August 2002

"How will we justify VOIP (Voice Over Internet Protocol)? The answer is.... Users want good interface design." -- David Willis, Network Computing magazine, July 8, 2002.

In the 1990's we watched a large number of companies offer slick "solutions" that had little or nothing to offer in terms of real productivity or performance improvements. The products they offered were merely window-dressing or just hype-driven fluff. Venture capitalists certainly found that their own "core competency" was giving money to glib spinmeisters, not finding real products with real value. A lot of companies changed their entire information infrastructure and got zero or even negative productivity growth to show for it.

Now we have the latest pseudo-craze in telecom-computer convergence, VOIP. The idea of putting all analog or even digitized voice information onto an Intranet may sound hip and leading-edge, but one wonders what the justification could be. On what basis would a company take one of the few remaining bastions of reliability in the world of small computers, the PBX or Private Branch Exchange, and replace it with a IP-based system with unpredictable and inconsistent levels of service? Is there a reliable ROI (Return On Investment) model to prove that this is a good idea?

Remember that the IP (Internet Protocol), being based on little packets of information that can take independent routes to their intended destination, has a statistical basis instead of a deterministic basis. In other words, you are at the mercy of the "odds" that your packets arrive in the correct order. You will have to delay reassembling the message until the packets arrive and are de-scrambled and returned to their correct original order that existed at the sending point. Your "QOS" or Quality Of Service depends on how many resources you are willing to devote to "promise" a certain maximum delay within a certain percentage probability. On the other hand, an ordinary digital PBX or analog phone circuit delivers near-instantaneous messages with essentially 100% probability for all calls, not just those given special allocations of "extra" resources (bandwidth, processing power, priority in the queue, etc.) to "promise" (the word commonly used is "guarantee") higher QOS (less probability of high delay).

So the so-called efficiency is produced by fudging on resources and only giving "enough" resources to truly "important" messages. The fun and games come from deciding which messages are "important enough" to be given "enough" resources. And how much will you pay to make more likely (since with IP you will never be certain) that you will receive the truly important message in a truly timely manner? And never mind the lower reliability of a typical Windoze-based VOIP system.

So where did the know-it-alls go wrong? When they decided to use a protocol from a non-real-time system (the Internet) for a real-time (telecom) application. Sure, there are certain parts of a PBX system (such as voice mail) that are not time-sensitive. But voice messages that are subject to digitized multipath distortion, variable message delay, and increased downtime (typical of a cut-rate Windoze system) are penny-wise and pound-foolish. The know-it-alls are selling reduced reliability and reduced performance on the premise of reduced costs....

Which is another bogus line of reasoning. The main cost of an information infrastructure is the infrastructure fixed cost and maintenance, not the transactional overhead. Tearing out a working infrastructure to replace it with another infrastructure -- even for FREE -- is no wiser than replacing a shaving razor with a pocketknife on the theory that "you'll save a lot of money by not buying blades." If the infrastructure is wrong, then the overhead savings are meaningless. Claiming that "the infrastructure is the Web" is not telling the whole truth. That's like saying the infrastructure of a PBX is the phone company.

So how do our friendly neighborhood yarn-spinners now plan to sell this bogus scheme to uninformed and misinformed managers everywhere? As the opening quote said, by focusing on user interface improvements. You remember the last big fiasco involving improved user interfaces, right? The one about being "intuitive" and "never needing training" because a "uniform GUI" would provide "user friendliness"? Yes, that was the Windoze party-line for a decade. Companies did not realize they were inserting an entirely new programming infrastructure between the hardware and the GUI, and that the GUI was only the rationalization or selling point. Windoze was never about making things easier, it was about selling more hardware. And so is VOIP.

Most recent revision: July 28, 2002
Copyright © 2002, Tom Nadeau
All Rights Reserved.