Forum

I would like to point out a few misconceptions conveyed in David Touretzky’s "Viewpoint" ("Free Speech Rights for Programmers," Aug. 2001, p. 23).

First, the notion that patents and copyrights are assured in the U.S. by Article 1, Section 8 of the U.S. Constitution, not by insinuating that lawyers have somehow usurped the public’s rights.

Next, the DVD CSS stream cipher was not cracked by some supposedly ingenious cryptography work by the purveyors of DeCSS. Rather, an incompetent implementation of the key management portion of a software player by a CSS licensee was discovered through reverse engineering activities performed by unnamed persons and subsequently disclosed by an adolescent Norwegian hobbyist who, interestingly as a foreign national, attempted to invoke protection under the guarantees of the U.S. Constitution.

Also, the financial penalties defined within the Digital Millennium Copyright Act refer to statutory damages that may be awarded as a consequence of convicting felons who traffic in circumvention devices and services. It is a financial penalty that may be imposed as the "court considers just" for essentially acts of intellectual property theft. The penalties are not, as Touretzky leads us to believe, the consequences of merely watching a movie in our own homes.

From Touretzky’s instructions and advice, it would appear that the creation of a C computer program that could be perfectly compiled and executed could contain binary objects consisting of digital images of Touretzky’s own book pages embedded within enclosing /* and */ statements. Free distribution of that program as an expression of free speech would seem to be acceptable to him. It is after all, a compilable program.

Touretzky’s defense of theft and felonious activities, defined as such by U.S. laws and upheld by the courts, as supposed threats to academic research and constitutionally assured freedoms calls into serious question his ability to discern the truths of these issues.

Scott E. Hamilton
Los Angeles, CA

David Touretzky Responds:
Hamilton has missed the distinction between simple copyright violation and unlawful circumvention (a DMCA violation). Republishing one of my books as a comment in a C program would be a copyright violation. But viewing a CSS-protected movie with an unapproved DVD player is an act of circumvention prohibited under 17 USC 1201(a)(1)(A). Unlike with copying, there is no "fair use" exception to permit harmless acts of circumvention in the privacy of one’s own home. And the statutory damages provided in 1203(c)(3) do not require any demonstration of economic harm. The MPAA admitted no harm was known to have resulted specifically from Eric Corley’s actions, yet it won its lawsuit.

Corley was found to have engaged in unlawful speech on the subject of circumvention technology. The implications for civil liberties and computer science may not trouble Hamilton, but they disturb many others. Eight amici (friend of the court) briefs were filed in support of Corley’s appeal to the Second Circuit. The sponsors include 46 noted law professors, the ACLU, the American Library Association, the Reporters Committee for Freedom of the Press, an ACM Turing Award winner, a past ACM president, several ACM Fellows, and the ACM’s own Law Committee. The briefs are available at www.eff.org.

While early DVD decryption software relied on a player key obtained through reverse engineering, newer programs, such as VobDec, employ an algorithm by cryptographer Frank Stevenson that breaks the weak encryption directly. Under the DMCA, Mr. Stevenson could be sued for having conducted his research, I could be sued for republishing it, and the New York Times and USA Today could be sued for linking to my Web site. That is why the DMCA must be overturned.

Back to Top

Outlawing Numbers Impractical

the August viewpoints "Free Speech Rights For Programmers," and "Pondering Pixelized Pixies" bring to light an interesting dilemma: Can certain bitstreams—or numbers if you wish—be illegal?

If such an outlawed bitstream (B) exists and it is exclusive ORed with a random bitstream (R) to yield B’, the combination of B’ and R together hold the outlawed data. Both R and B’ by themselves have no provable relation to B, but xOR(B’, R) yields the original outlawed bitstream B.

After the initial process, B’ and R are "symmetrical"; both are purely random bitstreams, neither of which can be said to contain more of the original bitstream B than the other. If I post B’ to my Web site and my friend posts R, no one can ever tell if it was me or my friend who must once have possessed the outlawed bitstream B. In court, I would argue I have posted a large random number for analytical purposes, and, of course, so will my friend.

No matter how immoral or threatening certain numbers may be, outlawing them does not seem practical.

Raymond Michiels
The Netherlands

Back to Top

Defining Information

I was pleased to read Peter Denning’s statement: "The distinction between ‘data’ and ‘information’, once carefully observed by computing professionals has all but disappeared" ("The IT Schools Movement," Aug. 2001, p. 19). It’s true; most computing professionals have never been exposed to Claude Shannon’s pioneering work on information theory, which explains many aspects of data processing, particularly, how five binary digits of storage might not contain five bits of information.

