Jotting Number 84, by Joe Podolsky:
From: joe_podolsky@hp.com
Date: June 21, 1999
Before you head for the emergency exits, there's also an airplane joke that takes the other side of the issue. According to my friend Earl Weiner, an expert in cockpit designs and processes, the cockpit of the future will have two living things in it: a person and a dog. The person is there to feed the dog. The dog is there to bite the person if it dares touch anything.
These stories are "old". They were designed in the days of process control, where we were concerned that artificial intelligence would be programmed into machines, doing not only the heavy lifting that most of us happily give up, but also the heavy thinking that some of us feel is the reason for the survival of the species.
We've gone well beyond that now. Computers and networks and software have become so intertwined, so complex, that we can't always know what these devices will do, nor where, nor when. Except for the engineers who make all these miracles happen, we don't need to know, and we don't want to know. We want the complexity hidden from us so that we can get on with using these wonders for our jobs. We don't need nor want to be aeronautical engineers or even pilots in order to gain the benefits of air travel.
Sure, there are enough horror stories around, just as there are occasional airplane disasters. For example, all the concern and expense about the Y2K "bug" indicates that computer systems have been far from perfect for a long time. The viruses and worms that infect our computers through the Internet are clear indicators that evil people can use this technology to do nasty things that good people can't easily stop. Playful or malicious hacking into our most secure government systems tells us that even our top computer science experts aren't completely in control of their own environments much less being able to help us with ours.
Yet, in spite all of these dangers, which of us would give up our word processors, our e-mail, our access to the Internet? We like ordering stuff on the Web, and most of us understand that putting our credit card number on a secure Web site is far safer than handing our credit card to a waiter in a restaurant.
But no matter how good things are, I do believe that we have a responsibility to keep on getting better. And improving systems safety and security is on top of the "things to improve" list. This is especially true, since the Information Technology industry, led by HP, is embarking on the most automated adventure yet, e-services.
We proudly proclaim that, with e-services, instead of our having to work the Internet, the Internet works for us. E-service application modules will join together, perhaps on the fly, to get a job done for us. Sounds good. We are harnessing the power and complexities of cyberspace to make our lives even more wonderful. Planes may still get delayed by weather problems, but our interconnecting rental car and hotel plans will all be automatically rearranged. Supply chain issues will be compensated for, and cash flows will be optimized according to our previous instructions. Yes, there will be screw ups. Yes, we will need to override the system sometimes. Yes, we will need access to experts who can help us with all these things, especially when these e-services are young. But, the benefits far outweigh the risks. And, again especially in these early days, we can learn and grow from our mistakes.
In fact, e-services are answers to the issues raised by the last generation of IT applications. In the May 10, 1999 issue of Fortune, venture capitalist Bill Gurley talks about "The New Market for 'Rentalware'". Gurley points out "that client-server software, which resides inside a corporate network, has become too expensive, too hard to implement, and too complex." E-services to the rescue. Application Service Providers (ASPs) are being formed to offer so-called "apps on tap" so that a corporate user can rent instead of buying the services they need. Some of these services are also being offered by the application vendors themselves. For example, SAP has formed a partnership with HP to offer SAP e-services.
Gurley points out that one of the most persuasive benefits of this rentalware trend is that the ASP can create linkages that connect users to other related services that they may need. For example, Gurley mentions that "InfoWave Technologies in Atlanta has created a unique Web-based service that links customers to the consultants who best match their needs. This kind of immediate connection is lean and fast compared with a cumbersome client-server solution." It's very much like Geoff Moore's concept of a "whole product," where a vendor solves the customers "whole" need through a web of alliances.
Peter Fabris, in his article, "A New Lease" in CIO Web Business, May 1, 1999, talks about some of the advantages of renting applications. Application renters can avoid upfront capital expenditures. Rented applications can be ready to use sooner than if implemented in-house. Renting can help solve staffing problems and maintenance costs. Renting allows the users to focus on the company's business, not on solving IT problems. "All we have to do is learn about using the application," says one of the users. "We don't need to know how to maintain the servers."
Fabris doesn't deny that there are control issues. One big problem is customization. To get the advantages of e-services, we have to follow the processes dictated by the software designers. Upgrades and modifications are done on the vendor's schedule, not necessarily according to our needs. And there are the ever-present concerns about security and privacy, although I think that the emerging authorization and authentication models becoming widely used are beginning to ease these fears.
What's interesting about the e-services world, however, is that we e-services buyers will have more choices. We won't be locked into a software solution by the huge up front installation costs. So, if we're not happy with the processes inherent in one e-service, we can switch to another at relatively modest effort.
Note, then, what happens. The point of control changes. We e-service consumers lose control of the detail intricacies of the underlying processes, but we have regained the power that all consumers have in the competitive marketplace. This is a big deal to us IT people who have made our careers working under the hood of customized systems. But it's a blessing to us businesspeople who have been at the mercy of unpredictable IT project schedules and bewildering budgets.
Within in this market-driven world, however, some people hold strong philosophical reservations. Ben Schneiderman, head of the Human-Computer Interaction lab at the University of Maryland, was profiled in the March 1999 issue of Scientific American. Schneiderman thinks that "enhancing computers' autonomy raises troubling questions about who will be responsible if machines controlling air traffic or medical equipment, for example, make disastrous errors."
In that same article, however, Patti Maes, an MIT professor and expert on computer agents, responds with a view of the benefits. Maes is concerned about responsibility issues, but she also feels that "The goal of agent research…is not to mimic human intelligence but to help the user suffering from information overload by providing simple, understandable, predictable programs that can act on his or her behalf."
Those are good justifications for automated e-service actions. Let's take this discussion one more step. Until now, we've assumed that computers were going to do "merely" what we told them to do. What about computers that we design to be "social beings"? Schneiderman hates the concept of computers acting like people, but that's what happens anyway. The May 1999 issue of Communications of the ACM focuses on "Persuasive Technologies." It was edited by B.J. Fogg of Stanford University, and raises many opportunities and concerns inherent in the field of "captology," which is the study of the interaction between computers and persuasion. There are lots of examples of these, many of which are e-services in the most general definition of that term. For example, there's a computerized doll for teens that is designed to simulate the difficulties of parenthood. There are programs to help people with more nutritious diets. There are monitoring systems to encourage handwashing after bathroom use by restaurant employees. All of these can be connected, via the Internet, to capability upgrades or to databases that constantly improve the service being offered.
These systems are technological marvels. Using them appropriately can dramatically improve our lives. But that doesn't mean that we need to trust them blindly. What safeguards would you insist on before using these e-services yourself? How, for example, would you assure appropriate data quality when looking at an e-service that quotes statistics about the health benefits of a particular diet? One way we do it now, is by having trusted watchdog private and public agencies. The world of watchdog e-services is already beginning to grow.
Someone once said that we design systems, and then they are our masters. That means that e-services become matters of public policy. What should we do to assure that the right people have the right debates on these issues? Who are the right people? This is the natural playground of politicians, academics, and journalists. But what is the appropriate place for us technologists in these discussions?
Our role in IT is evolving to be both the pilot and the dog, to provide the right amount of automation, and the right amount of human oversight. And to keep things from getting out of balance, from getting out of control... control... control...
Regards,
Joe