[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Poll results: User views on the FDL issue

Brian Nelson wrote:
> I recently conducted a poll on debian-user to get some input from users
> about the FDL issue.  The results are available here:
>   http://people.debian.org/~pyro/fdl_poll_results.txt

As some of you may know, I try to promote good practice in
surveys and consultations. Sometimes I even get so annoyed that
I flame people for bad surveys (sorry!). I would like to see
more good surveys done and better use made by debian of the
statistics that can be gathered about a large project.

Firstly, I thank Brian Nelson for running this survey. Most
of this looks to have been well-run mechanically. I feel the
analysis is incomplete: there is a problem with one question and
Brian over-emphasises one other result.  In future, it would
be good to see the methods used explained more fully, so that
others can try to replicate the results and comment on specifics.

What is the current number of readers of debian-user? I estimate
over 600 posting authors so far this month, which I guess is
only a fraction. So, I think 28 responses is a bit below what
I'd expect, if indeed this is a major issue for debian users.
Are there previous debian-user polls to compare with?

The preamble of the survey was terrible, in my opinion. I found
it in the mail log, but not in the results posted here or on
the poll results text file. I consider it incorrect on several
key points and it states a lot of the surveyor's beliefs. For a
self-selecting listener opinion poll (SLOP), this seems likely
to influence which listeners respond. Additionally, it should
have contained some details about what happens to responses,
especially how and when the results would be published.

I agree that question 1 gives a clear answer. I don't think
that the question is completely clear, but there's no sign of
widespread misunderstanding in the mail log. I hope it gives
some incentive for DDs to maintain FDL'd works in non-free
until they are replaced or relicensed.

> 2. Is it acceptable if all FDL documents were moved to non-free so that
> they could still be downloaded from Debian mirrors?  Note that they
> still would not appear on most Debian CDs.
> 28 responses
> Yes:        16  57.1%
> No:         11  39.3%
> Don't know:  1   3.6%
> My comment: Having documentation still available in non-free is OK with
> the majority of the respondents, though many stipulated that it was only
> acceptable as a last resort (i.e. they preferred keeping it all in
> main).

This is not significant. The sample is small, so please be
careful not to interpret a small majority of 27 responses as
being a strong signal that we should use to decide things.

[Let's assume that it's a 50-50 split among those who do
answer and we want less than a 1 in 10 chance of rejecting
that hypothesis erroneously if true: approximating Bin(27,0.5)
with Normal(0.5,0.25) suggests that P(X>0.571) = P(Z>0.142) =
0.4435, a long way above 0.1 - better tests can be used, but
I don't think any reject 50-50 split, based on this evidence.]

I agree that questions 3 and 4 give no clear signal.

> 5. Is Debian making a big deal out of a minor issue?  Should Debian be
> focusing more on creating a usable operating system and less on worrying
> about licensing issues?

This question is very troublesome. It's actually two different
questions and rather biased wording. The original post of it
was missing a word, too. I don't think it's safe to give much
weight to the answers. If a respondant read this question
before answering any (as they can with an emailed survey),
it may have biased all their answers even further. :-(


Reply to: