Make new statistical knowledge widely available, in order to provide benefits to society at large beyond your own scope of applications. Statistical methods may be broadly applicable to many classes of problem or application. (Statistical innovators may well be entitled to monetary or other rewards for their writings, software, or research results.)
Make new statistical knowledge widely available in order to benefit
society at large. (Those who have funded the development of new statistical
innovations are entitled to monetary and other rewards for their resulting
products, software, or research results.)
T.S. Lim
tslim@recursive-partitioning.com
February 25, 2001
The paragraphs of the ethical guidelines that you cite were among the most carefully and widely reviewed elements of the guidelines document. Before offering my personal interpretation of them, I would like to point out that my term as Chair, Committee on Professional Ethics has expired. That position is now held by William Seltzer at Fordham University, a Cc: on this message. Dr. Wang remains a member of the Committee. I can speak to the derivation of those paragraphs and the thought processes that led to the final wording. I do not speak for the ASA or the Committee in doing so.
The basic intent of these guidelines is to respect fully the intellectual property rights of innovators of statistical software, algorithms, mechanical devices, or "expressions" of theory that may be subject to U.S. (or other) patent or copyright law. Conversely, the document asserts that the scientific knowledge (e.g., the statistical theory or methodology itself) represented by a protected item must be placed in the public domain. The former consideration is a matter of law. The latter consideration is the ethical issue.
The first major writer on the subject of statistical ethics was W. Edwards Deming. A disciple of Walter A. Shewhart, he revolutionized and popularized the concept of statistical quality control. He realized that the science of statistics is an immense resource which must be made available for the use of all humanity and in all fields of application. A consultant to major industrial concerns, he consistently declared that any methodological development he might make in service to a particular client could not belong either to the client or to himself as a trade secret. It must be published in order to benefit the profession (and society) as a whole. Related to that very firm position, he retained close control over any description of his statistical methods and followed normal copyright procedures for his professional publications. He also was adamant in protecting the confidential data of his clients, including both proprietary knowledge and personal data.
Starting at the 2000 Joint Statistical Meetings, I teach a CE course in statistical ethics. When asked why a statistical methodological development cannot be ethically withheld from the public domain, I use the following reasoning:
Suppose that you own a statistical consulting firm which develops a brilliant innovation in industrial quality control. Your innovation allows your clients to cut the cost of their quality monitoring operations by 20-30%. Why should you not keep that a secret in order to retain an advantage over your competitors? My answer is that you would never fully exploit your innovation; by denying information about it to the rest of the profession, you would be preventing a potentially vast array of socially useful applications in areas you do not work in. For example, environmental monitoring employs much of the science behind quality control. If you publish or patent your innovation, you can assure yourself the credit and the appropriate financial rewards for your creativity. Beyond that, you allow others to improve the cost efficiency in applications such as environmental monitoring, which benefit all of society. The better environmental quality will be of value to society at large, including yourself, your children and their children.
The basic principle espoused here is also found in the Copyright Clause in Article 1 of the U.S. Constitution. That principle says that the basic factual knowledge of science and the useful arts cannot be privately owned. Expressions of that knowledge or specific means of practical implementation of that knowledge are rightly preserved to the innovators for some length of time in order to encourage such innovation and in order to place the knowledge itself in the public domain.
As is appropriate for an ethical guideline, the ASA ethics document
only states the basic principle. The detailed application in specific cases
is subject to considerations of law, technology, and economic realities.
In publishing the guideline, the ASA implies an ethical obligation to consider
the greater social and scientific good, in addition to those matters of
law, technology, and economic reality.
John Gardenier
JGardenier@cdc.gov
26 Feb 2001
If I publish a scientific paper describing my algorithm, then someone else or myself can program the algorithm and sell the software package. People who are not willing to pay for the software can program the method themselves.
Another related issue, what if the agency which is funding the research
decides to make the invention proprietary? So, people would have to purchase
the invention to use it. The invention, of course, won't be published in
any scientific journals.
T.S. Lim
tslim@recursive-partitioning.com
February 26, 2001
In general algorithms are not patented, but they are published in copyrighted works. All that is protected is the "expression", that is, the text of the document as published. Normally, it is the publishing journal that owns the copyright. Anyone is allowed to use the knowledge; they cannot reproduce and sell the document without permission. Also, no one else could claim credit for discovering the "Lim algorithm" if you are the first to publish it.
It seems from your latest message that you want to make sure that nobody else ever knows how you perform your statistical work, feeling that will give you a competitive advantage. The only way you can do that is to keep it a "trade secret" and never reveal it to anyone unless you have total confidence they will not divulge your secret.
There are two practical problems with that approach. It may be hard
to convince clients that you are doing responsible and reliable work if
you hide the methods in a "black box." Secondly, without benefit of peer
review, there may be important weaknesses in your algorithm which you do
not detect. And of course, from an ethical point of view, it would reflect
social and professional irresponsibility.
John Gardenier
JGardenier@cdc.gov
27 Feb 2001
True, but people who'd like to use your patented invention have to pay
license fee. Hence, IMHO, it limits the distribution of your innovation
quite a bit.
Algorithms have been and are being patented in the US. Go to http://www.recursive-partitioning.com
and click on Bibliography/Patents for examples. The situation in the UK
is different. It appears that it's much easier to patent algorithms in
the US.
I'm just playing the devil's advocate here. I can interpret the ASA ethical guidelines as ASA endorsing patenting innovations or making them proprietary. What do you think?
Proprietary algorithms are rather common in the data mining world. Data
mining people in general don't write papers when they get new ideas. Instead,
they code the ideas, start a company, and sell the software. I'm just afraid
that the "bad" habit of computer scientists would infect us.
