NRLgate -
Plagiarism by Peer Reviewers
Sections 4.9 thru 4.10
This page is part of the NRLgate Web site presenting evidence of
plagiarism among scientific peer reviewers involving 9 different peer review
documents of 4 different journal and conference papers in the fields of
evolutionary computation and machine learning.
This page contains sections 4.9 through 4.10 of "Evidence of plagiarism
in reviews #1, #2, and #3 of a paper on electrical circuit design submitted
to the Evolutionary Computation journal."
Go to top of NRlgate Home Page
Go to Abbreviated Table of Contents
Go to Detailed Table of Contents
Go to Complaint Letter
to the Evolutionary Computation journal
Go to top of Previous Page
Go to top of Next Page
4.9. The non-automatically-created time-and-date lines points to an
unusual sequence of events in creating these 3 review documents
Ordinarily, the precise time-of-day on which a peer reviewer happened to
dispatch his review by e-mail to a journal would not be noteworthy or remarkable
in any way.
Before examining the issue of the time-and-date lines, let me review the
mechanics of the transmission of reviews #1, #2, and #3 to me by the Evolutionary
Computation journal . The journal's first "call for papers"was
publicly issued in late July 1992. On September 23, 1992, I (and two co-authors,
Dr. Martin Keane and James P. Rice) submitted a paper entitled, "Use
of Automatic Function Definition in Genetic Programming to Facilitate Finding
an Impulse Response Function." The submitted paper presented a way
of enhancing the then-recently-developed technique of genetic programming
(GP) with the then-very-new enhancement of automatically defined functions
(ADFs) and applying these techniques to electrical circuit design.
On November 13, 1992, I received the following e-mail message from the North
American Associate Editor John Grefenstette.
- Date: Fri, 13 Nov 92 14:56:10 EST
- From: gref@AIC.NRL.Navy.Mil
- To: koza@cs.stanford.edu
- Subject: EVOLUTIONARY COMPUTATION manuscript
- Cc: dejong@AIC.NRL.Navy.Mil
- Dear John:
- I have obtained three reviews for your manuscript:
- Title: Use of Automatic Function Definition in Genetic Programming to
- Facilitate Finding an Impulse Response Function
- Manuscript Number: EC-JG-9209-0003
- which you submitted to EVOLUTIONARY COMPUTATION. As you can see from
- the detailed comments below, all the reviewers liked aspects of the
work
- and feel that the Genetic Programming paradigm is very interesting,
but
- they all agree that the current manuscript has significant shortcomings.
- Based on these reviews, I will accept the paper, subject to your
- revising it to address the concerns of the reviewers.
- Some of the major areas that require your attention are:
- 1. The intended focus and new contributions should be more clearly
- stated.
- 2. The motivation and significance of the control problem chosen for
the
- cas%g{Uereed to be addressed.
- 3. The evaluation of the method is incomplete. How does GP compare on
- this problem with, for example, random search, analytic methods, other
- heuristic methods, other genetic algorithms?
- 4. The limitations of the method and topics for further research need
to
- be addressed.
- 5. A more complete discussion of related work is needed.
- 6. The style of the presentation needs to be improved, as suggested
by
- the detailed comments below.
- I have enclosed the complete reviews. I'm sure you will find
these
- helpful in preparing a revised manuscript. I hope you can submit a
- revised version for inclusion in an early issue of the journal.
- John Grefenstette
- Associate Editor
- ^L
- ==============================================
- EVOLUTIONARY COMPUTATION
- Paper Review Form
- Title: Use of Automatic Function Definition in Genetic Programming
...
- Author(s): Koza, Keane, and Rice
- Date: 11/6/92
- Manuscript Number: EC-JG-9209-0003
- Reviewer: #1
(Emphasis added).
