I applaud Peter J. Denning's Viewpoint "Remaining Trouble Spots with Computational Thinking" (June 2017), especially for pointing out the subject itself is often characterized by "vague definitions and unsubstantiated claims"; "computational thinking primarily benefits people who design computations and . . . claims of benefit to nondesigners are not substantiated"; and "I am now wary of believing that what looks good to me as a computer scientist is good for everyone." Moreover, the accompanying table outlined various historic definitions of "computational thinking," including a comparison of what Denning called the "new" and the "traditional" view of the subject. However, my own interest in computational thinking differs somewhat from Denning's. First, I question the legitimacy of the term "computational" itself. Why say it, when the very subject is "computers" and the chief academic approach to their study is "computer science"? If one looks at how computers are actually used, it may come as a surprise to learn that few such uses actually involve computing. For example, applications that deal with scientific and engineering problems are of course heavily computing-focused, but, last I heard, they constitute only approximately 20% of all applications being developed worldwide. The most predominant applicationsthose for businessinvolve little computation beyond arithmetic. And systems programs like operating systems and compilers, the focus of much computer science study, historically at least, involve little or no computation and primarily concern manipulating information rather than numbers.
The problem is that computational-thinking enthusiasts, as Denning wrote, are driven to spread the subject across all academic majors. I certainly believe in the importance of programming and using computers for the variety of applications for which they provide benefit and that educational systems worldwide should provide the knowledge and skills that would help students move into the field, should that be their preference. But should computational thinking also be taught to artists, writers, poets, physicians, and lawyers? Not as I see it . . .
The faulty thinking behind the "computer science for all" approach to pedagogy is best seen in Denning's table, labeled "Traditional versus New Computational Thinking." Its entry on "domain knowledge" suggested traditionalists see domain knowledge as vitally important to the person doing the computational thinking, while "new" thinking says the importance of computational thinking is domain-independent. As a practicing programmer who has dabbled in many different application domains over a long professional career, I see it as beyond understanding how anyone could fail to see the importance of deeply knowing a domain to being able to solve problems in that domain.
Robert L. Glass, Toowong, Australia
Computational thinking is the habits of mind developed from designing computations. The meaning of computation has evolved from the 1960s "sequence of states of a computer executing a program" to today's "evolution of an information process." This changed meaning reflects the ever-expanding reach of computing into all sectors of work and life. Many of today's most popular apps feature computations well beyond arithmetic, as in, say, facial recognition, speech transcription, driverless cars, and industrial robots. The computational thinking developed by those who worked on these achievements is much more powerful than the handful of programming concepts offered as the definition of "new CT."
Peter J. Denning, Monterey, CA
Time to Retire 'Computational Thinking'?
Peter J. Denning asked, "What is computational thinking?" in his Viewpoint (June, 2017), then quoted the following definition by Al Aho: "Abstractions called computational models are at the heart of computation and computational thinking. Computation is a process that is defined in terms of an underlying model of computation, and computational thinking is the thought processes involved in formulating problems so their solutions can be represented as computational steps and algorithms." But as Aho's definition is highly circular, it reveals very little.
All disciplines rely on models. The only specifically computational word here is "algorithms." If we replaced it with similar words, like "procedures" or "sequences," we would arrive at such vacuous "definitions" as, say, "Medicine is a process that is defined in terms of an underlying model of medicine, and medical thinking is the thought processes involved in formulating problems so their solutions can be represented as medical steps and procedures." And "Drama is a process that is defined in terms of an underlying model of drama, and dramatic thinking is the thought processes involved in formulating problems so their solutions can be represented as dramatic steps and sequences." One could analogously "define" musical thinking, artistic thinking, chemical thinking, and so forth.
Unless somebody can come up with a more insightful definition, it is indeed time to retire "computational thinking."
Lawrence C. Paulson,
Toward a True Measure of Patent Intensity
In their article "How Important Is IT?" (July 2017), Pantelis Koutroumpis et al. described a methodology for assessing the importance of information and communications technologies (ICTs) compared to non-ICT technologies, using PatStat, a dataset from the European Patent Office of 90 million patents awarded from 1900 to 2014. Controlling for variables (such as patent office, year of grant, and patent family), they concluded ICT patents are more influential than non-ICT patents because they receive significantly more citations and a considerably higher PageRank.