As a science ethicist, I feel that all basic scientific knowledge, including the formulation, derivation, and testing of statistical algorithms should be in the public domain. As to how much information should be available in the patent records, I leave that to the Patent and Trademark Office of the Department of Commerce. As to when an algorithm has been "adequately" made part of the scientific record, I feel that must be addressed on a case by case basis.
Granted that there may be technical and ethical issues in the details
of specific patents, neither the Committee nor the ASA as a whole have
even attempted to sort those out. If you feel that there are specific cases
of algorithm patenting which are problematic and interesting, I would encourage
you to write those up as case studies.
John Gardenier
JGardenier@cdc.gov
28 Feb 2001
My personal opinion on this issue is that I'm against patenting statistical
innovations or making them proprietary. With the rise of data mining and
KDD, I believe we need to make it clear where we stand with respect to
patenting and proprietary algorithms.
T.S. Lim
tslim@recursive-partitioning.com
February 28, 2001
Thank you for raising the issue of the patenting of Statistical algorithms and for your persistence on the matter in dealing with John Gardeniers' responses. Thanks are also due to John for his persistence in addressing the points you raised. There is little I have to add to the discussion and I certainly welcome the Chamont's suggestion that this Q and A be added to our website.
I have only two points. First, a substantive one: People sometimes look to ethical guidelines as providing clear-cut rules for ethical behavior. While this is sometimes the case; often the situation is more complex. As the second paragraph of the preamble of the ASA Guidelines states "these guidelines may be partially conflicting in specific cases." In other words, the Guidelines embody a number of sound ethical principles that, taken together, can help us be ethically aware statisticians. The balance of the same preambular paragraph offers some advice how one might address, and perhaps resolve, the ethical issues involved in a specific situation in a responsible manner. Your exchange with John is part of such a process.
Second, I would invite other members of the ASA Committee on Professional
Ethics an opportunity to comment on the exchange so that the final posting
by Chamont could reflect the full diversity of Committee thought on the
matter.
William Seltzer
seltzer@fordham.edu
March 07, 2001
The exchange with Dr. Lim on patenting of algorithms is fascinating. Some time ago one of my students showed me a method of analyzing medical data that had been patented. I wouldn't even call it an algorithm--it was an approach that was interesting but didn't seem more than a variation of current procedures. I remember being flabbergasted that something like that could be patented. I don't know whether I can still find that material or whether it would be appropriate for me to share it with the Committee.
I agree with Dr. Lim that algorithms shouldn't be patentable, but I don't think we have the authority to make decisions with respect to patent law. I'd feel uncomfortable telling people that although they have the right to patent something under the law, it's unethical for them to do so.
Anyway, I certainly agree that this would be an interesting exchange for our web site.
Juliet Shaffer
shaffer@stat.berkeley.edu
March 9, 2001
To: seltzer@fordham.edu
I have a question about copyright, but it's different from the other issues being discussed here. If there's some more appropriate forum, I hope you'll let me know.
Here is my question. I published an article in Journal 1. I have a Paper in press in Journal 2 that uses a similar figure to that in Journal 1. The copy editor of Journal 2 says must get permission from Journal 1 to use the figure. I believe that this would be appropriate if I reproduced the figure exactly. However, the new figure is based on some if the same data, but is redrawn, retitled, etc. It doesn't seem reasonable that I couldn't publish, say, a scatterplot of data without permission from journal 1 if I needed it in a different article. What if I made the same figure but with twice as much data? Would Journal 1 still own the copyright?
I guess I'm asking whether there are guidelines somewhere about who owns a particular figure, or how different two figures would have to be to claim that Journal 1 didn't own the copyright. (To speed things along I am going to ask journal 1 for advice, but would appreciate knowing more about this issue for the future).
Thank-you for your consideration.
Paula Diehr
paula@diehr.com
June 26, 2003
Dear Paula Diehr,
Thank you for your question. This is a quick response to your query. After other members of the ASA Committee on Professional Ethics have had an opportunity to comment on your question, I may forward additional reactions to you.
I think the question you raise, as important and as personally frustrating as it seems to be, is really one of copyright law and practice rather than one of ethics.
The ASA's "Ethical Guidelines for Statistical Practice" is in fact silent on the specific issue of copyright. The Guidelines do contain the general admonition in Section C. of the Preamble that "All statistical practitioners are obliged to conduct their professional activities with responsible attention to: ... 5. Adherence to all applicable laws and regulations, as well as applicable international covenants, while also seeking to change any of those that are ethically inappropriate." In addition, in the body of the Guidelines, under "A. Professionalism", point 9 states: "Respect and acknowledge the contributions and the intellectual property of others."
From an ethical perspective the issue would seem to involve whether or not you are avoiding your legal responsibilities and ethical duties to acknowledge the "contributions and the intellectual property of others." These responsibilities and duties would appear, in turn, to depend on (a) how much independent effort Journal 1 put into rendering the figure as published therein and (b) what copyright law and practice have to say on the matter. I don't think ethics can answer these subsidiary questions. Rather I would look to a good copyright attorney for the most authoritative advice on copyright law and to experienced journal editors for advice on the relevant copyright practice and the effort of Journal 1 in rendering your figure.
In response to your interest in gaining more familiarity with some of the issues that may be involved, you may want to take a look at some of the material on one or more websites on the subject (for example, <http://fairuse.stanford.edu/> or <http://www.law.cornell.edu/topics/copyright.html>). In addition, many author's manuals (for example, The University of Chicago Manual of Style) contain guidance on copyright and permission issues. Of course, these general sources are unlikely to provide you with the most authoritative advice on the specific questions you have.
Again, thank you for your question.
Sincerely yours,
William Seltzer, seltzer@fordham.edu
Chair, ASA Committee on Professional Ethics
June 26, 2003