In part 1 of this interview with Miko Matsumura, vice president of SOA marketing for webMethods Inc., he talks...
about the role he sees SOA education playing in bridging the gap between IT professionals who may have embraced the service-oriented approach, and business people who may not have heard of it as yet. In part 2 on Monday, Matsumura discusses who will lead the way in SOA education in 2007 and why it will determine which organizations will be winners and losers in the race to implement SOA.
In your world travels in support of SOA technology, what are you seeing in terms of trends that will be important to watch in 2007?
I think we're increasingly seeing the dawning realization that there are two facets to the SOA equation. There's the machine facet and the human facet. We're increasingly seeing realization about what I call the "cybernetic nature of SOA." This is essentially the interfaces between the human side and the machine side. There are several different things that fall out from that. What sorts of things?
First of all you have to look at why this is happening and what is the trend line. The way I like to think about why the human factor is increasingly becoming part of the discussion is put simply if you look at the dividing line between humans and machines, what you're going to see is that machines actually have these very fundamental transformation vectors. For example, Moore's Law, which essentially says that the power in a processor is on a doubling curve every 18 months. Add to that increasing bandwidth and memory and the machine side gets more and more powerful every passing month. The thing that we don't have is an equivalent principle that drives the ability of human beings to work together better on a similar doubling interval. It would seem like carbon-based bipeds could become overwhelmed by the computing power that surrounds them?
Yes, and what's interesting about that is there are two philosophies about the interface between the human and the machine that seem to be relevant here. These two philosophies are exemplified by the Apple Newton handwriting recognition system. Back when Apple wasn't cool, they released this early PDA product called the Apple Newton. The nature of the handwriting system was essentially based on the premise that "computers are smart they'll be able to figure out what people are doing." Yes, it was supposed to learn your handwriting and translate it into ASCII, right?
Yes, and that, of course, was a dismal failure. What seemed to have caught on better was the Palm Pilot approach, which was "human beings are flexible and therefore they can adapt the way that they write text in a way that these dumb computers can start to recognize." You had to learn the graffiti code for each letter of the alphabet, correct?
Yes, you had to learn the graffiti script. But here's the take away from the Palm lesson at some scale, which is that while computers are getting faster and faster and more powerful, the notion of some kind of sweeping artificial intelligence ability that will universally resolve more complicated issues like data semantics, is basically a pipe dream. So where does that leave us?
The assumption I'm going to make is that it's easier to adapt human behavior. The thing that we human do best is adapt and that's something machines are not good at. Machines are good at accelerating speed, but human beings are good at being adaptable. And what does that mean for SOA?
Because SOA is basically a system, in order for optimal balance to be achieved there will need to be adaptation of human behavior towards an SOA world. So there is need for human adaptation, which is increasingly being recognized in the form of what people call "SOA education," which will be the buzz phrase for 2007. I predict that SOA education will be as important as SOA governance was 2006. Will that include educating the business people, who according to some surveys do not yet know what SOA even is?
Absolutely. This whole notion of SOA education extends to behavioral modification and conceptual absorption of SOA by people in business and technology functions. For the alignment of those functions both sides need to have common understanding and language. What will SOA education then be?
The interaction of the human world and the machine world produces powerful outcomes, but there is an expense. The expense is that human beings need to adapt slightly different behaviors. But on the benefit side, if you go back to the metaphor of the Palm Pilot, you can take advantage of things like the computers unerring memory, the prodigious capacity for remembering a near infinite number of phone numbers. So what I'm suggesting is the two dimensions of the machine world and the human world are such that the machine world is increasingly becoming the servant of the human world. But on the flip side, the successful organizations that are using SOA will be the ones that realize that there actually needs to be education and adaptation on the side of the humans in order to meet and take advantage of the power of the machines. So for an organization to successfully implement SOA, they need to retrain their human resources?
The organizations that are successful in taking advantage of the technology will be the successful organizations. This has been proven throughout history. Whenever you take two competing interests, you'll find that one of the interests is successfully leveraging technology and they are the ones that gain advantages and continue to prosper. So since SOA has this property, the boundary across which the property starts to emerge is the boundary whereby we are shifting the granularity from IT systems and IT services into business systems and business services. What's an example?
When you have long running asynchronous services, you end up with human beings as part of the service delivery infrastructure. This is discussed in a ZapThink paper called "The Mechanical Turk." Back at the turn of the last century, there was this chess playing robot. It was not powered by an IBM super computer, obviously, but, in fact, this chess playing robot was powered by a human being sitting in the back room like the Wizard of Oz. So they were cheating and tricking people, but if you think of a business service, for example, say I order a book on Amazon. The question becomes: "Will human beings touch my book?" Probably. "Will human beings be driving the vehicle that causes the book to move to my house?" Most definitely. I hope so. So as SOA services become business services you get a natural progression of interaction between machine based systems, services that are the IT granularity, and the business services that are more human driven. It happens because human beings are adaptable and flexible, which actually provides for a natural optimization that is pretty exciting.
What's happened in the past with business process is that there used to be this trend called "business process re-engineering." That was a big '80s and '90s trend?
Yeah, and a lot of what came out of that was a lot of expensive failures. The model was we're going to come up with an optimal process. Then we're going to overlay that process on whatever it is you are doing and everyone will now comply with the new regime. You're now going to behave in the optimal way and we'll all get rich. What ended up being the case is that while human beings are adaptable, they are not automatons. And it turns out that business is filled with ambiguities and intricacies and contractual arrangements. These are all very soft sciences for which the rigid processes are unsuited. So now people talk about agility being an actual thing. The way agility becomes manifest is when there is harmony between the models and patterns and the reality that it supposedly augments. What's happening in SOA land is that we're finally getting to the point where there's realistic convergence between business process modeling and actual system behavior. What's happened in this iteration, there's this change time metaphor in SOA that wasn't part of the pervious generation of business process. Business process in the past was "thou shalt X," but the business process of SOA is much more along the lines of here is a configuration that produces such outcomes and within that configuration there are n + 1 variants that can be dynamically spawned based on business requirements. And ultimately you end up with a system that can flexibly marshal human and machine resources to address diverse customer requirements.