How did you recognize the
need for a copycat search
engine?

I'm finishing a PhD in
biophysics. Back in 1994
and 1995, I was TA'ing a
couple of large lecture
classes at Berkeley. I
decided to make a site
that would allow students
to read each others term
papers and grade them.
They wouldn't be graded on
the reviews they received,
of course, but when I had
been an undergraduate in
rhetoric, I'd always wanted
to read everybody else's
papers, which you don't
typically get to do. We
did it under the auspices
of a double blind peer
review, in which papers
were posted on the Web
with no name. All students
were then randomly
assigned to read and
review two papers. They
were graded on the
reviews, and the papers
were graded by me, who at
the time was making the
equivalent of a Taco Bell
salary. A couple weeks
later, a second wave of
students began essentially
turning in other students.
People would come up to me
and say, "This guy got a
paper off the Web site and
turned it in to this
class; that guy turned it
in to the other class." I
heard of one guy who was
actually selling these
papers, telling people,
"You want a term paper on,
say, Alzheimer's disease?
I can get it for you." So
the students were stealing
each others' intellectual
property, and I was their
enabler.

Around the same time,
these sites like
schoolsucks.com, Evil
House of Cheat, etc.,
started to come online.
And there was nothing
going on to stop the
problem of students
posting their work on the
Web and having it stolen.
At Berkeley, between 1993
and 1997, plagiarism
increased by 744 percent.

How are you measuring
that?

That's the Office of
Student Conduct's own
numbers.

This year, in one class in
one semester, we caught 45
students plagiarizing.
That's three short of the
entire number of students
that the entire University
caught in all of 1995.

I had a professor who
busted a plagiarist, went
through the school library
to find all the stolen
material, took the guy to
academic court, and really
nailed him. I mean, he was
really on a mission. Have
you ever prosecuted some
student to the fullest
extent of the law?

We're just the technology
that the professor can use
to make it easier to find
out if a work is original.
We make it clear in the
beginning that neither we
nor the computer can ever
decide whether a work is
plagiarized. I would also
say that this technology
is so new and places such
an unfair burden on the
students who are subjected
to it that you might want
to consider grandfathering
this group of students and
using it as a learning
tool rather than ruining a
student's career.

Oh, come on. You're just
making the lazy bastards
toe the line.

You know, you're right.
And I would completely
agree with you on that one
year from now, after
everybody knows the extent
of the technology. There's
no excuse after that. But
the reason we caught 45
students in that one class
and are catching 20 to 25
percent of all AP high
school students is that
this technology blindsided
those people.

Couldn't a suspicious
professor just reproduce
this system by entering a
phrase from a student
paper — "hard,
gem-like flame" or
something like that —
into a search engine?

When a teacher enters a
phrase into a search
engine like AltaVista, he
may get 1,000 hits. Plus,
any search engine covers
only a fraction of the
Web, and teachers don't
have that much time to
track this stuff down. So
there are flaws in the
search engine strategy:
the fact that nobody has
time and also — the
biggest problem of all —
that a student might do a
cut-and-paste job taking
1,000 words from one
paper, 1,000 words from
another, etc. All those
papers might be 10,000
words long, but the
plagiarized version might
come to 5,000 words, so
it's incredibly complex,
and who has time to track
all that down?

This teacher I mentioned
before sure did. He took a
whole weekend searching
the library, underlining
and starring in the
margins, matching it all
up against this dumb
schmuck's paper. He was
really on a tear.

That's somebody on a
mission!

Oh yeah, he was divinely
inspired to get this guy.
He was bragging about
how the dean tried to talk
him out of it: Nobody
wanted to bother with the
trial, it was the first time
this student had been
caught, etc. And this
teacher was fuming: "This
is just the first time he's
been caught! He's been
getting away with this his
whole career!"

Well, just to give you a
comparison, we could do in
a matter of minutes what
took him that whole
weekend. We're negotiating
with Net Library to add
their content to our
database, but even without
that, the most popular
works on a topic get cited
and end up in papers on
the Web anyway, so we can
still find them.

One of our writers once
made up an SAT analogy:
"The Web is to plagiarism
what crack is to violent,
glassy-eyed babbling." Do
you strongly agree,
somewhat agree, somewhat
disagree, or strongly
disagree?

I guess I'd have to agree
with that. It may be that
the Web is to plagiarism
what marijuana is to eating
Ding Dongs. I'd like one
that's a little less
violent.

How did you decide on
eight words as the right
size for a phrase when
you're searching?

Statistical analysis. The
bajillions of tests we did
showed that when you drop
it down to seven words,
you just get too much
noise. And the phrases
that we choose are long
enough that the
probability that two
papers will use the same
phrase is pretty low.

I used to find there was
no point in doing a good
term paper, because
teachers think any paper
that doesn't sound like
it's written by an idiot
must be a work of
plagiarism.

By being able to cite
sources in an unambiguous
way, we make it easier for
professors to follow up
this stuff. The professor
won't be in the position
of saying, "This sentence
doesn't sound right to
me," which is completely
unfair to students. Get
this: I have thank-you
letters from students who
were accused of plagiarism
and can now bring in their
papers and say,
"Plagiarism.org says it's
clean. If you can do
better, bring it to the
table."

Did you ever plagiarize
when you were in school?

Nope, I took the hit. I
took the spanking. It
pissed me off when I saw
people doing it. I saw a
lot of it as an undergrad,
and a whole lot of it as a
grad student. And I'm glad
to be able to do something
about it.

There's always that trick
of grabbing a whole batch
of text and then doing the
thesaurus function on it.
At what point does
covering your tracks start
to constitute original
work?

It's a very good question.
In the end it's the call
of the instructor. If it
were up to me, I would
want no more than 20
percent of it underlined
— and that part would
have to be cited. And no
more than 1 percent of the
rest underlined as noise.

I used to use the old
cheating-in-plain-cite
technique, where you
"quote" a lot of material
— set it off with
single spacing and
narrower margins, but have
that stuff constitute the
bulk of your paper. If you
were good you could have
75 percent of the paper
written in citations,
connected by
fancy-sounding
transitions. Is there a
substantial difference
between that and
plagiarism?

Two of those 45 students
we caught did the same
thing. That stuff was
properly cited, etc. But
if the bulk of your paper
would end up underlined by
our system, that's still
intellectually dishonest.

Of course. I knew that
when I did it. I just
hoped nobody would
notice.

Well, in both those cases,
the papers got good
grades, because it didn't
click in anybody's mind
that this was what was
going on.

Have you ever caught
somebody plagiarizing a
paper that you yourself
wrote as an
under-graduate?

No.

Do you believe the urban
legend that that has
happened?

Well, after we were
mentioned on World News
Tonight, a professor from
the University of
Heidelberg sent us a paper
that he had found in the
Journal of the Royal
Society of Surgeons in
Edinburgh, which contained
paragraphs taken from his
own Web site. The Journal
has had to retract that
and write an editorial on
it. They're really going
to get slammed. And these
are doctors! They operate
on people.

Senator Joseph Biden got
in some trouble for
plagiarizing in his
student days. Have you
ever gotten a request to
examine any legislation
that he's written, to see
if maybe it was originally
a law in some other
country?

That's exactly what I want
to do. I want to get every
politician in the United
States pissed off at me.