Wegman and Said 2011, part 2

I continue the previous discussion of  unacknowledged antecedents in  Color Theory and Design by Wegman and Said (WIREs Computational Statistics, 2011), and examine the second half of the article in detail. There is an excessive (and partially unattributed) reliance on Marc Green’s web page on the subject.  An analysis of the list of references and figures shows a disturbing failure of the authors (who are also two thirds of the editorial team) to follow WIREs own guidelines. In all, at least 10 of 17 references appear to be spurious, and 12 of  the 17 figures are not properly attributed. All told, there are at least 12 different identified sources of unattributed text and figures, including five Wikipedia articles. This pattern raises questions concerning the fitness of the authors for editorial duties.

Introduction

In the interminable saga of the dubious scholarship of GMU profs (and  “hockey stick” report authors) Edward Wegman and Yasmin Said, surely one of the most astonishing developments has been their accession to the editorship of WIREs Computational Statistics (along with report co-author David Scott).

Recently I took a first look at this journal by examining Wegman and Said’s  Color Theory and Design. I showed that much of that article was based on a number of unattributed web-based antecedents dating back 10 years, as evidenced by “flow through” cut-and-paste material in a 2002 Wegman lecture. (That lecture, along with the rest of Wegman’s archived course lectures, has since disappeared, but it lives on at Archive.org [PDF]). Those sources were augmented by several newer online articles, mostly from Wikipedia.

As a reminder of the overall extent of dubious scholarship, here’s the big picture I presented last time around.

You can also follow along in the detailed analysis,  Dubious Scholarship in Full Colour, which includes a section with side-by-side comparison of most of the paper to its antecedents (starting at p.6).

A closer look

The second half of the paper does have less egregiously copied material as the first half. That could be because the first half relies much more directly on Wegman’s lecture, which itself was mainly cut-and-paste from the original sources. But the second half is certainly not without its fair share of problems.

I’ll start at p. 8 of the article and the section on Color Deficiencies in Human Vision. This and a subsequent sub-section on Hard-Wired Perception are acknowledged by Wegman and Said to be based on Marc Green’s SBFAQ (“should-be-frequently-asked-questions”) on colour vision. Interestingly, that’s the only one of the three main “flow through” antecedents that still exists on the internet.

Acknowledgment or not, the subsection on The Elderly follows Green very closely indeed, matching it sentence for sentence. Here is Green’s opening (as usual, identical wording are highlighted in cyan and trivial changes in yellow).

Vision declines with age in several ways, but the most relevant for color design is the  yellowing and darkening of the lens and cornea and the shrinking pupil size. Yellowing selectively blocks short wavelength light, so blues look darker.

Compare that to Wegman and Said’s version:

Deficiency of vision in the elderly is a result of yellowing and darkening of the lens and the shrinking of pupil size. Yellowing of the lens has the result of blocking short wavelength (blue) light so that the intensity of bluish colors is diminished.

As usual, there is the typical say-less-with-more style of paraphrasing (verbose repetition of “result of”  or “of the lens”). Not only that,  the meaning has been changed. While the original makes it clear that changes in the lens and pupil affecting colour vision are only part of declining vision, Wegman and Said’s version implies that all “deficiency of vision” is caused by these changes alone.

Things get worse in the next sentence.  Here’s Green again:

Moreover, the elderly have difficulty discriminating colors which differ primarily in their blue content: blue-white, blue-gray, green-blue green, redpurple, etc.

Wegman and Said give a different list of colours:

…[C]yan, blue-gray, light blue, magenta will be affected and more difficult to distinguish.

Here, they have completely missed the point; Green is saying that the elderly find it more difficult to discriminate between certain pairs of colurs, such as red and purple, or green-blue  and green, pairs which differ in the amount of blue present. Ludicrously, Wegman and Said’s reworking implies the very opposite, suggesting possible problems with cyan and magenta, which have a common blue element and differ in red and yellow green content.