The time-stamp on this e-mail message indicates that this message was sent
during normal business hours on a business day
- Date: Fri, 13 Nov 92 14:56:10 EST
and that it originated,
- From: gref@AIC.NRL.Navy.Mil
The sending address belongs to Code 5514 at the "Artificial Intelligence
Center" of the "Naval Research Laboratory" in Washington,
DC of the "Department of the Navy" which is a branch of the "Military."
That is, this message originated from a government computer.
Grefenstette's message shows an electronic "carbon copy" to Editor-in-Chief
Kenneth DeJong of Code 5510 of the Naval Research Laboratory.
- Cc: dejong@AIC.NRL.Navy.Mil
After leaving the government computer at the government office in the District
of Columbia, the e-mail message traveled to me via the Internet (which was,
at the time, a government-subsidized communication network) to its destination
in the state of California.
The North American Associate Editor's e-mail message consisted of four parts.
The first part contained the North American Associate Editor's summarization
of the reviews written by the 3 peer reviewers, his overall decision and
instructions to the authors, and was signed by him. This next part of the
message was separated from the remainder of the message by a dividing line
consisting of 46 ='s. The North American Associate Editor's message then
consisted of a heading and incorporated the "complete" written
by reviewer #1. Then, there was a second dividing line consisting of 46
='s followed by another heading and the text of the review written by reviewer
#2. Finally, there was a third dividing line consisting of 46 ='s, a similar
heading, and the review written by reviewer #3.
Returning now to the header of the overall e-mail message that I received
from North American Associate Editor Grefenstette:
- Date: Fri, 13 Nov 92 14:56:10 EST
- From: gref@AIC.NRL.Navy.Mil
- To: koza@cs.stanford.edu
- Subject: EVOLUTIONARY COMPUTATION manuscript
- Cc: dejong@AIC.NRL.Navy.Mil
This header is typical of the hundreds of e-mail messages that I receive
each week from all over the world and contains information common to all
e-mail messages transmitted over the Internet. Notice, in particular, the
specific stylized format of the "date plus time-of-day
"
line:
- Date: Fri, 13 Nov 92 14:56:10 EST
Peer reviewing at the Evolutionary Computation journal is anonymous
in the sense that authors of submitted papers are given the texts of the
reviews written by the peer reviewers, but not the names of the reviewers.
After the 3 reviewers finish writing their reviews, they transmit their
review to the journal (typically by e-mail over the Internet). Each of the
3 original peer review documents therefore has an e-mail header containing
the familiar information contained in the header shown above.
In order to prevent identification of the 3 reviewers to the submitting
author, the North American Associate Editor necessarily would delete the
"from
" line of each of the three e-mail messages
that he received from the 3 peer reviewers when he incorporated the texts
of their reviews into the his e-mail message to the submitting authors.
Instead of the 3 "from
" lines, we instead see the
3 uninformative phrases "reviewer #1," "reviewer #2,"
and "reviewer #3." And, presumably at the same time, the North
American Associate Editor would have deleted the uninformative "to
"
line deleted (since all 3 peer reviewers presumably would have sent their
reviews to the journal).
4.9.1 The absence of the time-of-day indicates that the 3 "date"
lines that the submitting authors received from the North American Associate
Editor were not the familiar "date plus time-of-day" lines that
are automatically created when an e-mail message is transmitted on the Internet.
Ordinarily, the precise time-of-day on which a peer reviewer happened to
dispatch his review by e-mail would not be noteworthy or remarkable in any
way.
The "date
" lines for the 3 peer reviews attached
to the North American Associate Editor's November 13, 1992 e-mail message
to the submitting authors are as follows (with spacing intact) for reviewers
#1, #2, and #3 (in that order):
- Date: 11/6/92
- Date: 21 Oct 92
- Date: Nov. 11, 1992
Notice the absence of the time-of-day. Ordinarily, the time-of-day (right
down to the second) is an automatically created timestamp that appears on
every e-mail message handled by the Internet.
