Identifying Colonel Burke

[Tom Keske]

IDENTIFYING COLONEL BURKE

The usenet troll known as "Colonel Edmund J. Burke"
has made thousands of stomach-turning, obscene
posts to more than 100 newsgroups, many of them laced with
racial epithets and anti-gay epithets.

I previously noted that multiple people on different newsgroups
had identified "Colonel Edmund J. Burke" as this person:

Eric Paul Burke
42404 Waterwheel Ct. W.
Northville, MI 48323
248-349-1212

Eric Cell: 248-767-9359

If he is just an internet troll, it might not matter much,
but he has openly bragged that he is getting paid to
try to destroy newsgroups. If so, his activities as a troll
are not a test of the limits of free speech, but would represent
an assault on free speech.

His use of racial epithets and vulgar obscenities might be
a *technique* for the very purpose of trying to drive people
away.

While he has been identified several times, independently,
there has at the same time been doubt expressed,
because he uses faked email addresses.

Probably, Burke thinks he is being very clever, but he is
actually rather careless. He probably thinks that he can just
laugh it off when identified, that it will boil down to someone
else's word against his, and that he can stonewall, and deny it.

Eric Paul Burke majored in computer science at Purdue, but
apparently didn't take any courses in statistics.
He does not seem to realize what could give him away,
even with all those faked addresses.

Many of his email addresses contain "burke":

***@yahoo.com
***@yahoos.com
***@yahooo.com
***@sneakemail.com
***@badboys.com
***@yahoo.com
***@yahoo.com
***@yahoo.com
***@asianbabes.com

Some of them contain no name, at all:

***@sneakemail.com
borkatemysausage.com
***@norad.com
ik823r5?@sneakemail.com
***@caribbean.net

Probably, most people would know intuitively that he
is not picking the name "burke" at random, and don't need
that fact proven to them. In the sampling found above,
there are no "Johnsons" or "Smiths", etc.

However, it is good to try to put a number on it,
just to get a better sense.

To make a ballpark estimate, I looked at the Boston area
white pages, which has about 718 pages total, of which about
1 page is occupied by "Burke". If Burkes are spread relatively
evenly throughout the country, it would be fair to estimate
that the chance of a random name being "Burke" is about
1 in 700.

For the first 3 email addresses which I found to all
contain "burke", by random chance would be
(1/700) * (1/700) * (1/700) , or about 3 chances
in a billion.

I made that calculation before finding that there were
at least 9 different email variations with "burke",
not just three.

So the gut feel is correct. The odds of random chance are off
the charts.

If you Google "email trace online free" you will find
there are numerous applications for tracing email addresses.

Most of the ones that claim to be "free" will only
tell you a little, then you have to pay money to get
more information.

The troll "Colonel Edmund J. Burke" has bragged that some of his
addresses were real enough, but stolen, and that
others were completely fictitious.

Like a 5 year-old, Burke boasted that his identity was "top secret".

Most of the tracing applications will tell you if an
email address appears to be valid or not. Burke's boast
seemed correct, because some were completely bogus.

However, of the first four addresses that yielded a positive
identification, two were Eric Paul Burke of Northville, Michigan;
one was an Elizabeth Burke, a 74 year-old woman, and the 4th was
another "Eric Burke" of Minnesota.

The Colonel might think that this result keeps his
identity hidden, but it is the other way around.

If someone looked naively at the result, they might suppose
that the odds of "Eric Paul Burke" being the right person
would be only about 50-50.

To estimate the odds that a specific individual,
"Eric Paul Burke" of Northville, MI, should show up twice in a
sample of four, I counted about 20 "E Burke" names in the
metropolitan Boston area. That is only Boston area, not even
counting the rest of Massachusetts.

If Burkes are spread roughly evenly in the U.S, there
might be about 20 * 50 = 1000 "E Burkes" in the country.

This becomes a simple exercise in binomial probability.
What are the odds of getting two more successes
in 4 trials, with a chance for success of 1/1000
for any one trial?

I added binomial probability to my own personalized
calculator program, a long time ago, but you can find the
same by Googling for "binomial probability calculator".

The odds are about 6 in a million.

That is just for "E Burkes" in the United States.
Since the Colonel is stealing addresses on the Internet,
it would really include all of Canada, England, Ireland,
Australia, Europe, etc.

And that is when limiting only to "E" Burkes. As some
of his faked addresses indicate, he could really pick
from *any* Burke. I found an estimate on the
web that there are about 32,774 people named as BURKE
in the U.S.

In that case, the chance of finding Eric Paul Burke
of Northville, Michigan twice, in a sample of 4 trials
would be about 6 in billion.

So, the posters who opined that they were 99% certain
that Eric Paul Burke was not the right person
would appear to have at least a 99.9994% chance of
being mistaken in that conclusion.