The next sub-section (on colour blindness) departs eventually from Green and veers into a discussion of “oculocutaneous albinism”, which is lifted, not from Wikipedia, but the All Experts website (with a small excision, presumably to avoid a confusing reference to previous discussion not shown).

But, in order to exhibit albinism, one must inherit a mutant allele from both the mother and the father (whereas males only receive an X chromosome from the mother),thus males and females are affected equally for autosomal disorders. The disease is characterized either by the failure to synthesize pigment proteins or the failure of integretory proteins to implement them into tissue.

Then it’s back to Green and a discussion of Color Design, beginning with a sub-section on Hard-wired Pereception. As noted before, Green’s influence on this sub-section is acknowledged in a general way. However, three figures used to illustrate the various points have no attribution; it turns out these are very much “after” Green’s original figures, shown below.

And here are Wegman and Said’s closely inspired versions.

This is followed by a meandering discussion of colour models in a sub-section entitled Color Design Based on the Color Wheel.

Wegman and Said’s figure 16 has no attribution, but does mention the images’ use under GNU Free Documentation License, a surefire tipoff that Wikipedia is the source.

Sure enough, the History section of the Wikipedia article on RYB Color Model provides both figures, albeit reversed. That also appears to be the main source for a very short discussion of the historical perspective, which still retains references to Newton’s Opticks, and two obscure “documents” by Goethe and Chevreul (thus boosting the rather meager list of references).

Less obvious is the provenance of the final figure, which illustrates the more modern RGB (red-green-blue) system of primary colours.

The left hand figure is easily identified as the RGB colour wheel, used in such Wikipedia articles as HSL and HSV. But the right hand one proved a little trickier. One clue lies in the meaning of the term HSV, which refers to the three elements of hue, saturation and value (sometimes called intensity or lightness). A colour wheel, though, only describes the hue element of the full HSV colour model; thus, it is somewhat bizarre to refer to an “HSV” colour wheel.

Indeed, the so-called HSV colour wheel has an “H” inscribed on the top, but is missing S and V. And so it turns out that the S and V portions of the diagram (which comes from a page about the Sputnik Wiki engine) have simply been excised!

Now that’s scholarly creativity and rigour for you. Since the previous RYB figure featured a “star” and “circle” pair, the authors have managed to find, with a bit of tweaking,  a way to produce the missing RGB “circle” not available on Wikipedia.

The last subsection covers Color Design Based on the Color Wheel. It lists and briefly describes six colour schemes:

1. Monontone achromatic
2. Monotone chromatic
3. Analogous hues
4. Complementary
5. Split complementary
6. Triad

Only one other source has this exact set of phrases and words; as you may have already guessed, once again it’s Marc Green’s SBFAQ. And, just like Green’s list of wavelengths from near the opening of the article, this  passage is completely unattributed, providing a fitting book end to my analysis.

Applying the Guidelines

All this dubious scholarship is bad enough, but matters only get worse when one steps back and examines it in light of the standards set forth in the WIREs Comp Stats Guide for Authors. There’s a separate guide for each journal in the growing WIREs series, signed by the respective editorial team. But since WIREs features a common “templated” format and mix of article types across all the journals, much of the guide is identical “boilerplate”.

The description of the Overview article type is in keeping with the WIREs philosophy of “citation-rich” scholarship:

Overviews will provide a broad and relatively non-technical treatment of important topics at a level suitable for advanced students and for researchers without a strong background in the field.
Average extent = 5,000-8,000 words, 10-16 figures/tables, 50-100 references, 10-14 pages.

Certainly, the material of Colour Theory and Design, with its mishmash of  non-scholarly internet sources, does not aim too high (although its confused structure and exposition will limit its accessibility). But 50-100 references is way over the paltry 17 Wegman and Said have provided.

And even those 17 represent less than meets the eye at first glance, so to speak. To understand this, here is the list from the article overview page (you can click on the reference tab to see the list online).

