and
> the market / rough consensus / running code is cordoned off into another
> area. I see all of those as inputs into the process.

Because it has little to no bearing on the decisions making process of a standards setting body (and the adoption of its standards). Getting standards adopted is more than proving that one standard is more scientifically valid than another: this is not a scientific journal.  

Standards are about human factors, social capital (relationships), history, and economics - science has very little to do with it apart from playing some minor rhetorical goal ("we showed that 23% bla bla bla, and only 4% of bar foo bar"). Scientific method may inform the process, but in the end people will make a decision in the best interest of the benefactors they serve (e.g., end-users, or their companies)… and sometimes they will deliberately poison the well (e.g., by using "ignore" tactics or outright hostile/subversive tactics, like the formation of the WHATWG).     

> > I'm wondering, what will you do if you find Google was right (i.e.,
> > 70% people are screwing up RDFa)? And even if you find the opposite,
> > I don't think it will prove much. Remember this is not the first
> > "study" Google has done [1] about this (see my comments at bottom of
> > that page, and my own outrage about that particular "study").
>  
> We are proceeding with the understanding that Google is right by
> publishing RDFa Lite (the "Trust" part). We are doing our own research
> to ensure that they are correct (the ", but verify" part).
>  
> If we find that Google was wrong, we will have to see if the changes
> that they're proposing would do more harm than good.


I'd be interested to hear how that would be evaluated.  
  
> > What might be more interesting (and scientific) is to see how many
> > people screw up RDFa vs Microdata vs Microformats (or other control
> > group). That way, you have a scientific basis on which to draw some
> > valid conclusions from. E.g.,
>  
>  
>  
> That's what we're proposing to do... so we're on the same page here. :)
Coolio.   
>  
> > And, of course, "screw up" would need to be defined (i.e., invalid
> > format or whatever). Anyway, proving Google wrong won't really
> > settle the matter. It might just come down to implementers personal
> > choice and perception about the technology:
>  
> This isn't about proving /Google/ right or wrong... it's about
> collecting data so that all of us can make more educated decisions when
> it comes to authoring standards.

Perfect.

Received on Monday, 28 November 2011 08:17:48 UTC