Re: tlug: TLUG S/N - an empirical analysis

[Thomas Joseph Haslam <tjhaslam@example.com>]

Reply to Professor Turnbull 
(and "noise" that can be ignored by most everyone else):

I enjoyed the gif but you haven't earned it yet.  The "signal to noise
ratio" comment that you *undertook* to empirically analyze was made in
regard to the TurboLinux "Advertising" tiff (an issue that most TLUG
members seem to regard as a non-issue, the comments about monkey mounting
aside). 

With that remark, which was probably meant more metaphorically than
literally, the particular TLUG member in question was expressing his
general satisfaction with the TLUG postings.  That is the matter in
question: member participation and satisfaction, not Shannon content per se.  

And this indeed could be empirically studied, although methodology would
not be as elegant or as simple (in the best possible sense) as your use of
information theory (re: Shannon, Claude E. and Warren Weaver. 1949.  The
Mathematical Theory of Communication. Urbana: University of Illinois Press).

Based upon my informal reading of the TLUG postings--I have NOT and will
NOT undertake the full empirical study required, here are some reasons why
TLUG has a perceived "high signal to noise ratio."

1.  Queries for information usually get answered within 24 hours--sometimes
even within 2 hours.
1.a.  The queries typically receive multiple answers--and these answers in
turn are generally perceived as both informed and useful.
1.b.  The person who originally posted the query often--although not
always--reports that his or her dilemma has been solved (or at least
alleviated) because of the information that he or she received from TLUG.

2.  TLUG postings cover a considerable range of Linux-related issues.
2.a.  Although there is a TLUG/TurboLinux connection (largely it seems to
me to the benefit of TLUG), users of any/all distributions of Linux receive
without prejudice advice and support from TLUG.
2.b.  Although there is a considerable SPARC contingency, users on any/all
platforms receive without prejudice advice and support from TLUG.
2.c.  Although TLUG has members with considerable (and perhaps even
specific) expertise (Chris Seikya--Sparc; Scott Stone--Turbolinux; et
cetera), Linux users of all levels are welcome to and do contribute to TLUG.

3.  TLUG postings that are not Linux or even computer-related generally
have some redeeming entertainment value.  

End reasons.  Here are two things might skew your results.

1.  The multiple and often similar answers to a common query.  
REMARK: Although the information is redundant--at least as you're measuring
it, the redundancy serves a valuable purpose.  Four radically different
answers might have a higher Shannon content.  But if three of those answers
are wrong or not particularly useful, then they certainly qualify as
"noise": and time- and resource- wasting "noise" at that.  The redundancy,
on the other hand, suggests that the proffered information is most likely
correct--or at least the best way to proceed, given the current knowledge.

2.  The somewhat extensive TLUG signature that ends every TLUG posting and
whose redundancy might lower the Shannon content.  REMARK: Does serve a
purpose--and did you account for it?


End that.

Based upon my reading of the TLUG postings (I should say my interpretation,
not "analysis"), I would deem TLUG very vibrant and generally successful,
and I would understand its members as generally satisfied: hence "high
signal to noise."  I would regard your *empirical analysis* as missing the
question--and hence as more an intellectual curiosity than an actual report
on conditions.  

But I write the above for your entertainment and amusement. Cheers, and

Best to all with all,
TJH




--------------------------------------------------------------
Next Meeting: 8 August, Tokyo Station Yaesu central gate 12:30
featuring Linux on multiple platforms:
i386, Sparc, PA-Risc, Amiga, SGI, Alpha, PalmPilot, ...
Next Nomikai: September, 19:30 Tengu TokyoEkiMae 03-3275-3691
--------------------------------------------------------------
Sponsor: PHT, makers of TurboLinux http://www.pht.co.jp