Forum

I enjoyed the special section on Aspect-Oriented Programming (Oct. 2001), but I disagree with the editors’ claim that "Object technology has difficulty localizing concerns involving global constraints and pandemic behaviors, appropriately segregating concerns, and applying domain-specific knowledge." The difficulty is not with object technology as a whole, but with the specific tools and methods that attempt to define an object’s entire behavior through its class. Object technology provides alternatives.

Asked whether AOP will replace OOP, Gregor Kiczales and others reply with an emphatic no. AOP extends current technology to allow programmers to localize some aspects of object behavior in modular constructs outside an object’s class. This is a neat trick, and, as the section argues persuasively, a great improvement over the alternative of replicating code.

I believe the same benefits could be achieved in a language such as Smalltalk by defining an AspectSpecification object analogous to the aspect specifications in AspectJ. An AspectSpecification can be implemented either via reflective programming to augment the methods of affected classes, or via runtime checks that invoke aspect behavior as needed before and after each method invocation.

Object technology can be used for AOP. The difficulty is in engineering tools that are both easy to program with and yield efficient code.

Richard Staehli
Portland, OR

I read the special section on AOP (Oct. 2001) with great interest. Was this the next programming revolution?, I asked myself. After all, I have seen many fads born in academia come and go during the past few decades. Some succeeded; most failed.

I remember the promises of Ada, AI, ego-less methods, persistent databases, and many other good ideas. Most did not achieve success because they failed to address the problems industry was facing as it built real systems for real customers. Post-mortems for failures I have been part of all pointed to the need for more and stronger practitioner involvement. Practitioners provide help by identifying real issues researchers need to be concerned with. The classic text on technology transfer, Crossing the Chasm, by Geoffrey Moore, points out that a conscious effort needs to be mounted as the technology is developed to get the early majority (practitioners) involved. The pilots conducted by academics and research scientists to prove concepts and illustrate potential gains attributed to a technology typically fall by the wayside. Practitioners aren’t researchers or graduate students working on a problem in a laboratory. Instead, they are professionals who build production software systems for a living. Such systems today are typically large, complicated, and involve considerable interactions (systems of systems).

To determine the potential of AOP concepts, I reread the articles to see if the 20 or more professors, graduate students, and research scientists involved in writing the articles had adequately addressed the issues of consequence for industry professionals, including interoperability, system performance, issues of scale, and technology transition). Needless to say, I believe they didn’t. "Why not?" Simple: They fell into the same traps researchers and academics have fallen into for years. They were so busy pursuing their ideas, they failed to plan for transition to industry. Such transitions take years and years. Therefore, I am extremely skeptical their work will fulfill the promises made in their articles. Of course, I will continue to follow aspects. But, I really believe that until the research community bridges the gap and works more closely with practitioners, we will continue to see repeat promising technologies sit on the shelf as they await practitioner approval. I wish the situation was different, but it isn’t. I have doubts that the promise of AOP can be realized before the research community bridges the gap.

Donald J. Reifer
Torrance, CA

Authors Respond:
As Reifer’s interesting letter points out, some innovations born in academia and research facilities have succeeded. I hope that aspect orientation will be one of them. Success is dependent on the merits of this new technology, which by the quick growth and attention it has gained, is promising. It is also dependent on success in technology transfer cycling. Here, hope is not enough; I agree with Reifer that a conscious effort needs to be developed as the technology is matured to get the early majority (practitioners) involved.

AOP has been deeply industry-connected since its beginning, trial projects are already well under way and commercial systems already use AspectJ and Hyper/J. The various research groups today have active efforts in this direction where collaboration with practitioners provide a good trial run of the ideas in the real world. (For more details, please contact the project leaders directly). You may have noticed that AOP has matured to AOSD or aspect-oriented software development, to support aspect orientation throughout the software life cycle. The community is organizing two classes of activities: workshops dedicated to aspect-orientation research to benefit from wisdom gained on the fundamentals of our science and workshops designed to get the attention of practical applications. We are also collaborating with industry, such as Rational, on AOSD/UML mapping. This, I hope, will accelerate the technology transfer cycling. I share the belief that the gap between the research community and industry should be bridged. Like marriage, a mutual conscious effort needs to be mounted. It requires both sides to reach out. I have no doubts that cross fertilization among researchers willing to share their innovations, make an effort to deliver it, take benefit from the trial run done by real-word applications, and practitioners willing to invest in long-term efforts and willing to put the time and effort required into the learning curve of new technologies and share their experience are the winners.

Social and Technical Aspects of Information System Design and Deployment

I was gratified to read in the column by Peter Denning and Robert Dunham ("The Core of the Third-Wave Professional," Nov. 2001) that "the social webs of relationships among people" are recognized as important to successful development, deployment, and use of computer-based systems.

The current growth of a distributed computing infrastructure and of applications and services supporting information and knowledge management serves to underscore the interdependence of the social and technical aspects of information system design and deployment. As a software engineer who has contributed to several successful customer- and user-centered product development projects, I believe information systems and software engineering curricula must connect system design and construction skill and competence to a grounded, empirical understanding of users’ work and of system usability in the users’ work environment.