I believe that any competent statistician would be
comfortable with the conclusion that the allegation
about Eric Paul Burke was correct, the first time.

------------------------------------------------------------------

[Colonel Edmund J. Burke]

Post by Tom Keske
IDENTIFYING COLONEL BURKE
The usenet troll known as "Colonel Edmund J. Burke"
has made thousands of stomach-turning, obscene
posts to more than 100 newsgroups, many of them laced with
racial epithets and anti-gay epithets.
I previously noted that multiple people on different newsgroups
Eric Paul Burke
42404 Waterwheel Ct. W.
Northville, MI 48323
248-349-1212
Eric Cell: 248-767-9359
If he is just an internet troll, it might not matter much,
but he has openly bragged that he is getting paid to
try to destroy newsgroups. If so, his activities as a troll
are not a test of the limits of free speech, but would represent
an assault on free speech.
His use of racial epithets and vulgar obscenities might be
a *technique* for the very purpose of trying to drive people
away.
While he has been identified several times, independently,
there has at the same time been doubt expressed,
because he uses faked email addresses.
Probably, Burke thinks he is being very clever, but he is
actually rather careless. He probably thinks that he can just
laugh it off when identified, that it will boil down to someone
else's word against his, and that he can stonewall, and deny it.
Eric Paul Burke majored in computer science at Purdue, but
apparently didn't take any courses in statistics.
He does not seem to realize what could give him away,
even with all those faked addresses.
borkatemysausage.com
Probably, most people would know intuitively that he
is not picking the name "burke" at random, and don't need
that fact proven to them. In the sampling found above,
there are no "Johnsons" or "Smiths", etc.
However, it is good to try to put a number on it,
just to get a better sense.
To make a ballpark estimate, I looked at the Boston area
white pages, which has about 718 pages total, of which about
1 page is occupied by "Burke". If Burkes are spread relatively
evenly throughout the country, it would be fair to estimate
that the chance of a random name being "Burke" is about
1 in 700.
For the first 3 email addresses which I found to all
contain "burke", by random chance would be
(1/700) * (1/700) * (1/700) , or about 3 chances
in a billion.
I made that calculation before finding that there were
at least 9 different email variations with "burke",
not just three.
So the gut feel is correct. The odds of random chance are off
the charts.
If you Google "email trace online free" you will find
there are numerous applications for tracing email addresses.
Most of the ones that claim to be "free" will only
tell you a little, then you have to pay money to get
more information.
The troll "Colonel Edmund J. Burke" has bragged that some of his
addresses were real enough, but stolen, and that
others were completely fictitious.
Like a 5 year-old, Burke boasted that his identity was "top secret".
Most of the tracing applications will tell you if an
email address appears to be valid or not. Burke's boast
seemed correct, because some were completely bogus.
However, of the first four addresses that yielded a positive
identification, two were Eric Paul Burke of Northville, Michigan;
one was an Elizabeth Burke, a 74 year-old woman, and the 4th was
another "Eric Burke" of Minnesota.
The Colonel might think that this result keeps his
identity hidden, but it is the other way around.
If someone looked naively at the result, they might suppose
that the odds of "Eric Paul Burke" being the right person
would be only about 50-50.
To estimate the odds that a specific individual,
"Eric Paul Burke" of Northville, MI, should show up twice in a
sample of four, I counted about 20 "E Burke" names in the
metropolitan Boston area. That is only Boston area, not even
counting the rest of Massachusetts.
If Burkes are spread roughly evenly in the U.S, there
might be about 20 * 50 = 1000 "E Burkes" in the country.
This becomes a simple exercise in binomial probability.
What are the odds of getting two more successes
in 4 trials, with a chance for success of 1/1000
for any one trial?
I added binomial probability to my own personalized
calculator program, a long time ago, but you can find the
same by Googling for "binomial probability calculator".
The odds are about 6 in a million.
That is just for "E Burkes" in the United States.
Since the Colonel is stealing addresses on the Internet,
it would really include all of Canada, England, Ireland,
Australia, Europe, etc.
And that is when limiting only to "E" Burkes. As some
of his faked addresses indicate, he could really pick
from *any* Burke. I found an estimate on the
web that there are about 32,774 people named as BURKE
in the U.S.
In that case, the chance of finding Eric Paul Burke
of Northville, Michigan twice, in a sample of 4 trials
would be about 6 in billion.
So, the posters who opined that they were 99% certain
that Eric Paul Burke was not the right person
would appear to have at least a 99.9994% chance of
being mistaken in that conclusion.
I believe that any competent statistician would be
comfortable with the conclusion that the allegation
about Eric Paul Burke was correct, the first time.
------------------------------------------------------------------

LOL