The absence of the time-of-day indicates that the 3 "date
"
lines that the submitting authors received from the North American Associate
Editor were not the familiar automatically created "date
plus time-of-day
" lines that are routinely created when an e-mail
message is transmitted on the Internet.
4.9.2 The unusual format of the 3 "date" lines were not the
familiar "date plus time-of-day" lines that are automatically
created when an e-mail message is transmitted on the Internet
Ordinarily, the precise date on which a reviewer happened to e-mail his
review in to the North American Associate Editor would not be noteworthy
or remarkable.
Note that none of the 3 dates appear to be in the familiar automatically
created format. Note, in particular, that there are 11 spaces preceding
the hand-created date line for reviewer #2 and there is a period after "Nov."
for reviewer #3.
Whatever the three "date
" lines on these 3 review
documents may be, they are not the untouched, virginal, automatically created
"date and time-of-day
" lines that are automatically
created when electronic mail is ordinarily transmitted via the Internet.
Instead, they are hand-created dates.
One cannot determine the exact detailed sequence of events that took place
to create each of the four parts of North American Associate Editor's Grefenstette's
e-mail message merely by looking at it in a vacuum and without additional
evidence. However, one can consider the possibilities.
Why would the North American Associate Editor send the submitting authors
the "complete reviews" of the 3 peer review messages intact ---
but not send the original, virginal, intact "date plus time-of-day
"
lines that were effortlessly available?
One possible explanation is that the original, virginal, intact "date
plus time-of-day
" lines of these 3 "independent" peer
documents were unfit for the eyes of the submitting authors --- perhaps
because they were embarrassingly contemporaneous with the North American
Associate Editor's own summarizing message to the submitting authors at
precisely
- Date: Fri, 13 Nov 92 14:56:10 EST
Another possibility is that one or more of these 3 "independent"
peer documents was not, in fact, ever "received" by e-mail from
the outside from any of the 32 geographically dispersed members of
the editorial board of the journal (none of whom, it will be recalled, are
co-located in Washington, DC with the Editor-In-Chief and North American
Associate Editor at the Naval Research Laboratory). That is, perhaps one
or more of these 3 "independent" peer documents never arrived
from outside of the NRL office and therefore never originally bore any automatically
created "date plus time-of-day
" line of its own.
This could happen if these documents came from a computerized file residing
within the very same NRL computer workstation that dispatched the e-mail
message to the submitting authors (e.g., the North American Associate Editor's
own workstation). To be specific: a piece of text that was scooped from
one file of a computer to the e-mailer of the very same computer would not
bear any "date plus time-of-day
" line of its own.
Similarly, a piece of text that was scooped by a direct file access (e.g.,
via a local area network) from another computer located under the same roof
would not bear any "date plus time-of-day
" line of
its own. If one or more of these 3 "independent" peer documents
did not bear a "date plus time-of-day
" line of its
own, that would explain why some kind of "date
" line
would have to be added.
These 3 "date
" lines were not in any standard automatically
created format. The question arises as to why --- in typing these 3 "date"
lines (possibly typed by one hand in rapid succession) were 3 different
styles employed --- namely
- Date: 11/6/92
- Date: 21 Oct 92
- Date: Nov. 11, 1992
There are 11 spaces in the second date and there is a period after "Nov"
in the third date.
Perhaps the typing of these 3 dates, in rapid succession, in 3 different
non-standard styles was part of an effort to make them "look different."
4.10. The 4 appearances of the submitted paper's title points to an
unusual sequence of events in creating the review documents. The 2-line
title line of review #3 is an exact "cut-and-paste" copy of the
two-line title in the section written by the North American Associate Editor,
but the abbreviated 1-line title lines of reviews #1 and #2 were not.
The title of the submitted paper is contained in the introductory portion
of North American Associate Editor Grefenstette's e-mail message and it
is also contained in each of the 3 attached reviews.