1 Goldsmith TH.  Optimization, constraint, and history in the evolution of eyes Q Rev Biol 1990 65:281-322
2 Munsell AH.   A Color Notation G. H. Ellis Co Boston, MA 1905
3 Munsell AH.  A pigment color system and notation Am J Psychol 1912 23:236-244
4 CIE  Commission Internationale de l`Eclairage Proceedings, 1931 Cambridge University Press Cambridge 1932
5 Smith T, Guild J.  The C.I.E. colorimetric standards and their use Trans Opt Soc 0 33:73-134
6 Wright WD.  A re-determination of the trichromatic coefficients of the spectral colours Trans Opt Soc 1928 30:141-164
7 Guild J.  The colorimetric properties of the spectrum Philos Trans R Soc Lond (Ser A) 1931 A230:149-187
8 Lee H-C.   Introduction to Color Imaging Science Cambridge University Press Cambridge 2005
9 Crayola Website. http://www.crayola.com/colorcensus/history/chronology.cfm. (Accessed November 30, 2010).
10 Jameson KA, Highnote SM, Wasserman LM.  Richer color experience in observers with multiple photopigment opsin genes Psychonom Bull Rev 2001 8:244-261
11 Hayutin A.   How Population Aging Differs Across Countries: A Briefing on Global Demographics Stanford Center on Longevity Stanford, CA 2007
12 Green M. 2004. http://www.visualexpert.com/. (Accessed November 30, 2010).
13 Luo Q, Wegman E, Fu X.  CrystalVision, a Visual Data Mining package for Windows  2000
14 Newton I.   Opticks The Royal Society London 1704
15 Goethe JWv.   Zur Farbenlehre Harenberg Kommunikation Dortmund 1810
16 [link] Chevreul M-E.  De la Loi du Contraste Simultané des Couleurs et de l`assortiment des Objets Colorés  1839
17 Mante H.   Color Design in Photography Van Nostrand Reinhold New York 1972

A full ten references are found in the same Wikipedia articles previously identified as antecedents. For example, the Munsell references (2 and 3) are listed in the Wikipedia article on the Munsell Color System that was identified as the likely source for text and figure appearing on p. 4 of Wegman and Said 2011. And the historical references (14-16) are those found in the Wikipedia RYB Color Model, although further research was apparently necessary to come up with the original German and French titles.

This pattern strongly suggests that these are not bona fide references, and are simply padding and obfuscation. Meanwhile, of course, the real Wikipedia sources are unacknowledged.

Leaving aside for the moment the special case of Marc Green (reference 12), I’ll reserve judgment on the other six references; some may well be bona fide sources, if somewhat odd for an encyclopedia article, such as the Crayola website (reference 9).

Now let’s turn to the list of 17 figures (again click on the appropriate tab at the Wiley article page to see the whole list). They fall into three categories (each figure number has been linked to the corresponding figure image):

• Fig 2, 3, 6, 11, 14. Properly attributed and acknowledged (including photos from Wegman and Said, screenshot of CrystalVision). Total: 5.
  
• Fig 1, 7, 8, 9, 10, 16,17. License format is given, but no attribution; all from Wikipedia. Total: 7.
• Fig 4, 5, 12, 13, 15. No attribution; reworking of figures from Ted Park Color Theory Page (4, 5) and Marc Green (12, 13, 15). Total: 5.

One might think that proper attribution or credit is so basic that it need not be mentioned. But in fact the Authors’ Guide is quite explicit.

Please include the necessary credit lines in the appropriate places in your article and send us the completed permission request forms when you submit your manuscript. The language should be the exact language used by the copyright owner, or, if nothing is specified, should include the title, author’s name, previous publisher, and the date of copyright. … Credit lines will often accompany figures and illustrations; they should be included at the appropriate place in the figure legend or text.

In other words, the source must be credited and permission obtained. In the case of the Wikipedia figures, blanket permission is already given by various open licenses (usually Creative Commons and/or GNU). Somehow this preclearance has been interpreted as obviating the need for any credit at all!