Participatory design, a sociotechnical approach to collaborative system design and development, has been around since the 1970s and has contributed to many of the current user- and work-centered design perspectives. A good source of case material (and further references) on participatory design projects was published in the July 1993 Communications.

William L. Anderson
Rye, NY

Classroom Teaching Vs. Online Courses

In his column, "Revolutionizing the Traditional Classroom Course" (Dec. 2001) Roger Schank attempts to prove the superiority of online courses over classroom courses. To do so, he essentially tries to equate classroom teaching with plain lecturing. It is difficult to agree with him, since classroom teaching goes far beyond lecturing, at least with teachers who do it professionally and responsibly.

As obvious as it is for many readers, it seems proper to point out that classroom teaching is a creative intellectual activity—as creative as the work of a performing artist, such as a musician. The composer writes the score, and the performer uses his or her creativity, talent, experience, and inspiration to interpret the music and communicate it to the audience, often in a unique way. Teaching is similarly creative; conveying knowledge generated by others requires ingenious interpretation for an audience—interpretation that demands careful study and planning, interaction, constant feedback and adjustment, that influences, motivates, and inspires for further work.

Besides, classroom teaching of any subject could and should be entertaining. A good lecture on C++, for example, can be as entertaining as a good lecture on anything else (though it is probably more difficult to achieve a good lecture on C++ than, say, a good lecture on art history). So, is anyone able to provide all attributes of creative classroom teaching in a purely online course? I do not think so, at least not yet. However, there is a way a purely online course would be a proper substitute for a good classroom course.

Schank refers to the authority of Aristotle, Galileo, Dewey, and Einstein, and states that real learning is by doing. While true, it is not more than half the truth. In addition to doing, one needs to be thinking as well. As John R. Pierce puts it, "Human beings learn by doing and by thinking about what they have done." Capable and responsible teachers are good at provoking and stimulating thought and a desire for further study. As a matter of fact, this is the main purpose of a good teacher, not just plain lecturing of what is already in the book. Online courses can be an efficient substitute for plain lecturing, of course, but how much more can they do?

I agree the computer is a learning-by-doing device, and is going to revolutionize education. However, I believe the computer should not be a complete substitute for classroom teaching. To me, the computer is a great technological enhancement to teaching and learning—nothing more, nothing less. Enough destruction of creativity and individualism is caused by the TV networks. One can just hope that online education will not choose to have a similar effect on young people. Unfortunately, it could, especially if the role of the machine is overestimated and the role of the human underestimated.

Atanas Radenski
Orange, CA

No Unsolicited Email Ads for Me, Please

The article "Admediation: New Horizons in Effective Email Advertising" (Dec. 2001), by Gopal, Walter, and Tripathi seems to contain a single enormous flaw: the assumption that receivers of unsolicited commercial email (UCE) value some of what they receive.

The only commercial solicitations I have valued at all have come from companies with whom I have already done business; thus, these solicitations do not qualify as UCE. I have received countless UCE messages in the last couple of years, and I haven’t valued a single offer: pictures of naked people, fake degrees from any university and curriculum I care to name, prescription drugs without a prescription, get-rich-quick pyramid schemes, and illegal tax-evasion schemes.

In short, most of the UCE that I have seen makes illegal offers; and the rest offers only pornography. I don’t want any of it in my email in-box.

Rick Simkin
Chicago, IL

Authors Respond:
Contrary to the claim by Simkin, our operating premise is that UCE may be valued by some of the email receivers (albeit a few; typically much less than 1% of them). The low cost of sending commercial solicitations via email make them viable even at such low response rates. Therein lies the economic appeal, especially for those peddling pornographic and other objectionable products and services. These senders care little about building long-term relationships with consumers or producing negative public stigma.

We do agree with the sentiments expressed by Simkin that most users do not value any of the solicitations they receive. In fact, a key recommendation of the article is to push for legislation to make UCE illegal. Moreover, the concept of admediation we espouse works best when users are not harassed with UCE. Our underlying contention is that legislation against UCE does not mean an end to advertising via email; rather that effective legislation will indeed foster legitimate forms of email advertising that benefits both users and legitimate sellers through mechanisms such as admediation. The long-term success of email as an advertising channel depends critically on users being able to control the volume, nature, and timing of the email messages they receive.

A Chilling Virus

I just wanted to compliment Hal Berghel on a fine column ("The Code Red Worm," "Digital Village," Dec. 2001). A chilling story indeed. It occurred to me that worms and other virus artifacts are usually recognized only by their pernicious behavior. But what if they didn’t misbehave? I remember reading an excellent book called Plagues and Peoples, by William McNeill, identifying the necessary conditions for plagues to occur. He also identified what he called (I think this was the phrase he used) "macroscopic plagues" such as, say, Attila and his Huns. It was an interesting concept he associated with the more common use of "plague," which is the bacteriological or viral "microscopic" epidemic. His point was that, while plagues exploit limitations in the human immune system (micro) or societies (macro), they are also dependent on a set of social (read "system") conditions. He came close to asserting that such epidemics are a natural consequence of certain kinds of systems’ growths. Intuition tells me that the same is happening with the Internet. Perhaps we should add another flavor of plague, perhaps "infoscopic" or "Weboscopic."

Thanks for the column. Excellent work as always.

Phil Armour
Deer Park, IL