It’s Dec 31, and time to see how my past predictions have done.

I predicted that XMethods would list less than 400 services by today, and lo-and-behold, that’s the case; they list 366.

Am I psychic? Have I hacked XMethods? Nope, I just performed simple linear extrapolation. Why linear? Because Web services have extraordinarily poor network effects; their growth has been roughly a function of time, not a function of the size of the existing population as you’d expect to see in an Internet scale network (like email, instant messaging, or the Web) which benefits from Metcalfe’s law.

I also predicted that the TAG would reject the Web Services Architecture document. Of course, that document hasn’t progressed very far, and will apparently only be published as a Note, meaning it will never be reviewed by the TAG.

So here’s a challenge to Web services promoters; make a prediction about the number of Web services available on the Internet by the end of 2004 and/or 2005. If they’re so great, then surely, at some point, there’s going to be thousands of them, right? When will that be? At this rate, they won’t get to 1000 until 2010 … assuming the hype – which is the only thing keeping them even linear, IMO - lasts that long.

… to the land of blogdom.

telnet proxy.markbaker.ca 80
MONITOR http://www.pacificspirit.com/blog/index.rdf HTTP/1.1
Host: www.pacificspirit.com
Reply-To: http://www.markbaker.ca/private/
Content-Length: 0

aka “subscribed”

It may only be an automata, but I’m gettin’ good lovin’ from The Classic IQ Test;

Mark, you are a Visionary Philosopher. This means you are highly intelligent and have a powerful mix of skills and insight that can be applied in a variety of different ways.

Update; they ask for personal info on page three, but you don’t need to enter accurate information (i.e. email address) to get your results

The Web Services Description WG offered up their latest attempt at an HTTP binding for WSDL last week. I think it’s about as good as you can hope something like that to get, without rejecting Web services. But it seems the group asked itself some good questions while reviewing it, in particular this one;

Is it good practice to extract part of your content to parameterize your URI? If not, what is the best way?

To which I responded;

Invoke GET on the URI to retrieve a document, and include in that a declaration of how the URI can be parameterized. This is what forms are for (well, “GET forms” anyhow); see XForms or RDF Forms.

If you didn’t recognize it, this is a textbook case of bootstrapping; see a URI, GET it, etc.. It reminds me of the Java Beans component model, which, for all its flaws (namely, that it wasn’t useful for much more than GUI stuff), had its own wonderful bootstrap mechanism; the zero argument constructor (sorry, no good URI for it). This permitted anybody who was in possession of a bean class to instantiate it without a priori information about it. Some might argue that this was just deferring binding, and they’d be right, but there’s no “just” about it; deferred (aka late) binding permits more functionality to be internalized into the system. In the case of beans, this was realized via the use of a wizard which could be triggered when instantiating beans that needed parameterizing.

Bootstrapping is your friend.

John Beatty reminded me that a friend told me a great business (aka non-plumbing) description of the differences between transport and transfer in the supply chain management space, where they use those very same terms. Paraphrasing, he said;

Transport refers to the movement of goods between locations. Transfer refers to the change of ownership of those goods.

Compare that with this REST vs. SOA diagram.

Compare and contrast.

Werner Vogels writes;

A common trend still is that there are no generally accepted definitions of (web) services, and people spent way too much time criticising each others definitions.

Dan Connolly wrote, this in response to a question regarding the value of dereferenceable URIs;

my answer to “but what if I don’t want to publish representations or descriptions of the thing” is: then you’re not doing what you can to establish consensus around your terms.

My point being that perhaps if there were more Web services to which people could look for definitions, conferences like the one Werner’s attending (not to mention the Web Services Architecture WG and indeed most other Web services standardization groups) could be more productive.

Nothing facilitates concensus quite like running code.

Don’t chicken-and-the-egg problems suck?