By Jack Vaughan
I would like to talk today about control theory, the economy, BPM, event processing and some other strangeness.
While reading a recent science journal I chanced upon a story on control theory and its limitations when applied to the world economy. In “Everything is Under Control” by Brian Hayes in American Scientist [May-Jun, 2009], the author discusses varieties of feedback control and then analyses how these might apply over some or the other snapshot of time to related feedback algorithms in the system that is the world economy.
This often controversial process – that is, converting the economic system into a divinable process – has never quite worked, and is under renewed scrutiny given large-scale economic turmoil at the moment. Put more simply: the world economy is in an uproar and no one seems quite to know why exactly, and few trust computer programming to figure a way out.
Anyway, what especially caught my attention in Hayes’ piece was a depiction of MONIAC, for Monetary National Income Analogue Computer, an early analog computer that was based on – get ready if you haven’t heard this one – hydraulics! Yes, water, rather than electrons was the goop that moved through this system. The MONIAC system has an unfortunate resemblance to an early toilet. Still, it was a valid stab at a solution for its time.
Which brings us to BPM. At heart it is a cousin to control theory. ‘Let’s figure out the process and put a master control upon it,’ the BPM practitioner, like the MONIAC developer, might say. As patently obvious as it may seem right now, BPM is full of streams and eddies that are yet to be discerned. There are many questions. How do rules engines interplay with BPM? How will front-end business modeling ever truly connect with ‘up-from-the-stack’ business process execution? Is BPM related to event processing?
Hard to easily ascertain questions – no doubt. But as we recently asked the last of these questions to Neil Ward-Dutton (of Macehiter Ward-Dutton Advisors fame) we share his thoughts here on event processing and BPM – and how tenuous their relation is.
Event processing might sound like something that you might use BPM technology for, but in practice it’s implemented quite differently because the runtime conditions in play (and hence the design philosophies) are very different.
BPM technology typically focuses on highly structured flows of work involving the coordination of multiple systems and/or people. Although there might be high throughput at runtime (in straight-through processing in financial trading scenarios, for example) those flows can be mapped out in advance and centrally orchestrated at runtime.
Event processing technology, in contrast, is optimized to detect, filter and analyse and then act on events that might occur in unpredictable ways, in unpredictable sequences, and at very high volumes in highly distributed environments.
Neil had more to say but here we’d forward you to Macehiter Ward-Dutton Advisors Web site for more analyses on issues of this era. Right on Neil! As well, you might check out the blog of Brian Hayes. My two cents worth: Brian Hayes is something of a national treasure. He continually finds worthy near-philosophical topics in the often dry dessert of computing and math. His work can be found in American Scientist, and on his blog http://bit-player.org/.