It would be understandable if
- all 4 of these titles were entirely different (presumably because they
were typed separately by the 3 reviewers and the North American Associate
Editor), or
- all 4 were identical (presumably because they were part of the paper
review form disseminated by the journal).
But the documentary evidence shows that neither of these situations
prevailed.
The introductory portion of the e-mail message from North American Associate
Editor Grefenstette is
- Dear John:
- I have obtained three reviews for your manuscript:
- Title: Use of Automatic Function Definition in Genetic Programming to
$
- Facilitate Finding an Impulse Response Function
The title line of review #3 is
- Title: Use of Automatic Function Definition in Genetic Programming
to $
- Facilitate Finding an Impulse Response Function
The title line of review #2 is
- Title: Use of automatic function definition in Genetic Programming$
The title line of review #1 is
- Title: Use of Automatic Function Definition in Genetic Programming
...$
Note that we indicate the carriage return symbol by a "$".
First consider the two-line titles appearing in North American Associate
Editor in Grefenstette's introduction and review #3. Notice
that both of these two-line titles consist of the complete 15-word title
of the paper and that exactly the same 11 words are capitalized. Now notice
the 8 spaces before the word "Facilitate" on the second lines
of these two titles. Notice the single blank space after the "to"
on the first line of both titles. Notice the carriage return symbol (made
visible here by a "$") after the "to
"
on the first line of both titles. In other words, both of these two-line
titles are identical (character-for-character and space-by-space).
In most computerized word processor systems, the computer software automatically
drops down to the next line based on the preestablished width of the paper
(or computer screen) and then continues the ongoing sentence on the new
line. The person doing the typing does not normally control this breakage.
Instead, the typist normally only specifically marks desired paragraph breaks
(usually by hitting the "carriage return" key). Of course, if
the typist wants to force the creation of a new line, he may type the "carriage
return" key.
The 8 hand-introduced spaces at the beginning of the second lines of these
two titles are apparently an attempt to make the masthead look more attractive
by indenting "Facilitate" (the 10th word of the title) in one
space with respect to "Use" (the first word of the title). To
accomplish this indenting, a space and a carriage return were added after
the "to" on the first line and then 8 spaces were introduced,
by hand, at the beginning of the second line.
Now consider the one-line titles of review #1 and review
#2. Notice that both of these titles are abbreviated to the first 8
words.
The one thing that is certain is that two-line title of review #3 ---
with its 8 hand-introduced extra spaces before "Facilitate" ---
is an exact "cut-and-paste" copy of the two-line title in the
introductory section of North American Associate Editor Grefenstette's e-mail
message to the submitting authors --- whereas the abbreviated one-line titles
of reviews #1 and #2 were not.
That is, Grefenstette's introduction and review #3 contains common "cut-and-pasted"
material.
It would be understandable if all four of these titles were identical and
it would be understandable if all four were entirely different. However,
it is difficult to visualize any consistent process that could have given
rise to the four titles that we see. Apparently whatever sequence of steps
gave rise to Grefenstette's introduction and review #3 was very different
from the sequence of steps that gave rise to reviews #1 and #2.
When North American Associate Editor Grefenstette's began his e-mail message,
- I have obtained three reviews for your manuscript...
(Emphasis added)
did we have a specific reason to carefully choose the unusual word "obtained,"
instead of the more commonplace "received"? Is it possible that
one or more of these review documents were, in fact, never "received"
from anywhere?
One cannot determine the exact detailed sequence of events that took place
to create each of the four parts of North American Associate Editor Grefenstette's
e-mail message merely by gazing at this message in a vacuum. Additional
factual evidence is necessary to establish the precise explanation for this
unusual outcome.
Author: John R. Koza
E-Mail: NRLgate@cris.com
Go to top of NRlgate Home Page
Go to Abbreviated Table of Contents
Go to Detailed Table of Contents
Go to Complaint Letter
to the Evolutionary Computation journal
Go to top of Previous Page
Go to top of Next Page