GPL Revisited (Asbestos underwear activated)

Laura Creighton lac at cd.chalmers.se
Thu May 10 15:31:10 CEST 2001


(Chris Gonnerman) writes:
>One thing I find strange is that there seem to be many who are adamantly pro
>GPL, and many who are adamantly anti GPL, but there seem to be few like me
>who have no strong opinions either way.

This is not strange; it is sampling error, and you have to take it into
consideration at all times no matter what you do or you will really screw
up your life.  Unfortunately, most people do not know about it, and it
is one of the big reasons why the world is as screwed up as it is.

This is the difference between probability sampling and
non-probability (convenience) sampling.  With probability sampling,
all elements in the population have some opportunity of being included
in the sample, and the mathematical probability that any one of them
will be selected can be calculated. With nonprobability sampling, in
contrast, population elements are selected on the basis of their
availability (because they volunteered) or because of the researcher's
personal belief that they would be cool to have in the study.  (Which
is generally phrased in a more learned-sounding way, but which
amounts to just that.)  The consequence is that an unknown portion of
the population is excluded.  The extent to which a convenience sample
- regardless of its size - actually represents the entire population
cannot be known.  EVER.  And no amount of statistical fiddling with
your data will EVER fix the problem.  So you must NEVER EVER use 
convenience sampling if you want to make conclusions about the population
as a whole.

A great deal is known already about what happens when you let a
set select themselves.  People with strong opinions care enough to be
counted.  In general, people with negative opionions care more than
those with positive opinions -- which is why if your political party
is in power, you want Sunny weather on election day, but if the
country is being run by your opponents you want a Snowstorm.  (Parties
in power should not call elections in the middle of the summer when
everybody is at their summer house as well).

Guido is a beloved dictator because he *is* aware of the sampling
problem, though here we have more people who want to *change* python,
rather than just _hate_ it.  A smaller number of posts come from
people who do not want changes, just to remind Guido that there are
people who do not want change.  But nobody is making c.l.p postings
saying hmm, I just read PEP XXX and I don't care if it goes into the
language.  If you do not care you do not say anything.  Also, a vast
number of people are way too busy writing code to be available for any
sampling in c.l.p.  They matter as well, but how can we ever sample
them?  All we can do is guess, and if it turns out we guessed wrong
try to make that easy to find out.  After 2.0 came out, people returned to
c.l.p to say, more or less, ``Sorry if this has been talked to death,
I haven't been reading c.l.p for a while, but ... this new string
syntax?  I think it really sucks.''  Enough to decide it was a
mistake?   To rip it out? Glad it is not my problem.

-------- 

So back to your original question.  Unless you have been trying to do
some probabibility sampling, what you are seeing is sampling error.
(If you _have_ been trying to get a representative sample, then I
would say there is something severely wrong with your scientific
procedure; try standing someplace other than outside the electrical
engineering building <wink> before you start sampling people at
random; your sample set is wildy non-representative.)  This error
masks a different question, the question of why do the people who care
about the GPL care so much about it, which I think is the question you
are more interested in.  This is a subset of the more difficult
question ``Why do people care about the things that they care about?
Why do we love some things and hate other things?''  You need a
biologist and a psychologist and I don't know what else to go after
that question -- it is an inter-disciplinary problem, and a tough one.

If you decide to interview GPL lovers and haters, trying to find out
why they care, could you please do it some place other than c.l.p?  I
care __far__ more about _discussing_ the GPL than I do _about_ the
GPL, to state for the record: I have had enough GPL discussions to
last me at least 10 lifetimes and I never want to hear anything about
it again.

Laura







More information about the Python-list mailing list