When one publication (not just those involving patents) is cited more often than some other publication, the more-cited one is thus more influential. However, patent publications are unique because they not only describe novel systems and methods but also hold commercial value and represent licensable assets for their holders. A patent may be cited hundreds of times yet still have relatively low financial value; on the other hand, a patent may be cited only rarely yet reflect enormous valuation.
Consider that in 2013, Kodak, the company that invented the digital camera, sold its portfolio of 1,100 digital photography-related patents to multiple licensees for $525 million (or $477.3K per patent). Earlier, Google bought Motorola Mobility and its 17,000 patents for $12.5 billion (or $735.3K per patent), and Microsoft acquired 800 patents from AOL for $1.06 billion (or $1.33M per patent). Snap paid the exceptional price of $7.7 million for Mobli's Geo-filters patent, believed by TechCrunch to be the highest amount ever paid for a patent from an Israeli tech company. However, the valuations of most patents are unknown until they are indeed auctioned or sold off. For instance, ICT-related patents (such as those involving Google's and Microsoft's methods for faster Internet browsing)1 may have impressive valuations, but those valuations are difficult to predict before actually being auctioned or sold off.
Considering non-ICT patents, the revenue streams of several pharmaceutical companies depend on patents and their corresponding expiration dates, and one patent could be worth billions over the course of its licensing period. Notable patented medications include Pfizer's Lipitor (for lowering fatty acids known as lipids), Bristol-Myers Squibb's Plavix (for preventing heart attacks and strokes), and Teva's Copaxone (for treating multiple sclerosis). Other non-ICT patents that have significantly and directly improved people's lives are cited only rarely, including those related to agriculture, transportation, and creation of new materials.
In the most recent U.S. Patent and Trademark Office's economy update,2 the non-ICT "basic chemicals" category ranked first, with $64.5 billion in merchandise exports of selected intellectual-property-intensive industries, while "semiconductors and electronic components" was second at $54.8 billion. Most industries involve non-ICT technology. As for "patent intensity," or the ratio of patents to employees measured as patents/thousand jobs, "computer and peripheral equipment" and "communications equipment" topped the list, though this was due directly to the relatively high number of patents issued in the industry versus the industry's relatively low number of employees. Conclusions regarding level of influence of ICT technologies versus other types of technologies should thus be reported with care when a comparison is based solely on number of inventions and citations.
If such influence is indeed the basis for a comparison, then additional covariates should be controlled for, including the mean estimated valuation per patent, number of employees in the industry, and additional financial and industry-specific characteristics.
Uri Kartoun, Cambridge, MA
Although there may be some correlation between patent price and technological influence, the relationship is neither clear nor systematic. Patent prices are more likely driven by how incremental/radical/breakthrough it is, whether its value is standalone or as part of a bundle, projected commercialization timescale, cost versus risk, bidder's experience, patent age, rate of technological change, and substitution and reverse-engineering risk, to say nothing of broader economic factors. Perhaps our technological-influence measure could thus be used to help understand patent pricing.
Pantelis Koutroumpis, London, U.K.,
Aija Leiponen, Ithaca, NY, and
Llewellyn D W Thomas, London, U.K.
2. U.S. Patent and Trademark Office. Intellectual Property and the US Economy: 2016 Update. U.S. Patent and Trademark Office, Washington, D.C., 2016; https://www.uspto.gov/sites/default/files/documents/IPandtheUSEconomySept2016.pdf
Communications welcomes your opinion. To submit a Letter to the Editor, please limit yourself to 500 words or less, and send to [email protected].
©2017 ACM 0001-0782/17/09
Permission to make digital or hard copies of part or all of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and full citation on the first page. Copyright for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, to republish, to post on servers, or to redistribute to lists, requires prior specific permission and/or fee. Request permission to publish from permission[email protected] or fax (212) 869-0481.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2017 ACM, Inc.