The third group of figures, all of which I covered above and in part 1, are even more problematic. Since there is no attribution (let alone permission), the presumption would be that these are the original work of the authors. While the authors may have regenerated the figures from scratch, they certainly are not original. The Authors’ Guide makes the following point:

You must have permission to use any material from a copyrighted source. Redrawing an illustration is not enough—even if you use someone’s illustration only as a basis for your own, you must obtain permission to make a new version.

It’s true that Marc Green, unlike all the other main antecedents, was partly acknowledged. But even here, we have shown two passages and three figures that were not credited. And some of the acknowledged sections hew uncomfortably close to the original and are derivative in the extreme. Indeed, one is almost tempted to say that Green should have been a co-author, as he appears to have unwittingly contributed as much material as either of the putative authors.

In all, no less than twelve unattributed antecedents, including five Wikipedia articles, have been identified (see also Part 1 and Dubious Scholarship in Full Colour full analysis for further details).

Year	Source (with hyperlink)	Wegman & Said (2011)
1999	Color Theory Page (Ted Park)	p. 1, p. 2, p.3
1999	Kodak Digital Color Tutorial, Chapter 2, Digital Color Theory	p.2, p.3, p.4, p.5, p. 7
2000, 2004	Marc Green Basic Color & Design SBFAQ	p. 1, 14 (figures), 15
1999	Markemson, 1999	p. 7
2001-11	Wikipedia – Eye	p.1 (text & figure), p. 2 (text),
2002-11	Wikipedia  – Munsell Color System	p. 4 (text & figure)
2006-11	Wikipedia  – 1931 CIE Color Space	p. 4-6 (text & figures)
2005?	Wikia – Computer graphics – Saturation	p. 6-7 (text & equations)
2004	AllExperts: Genetics/colour blindness and albinism	p.9 (text)
2004-11	Wikipedia – RYB Color Model	p. 13-14 (paraphrased text, figures)
2005-10	Wikipedia – Normalized Color Co-ordinates	p. 14 (figure)
	Sputnik Color Schemes	p. 14 (figure)

What’s next?

Clearly, there are deep problems at WIREs CS that go well beyond one problematic article. If this is the example that two of the three editors are setting, then the entire project is obviously in need of a complete review. Surely even a minimum of  editorial control and competent peer review should have spotted glaring red flags in this article.

And it should be noted that this is not the only dubious work published here. The other Overview article by Said and Wegman, Road Map for Optimization, contains a scant seven references (including both the original and republished versions of Bellman’s Dynamic Programming). And a first look appears to reveal, once again, both derivative text and glaring errors.

To be sure, it is unlikely that the abysmal scholarship exhibited by Wegman and Said has infected the rest of the journal roster. But a glance at the list of authors does show a remarkably high preponderance of Wegman’s social network of students. And the list of published and proposed articles is an odd mix of obscure minutae and overly broad excursions into broad swathes of computer science. Thus, a detailed review of the entire WIREs CS project would appear to be warranted, including the projected Wiley Blackwell-Encyclopedia of Computational Statistics to be released in 2012.

But all that will have to wait. For it turns out there is yet another mother lode of questionable scholarship from Wegman and Said – one that falls in  their putative areas of expertise, and dates back to 2005, well before they became  embroiled in the “hockey stick” controversy.

And, hopefully sooner rather than later, there will undoubtedly be developments in the mess that started this all off, namely the Wegman report and its followup Said, Wegman et al 2008 in Computational Statistics and Data Analysis, if emerging credible reports of a pending retraction of the latter are anything to go by.

To quote my compatriot Randy Bachman, “You ain’t seen nothing yet”.

===============================================

Main references:

• Edward Wegman and Yasmin Said, “Color Theory and Design”, Wiley Interdisciplinary Reviews: Computational Statistics, Volume 3, Issue 2, pages 104–118, March/April 2011. Online Feb. 4, 2011.
• Deep Climate, Dubious Scholarship in Full Colour: Antecedents of Wegman & Said (2011) and Wegman (2002), March, 2011 [PDF]