Unfortunately, I was dismayed when I read further and found the column’s definitions of these foundation terms.

Denning’s definition of the term "data" (symbols inscribed in formalized patterns) seems to exclude analog data and leaves a key concept (formalized patterns) completely undefined.

His definition of "information" ignores Shannon and implies that a sign bearing the symbol "WC" contains information if there is an observer who needs a washroom but not otherwise. This is neither a standard nor a useful definition of information.

His definition of "knowledge" (the capacity for effective action in a domain of human practice) implies that a paralyzed person who cannot speak has no knowledge.

David Parnas
Ontario, Canada

Peter Denning Responds:
Parnas and I agree that our foundational terms have lost their meanings and that it has consequences for education and practice in IT.

We don’t disagree about data. Although I was speaking from a purely digital context, I did not mean to exclude analog data. By "formalized patterns," we would both mean that the symbols are machine-recognizable and the allowable patterns of the symbols can be described by a grammar.

We also don’t disagree about knowledge. The ancient Greeks gave us the distinction between "knowing that" and "knowing how." The former is oriented to description, the latter to action. I defined "knowledge" in the knowing-how sense, which is operational and important for professional practice; show me your knowledge by demonstrating it. People who cannot speak can still demonstrate knowledge via devices they control; were they totally paralyzed, however, they would be unable to demonstrate knowledge. I recommend The Tree of Knowledge (H. Maturana and F. Varela, Shambala Books, 1988) for those who want to study the action interpretation of knowledge more deeply.

Parnas and I agree that information has several definitions. In some contexts, information has a precise, mathematical definition. In others, information depends on the meanings assigned to data by human observers. Shannon’s information is the most famous example of a precise definition; it measures the uncertainty of a message source as entropy over the probability distribution of the messages that can be sent by the source. Shannon, and his colleague Warren Weaver, explicitly intended to address only the accuracy of transmission of data, not the accuracy of the receiver’s interpretation (see C. Shannon and W. Weaver, The Mathematical Theory of Information, University of Illinois Press, 1963).

But the term "information" is commonly used in much broader senses, where the receiver’s interpretation plays a role. MIS designers are concerned with Drucker’s dictum: Get the right information to the right people at the right time. User interface designers are concerned that the information a user derives from reading the displays leads to safe action. Biologists are concerned with the information carried by DNA. The information for making management decisions is different from the information grasped by a user from a computer screen, which in turn is different from biological information; all three are different from Shannon information.

Some people are troubled by the ambiguities of the meanings of information. I am not. Information is at the boundary between the capabilities of machines and the capabilities of humans. Data is on the machine side of the line, knowledge on the human side; information overlaps both sides. Information is to IT people as life is to biologists. Biologists have no generally accepted definition of life; they use half a dozen criteria to assess when life is present. Although they may agree on the structure of an organism, they may not agree on whether that organism is alive. In the same way, IT professionals may agree on the structure of data and data processing systems, but not on what information is produced by the system.

I prefer the definition of information as the meaning assigned to data by human observers because I am concerned about action and design. I am better able to understand what actions people take with computers, and shape the computer to help them toward more effective action, if I listen carefully to their interpretations of the data processed by the computer.

Back to Top

Special Needs and Sound Scholarship

I have a long-standing interest in computing for people with special needs, and for many years I have been dismayed at the comparative neglect of this topic in discussions of related subjects. I still have some notes I wrote perhaps 10 years ago about a then-recent Computing Surveys article ("Human-Computer Interface Development: Concepts and Systems for its Management," 1989). They begin like this:

• "In an extensive review, [authors] Hartson and Hix describe the recent state of the art of computer interface management. They manage to do so while making no mention at all of the requirements of people with any sort of disability. …"

It is all the more curious in that many of the points raised by Hartson and Hix carry over smoothly into the area of interfaces adapted to people whose abilities are different from the norm.

Things have not changed. In the March 2000 Communications, in a special section on perceptual user interfaces, there was hardly any mention of rehabilitation topics. In that section, Reeves and Nass ("Perceptual Bandwidth," p. 65) discuss the extension of the range of conventional computer interfaces by adding "speaking, touching, gesturing, emoting, and gazing." They comment that, when involved in perception, "the computer is a stimulus machine, reconfiguring reality in different ways, and making it available to more, fewer, or different perceptual systems than might otherwise be engaged." This is precisely the function of communication systems used by disabled people.

The theme of the March 2001 Communications was "the next 1,000 years." We might infer from its contents that through the next millennium disability issues will continue to be ignored. Rita Colwell ("Closing the Circle of Information Technology," p. 31) hopes something will be done; there’s an accompanying picture of an embosser for Braille and tactile graphics, which seems to have little relevance to the adjoining article, and doesn’t begin to mention the real problems of tactile graphics. Are there others? If so, I’ve missed them, but there can’t be many.

