From Newsgroup: sci.lang

Recently, I read about

"When Behavioral Data Betray Users: A Diagnostic and
Protective Framework Against Social Interaction Leakages"
(2026) - Yan Leng et al.

. The report said, readers can identify who your friends are on
the fact that friend often write reviews of the same length.
Well, I'm not sure how this should work when there are thousands
of users, may be there is something I miss.

However, I have checked some newsgroups using linguistic means
to find whether any poster uses "sock puppets" and I found
that posters can be linked strongly to the average length
of their posts and the average percentage of material quoted.
When some poster uses slight variations of his name, such as
"John Doe" and "JohnDoe", which are not intended to hide the
face that these two names belong to the same person, these two
often appear close together when sorted by these two values!

I also added another layer taking preferred and avoided words
and phrases into account.

Using this means, I found that my system often correctly
groups slight variations such as "John Doe" and "JohnDoe"
together (without taking the similarity of the names into
account), which confirms that it is correctly identifying
a kind of significant linguistic fingerprint.

On the other hand I found that regulars posting under different
names trying to hide that this is the same person ("sock
puppets") are very rare. Usually, regular posters post under
one single name and not more. However, my system might have
missed one-shot fakes because it only can identify someone
linguistically when there is enough text written by him.


--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: sci.lang

ram@zedat.fu-berlin.de (Stefan Ram) wrote or quoted:

When some poster uses slight variations of his name, such as
"John Doe" and "JohnDoe", which are not intended to hide the

I just observed a real example of this. A man posting as

Johann 'Myrkraverk' Oskarsson

and then later as

Johann \"Myrkraverk\" Oskarsson

in comp.lang.misc. This is very probably the same man /not/ in-
tending to hide the fact that both names refer to the same man.


--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: sci.lang


ram@zedat.fu-berlin.de (Stefan Ram) posted:

. The report said, readers can identify who your friends are on
the fact that friend often write reviews of the same length.

"Were these poems/plays all written by the same man?" is a topic
that has been explored in several college classes. Often the focus
is specifically on the Shakespeare Authorship Controversy. It might
be interesting to comment on those efforts; unfortunately it was decades
ago that I read some of those papers and I won't try to hunt for links now.

Of course attention focused especially on Edward Earl Oxford, the best candidate for true Author.

To avoid comparing apples with oranges, one would want to compare plays
with plays or letters with letters. But difficult for Oxford-Shakespeare comparison:
-- Letter vs letters? ZERO letters by "Shakespeare" exist if he was
literate at all.
-- Plays vs plays? Although Oxford was certainly a renowned playwright,
zero of his play-names are known, let alone play-scripts. Excepting,
of course, plays he wrote under the pseudonym.
-- Poems vs poems? Many or most of the few poems definitely ascribed
to Oxford were written in his teens, while he was in his 40's when "Shakespeare"'s first poem was published.

Despite these difficulties (and despite that even Oxford's letters use hendiadys, and other techniques for which Shakespeare is famous)
some of those papers conclude that Oxford/Shakespeare identity is 0.0001% chance or so. *One professor whose class came up with near-impossibility* *confessed that he personally thinks Oxford was the Author!*

One student reported a 12-sigma deviation (or such) between the two authors! His metric was so ridiculously complicated I concluded he must have
explored and cherry-picked to maximize the sigmas.

Cheers,
James
--- Synchronet 3.22a-Linux NewsLink 1.2