EAI, web services and high school grammar classes

My one and only new year's resolution this year is to take the time to
write down all of the silly questions I have in my head that I would
like to have answered. The kind of questions we knowledge workers do not
ask willingly because of fear that we will look silly or ill-informed.
Here are a couple from my ever-expanding list:

1) Archaeologists must dig down into the earth to discover stone age
artifacts. Where did all of the new earth come from to bury everything?

2) Is there a scientific explanation for the fact that the moon is just
the right size to totally eclipse the sun, and that it is at just the
right rotational speed to ensure we only see one side of it from Earth?

3) Why are there two tides a day? (I have read a few explanations for
that one but still don't feel that I understand it properly.)

I remember studying English grammar in my youth and finding my head
exploding with "why" questions I was too diffident to ask. Here are some
I remember:

1) Are irregular verbs unavoidable? Wouldn't language acquisition be so
much easier without them? Why do they exist?

2) Why are there so many "special cases" that break the rules?

3) Why are there so many more nouns than verbs?

Let's skip the question about the avoidability of irregular verbs and
the plenitude of special cases. (I still have not come across a good
explanation of these phenomena!). Instead, let's concentrate on the
question of the disparity between the number of verbs and the number of
nouns.

It sounds plausible that in the classic subject/verb/object scheme of
things common to almost all languages, there are many more objects in
the world than there are actions you would want to perform on objects.
By combining a small number of verbs with a large number of nouns
(objects), you create a very large communication space with the minimum
of fuss. As long as you know the general rule for applying verbs to
nouns you can do so without any extra language machinery beyond the
general rule. I'll buy that.

Believe it or not, this brings us nicely onto the subject of enterprise
application integration (EAI) in general and web services in particular.

The web can usefully be thought of as a large collection of nouns. These
nouns are known as URLs. Basically, things that have addresses, created
using a unified, universal naming convention.

The web also has a small set of verbs. Some of the most common are GET,
PUT, POST and DELETE. The HTTP protocol is basically the means by which
we apply this small set of verbs to a vast sea of nouns on the web.

Recently however, we have seen the emergence of the web equivalent of
irregular verbs. Hoards of them in fact. Each of which must be learned
on a case-by-case basis. There is no rhyme or reason to how these verbs
work. You just have to learn them just like the irregular verbs of your
grammar classes.

Examples include the verbs "To search using search engine X", "To create
a PDF", "To list Italian fiscal codes". . . A sample list of these so
called web services are available on the Xmethods[1] website.

Web services technology - in the sense of SOAP/WSDL - is basically a
recipe for creating an uncountable number of irregular verbs - one for
each noun on the Web!

But look back at what we started with on the web. No irregularity. Just
a vast communication space created out of sheer simplicity. A small
number of standard verbs applied uniformly to a large number of nouns.

So here is the big question. We have learned from the real pain of
enterprise application integration that irregularity really hurts. Every
CORBA IDL was basically an irregular verb. Every EDI bilateral exchange
format was an irregular verb. The combinatorial explosion of effort
involved in linking one irregular verb to another is unacceptable.
Irregular verbs hurt.

And yet we may have living proof that it does not have to be so. Perhaps
irregularity - at least in IT languages is not unavoidable. We have the
web.

If things continue in their current trajectory in the web services
world, grammar classes for programmers in the future are not going to be
a pleasant experience. Those that find the experience of learning this
stuff "off by heart" grindingly time consuming and boring will ask "does
it really need to be this way?". "Is all this money really being well
spent learning all this stuff?" "Why is it that EAI still costs ten
times what the hardware/software cost - just like in the CORBA and EDI
days?" "Why can't things be as simple as the original web?".

The good news is that there is a growing appreciation of *why* the Web
works, thanks in no small part to Roy Fielding's dissertation on the
REST architectural style that underlies the runaway integration success
story that is the web. At the heart of REST is the idea that the web
works precisely because it uses a small number of verbs applied to a
large number of nouns.

Examples of how this idea can be applied to web services are emerging -
building on the lessons of the web rather than repeating the mistakes of
EAI eras past.

As a starting point, I recommend Roger Costello's REST tutorial[2]. From
there I recommend Mark Baker[3], Paul Prescod[4] and Bill de
h