I am moved to action. My principal subject is: When are computists going to take seriously the issues related to computer use by people with disabilities? I also have a second subject, the importance of which has impressed itself upon me over the years: Extension of topics to include extreme cases is essential for a healthy scientific discipline.

Why should computists take rehabilitation systems seriously? One answer is that rehabilitation got there first. All five modes mentioned by Reeves and Nass have been considered in some fashion in rehabilitation computing, in some cases for decades (see, for example, papers in A.D.N. Edwards (Ed). Extra-Ordinary Human-Computer Interactions, Cambridge University Press, 1995), as have most items in a list of future interaction modes given by Donald Norman in the March 2001 Communications: "by hand, foot, and body motion; by the speed and forcefulness of our activities; by our thoughts, feelings, and emotions; by where, how, and when we look; by speech and sound; by music and touch." This isn’t an accident; rehabilitation communication systems are used not merely for computing but for communication—using the telephone, chatting to friends, getting about, shopping. The system requirements for such normal human interaction go far beyond the conventional HCI topics (see Creak, G.A., "When HCI Should be HCI," Info. Tech and Disabilities, Nov. 1999).

The March Communications also introduces my second subject, albeit negatively. Edsger Dijkstra ("The End of Computing Science?," p. 92) asserts that the academic study of computing has failed—or has not yet begun. He mentions "there is a widespread belief that computing science as such has been all but completed"; they thought the same of physics in the late 19th century. What do we really do in computing? A few of us address theoretical topics and have built up a significant discipline (another subject also essentially absent from the issue). The majority go ahead as good craftsmen do, devising ways and means of working with their material, much as the woodworkers and stonemasons of the Middle Ages must have done. Their craft is drawn from experience of working with the material rather than from the properties of the material itself.

The craftsmen, diligent and intelligent, get lots of things done, which work, most of the time. So did the woodworkers and stonemasons; some of their magnificent achievements are still with us (if only some modern software were as beautiful), but they worked with skill and intuition, common sense, experience, and rules of thumb, and their constructs, however beautiful, were inefficient and too heavy, dark, and inconvenient and fell down in earthquakes. We had to wait for physics before we could understand in a different way how the buildings stood up and were able to design them in structurally superior ways. (Beauty is another issue.)

Most of computing today is at the craft level. Where might we find the science? Where did we find physics? Consider Sir Isaac Newton. The story of his development of the theory of gravitation from watching an apple fall to the ground might be apocryphal, but it is nevertheless plausible. You ponder an observation, and wonder why it is so. Newton postulated an attractive force between the apple and the Earth and formulated his theory of falling apples. How did that become a theory of universal gravitation? One significant fact was Newton’s realization that the same force could account for the moon’s stable orbit around Earth. You believe you have a real theory when you find it applies equally to a wide range of dissimilar systems.

Apply that thought to your pet topic. Rehabilitation systems are my pet topic, but the principle is universal. HCI is the discipline behind communication systems for rehabilitation. If we want a science of HCI, we should study as wide a range of interfaces as we can. The human side of the interface has been thoroughly studied, but on the computer side the craftsmen have almost universally stuck with what they know, which is two ordinary agile hands on fully operating arms, accompanied by good color vision and adequate hearing. Not surprisingly, the interfaces we get work well if you have this equipment.

If we want science, we have to cast our nets wider. If we get science, we can expect it to apply to all cases, which is just what we want for rehabilitation systems. We would like a real scientific base for the HCI that we can use to ensure interface design works as well as it can for everybody. To do so, we must study the whole range of possible interface states, not merely the comfortable middle ground; the apple alone wasn’t enough to support a universal theory. This is why special interfaces, such as those used by people with disabilities, are of importance; they provide extreme cases of communication systems. That’s why I was so worked up by the Harton/Hix article; it’s called "Human-Computer Interface Development: Concepts and Systems" but neglects the rehabilitation field despite its significance as an extreme case.

Which brings us back to where we began. Perhaps it’s appropriate to conclude with another short excerpt from my ancient notes—and another special section of Communications.

Glinert and York ("Computers and People with Disabilities," May 1992, p. 32) introducing a special section on computers for people with disabilities, describe the purpose of the section as follows:

• To show that research related to computers and people with disabilities is a valid area of scientific endeavor, and in particular of computer science; and
• To show that "design for disability" in computer and telecommunications is not only cost-effective but also leads to beneficial innovations for users who are not disabled."

When do we start?

Alan Creak
Auckland, New Zealand

Back to Top