[Koha] Koha license upgrade ballot design

[Thomas Dukleth]

Reed Wade has rightly recognised that discussion about voting methods at
least partly involves some presumptions ballot design which may not
necessarily be warranted.

Reed suggested a particular ballot design which I find to be mistaken,
however, his thought of conceiving of the ballot design as multiple binary
questions for approval or disapproval is worth considering.  Maybe the
question of license upgrade is amenable to binary questions ballot design
which people would not find too tedious and difficult.  We should also
consider the issue of whether such a binary questions ballot design might
unfairly favour some options over others.


1.  BINARY REDUCTION.

Most everything could be reduced to a set of binary possibilities.  A
ballot question with a few options could be expressed as multiple ballot
questions for approval or disapproval, yes or no.

Enumerating all the possibilities needed for proper binary reduction is
often tedious and difficult to follow.  An important reason we humans
prefer high level languages to low level binary communication is that
enumerating our communication in a low level binary language would be
difficult.

We have been discussing voting methods have been devised to allow complex
expression of preferences while avoiding excessively long ballots. 
Dividing questions into multiple binary questions, in which every option
of the undivided question is presented in comparison with every other
option individually for approval or disapproval can lead to excessively
long ballots.  We have also been discussing the possibility of expressing
the strength of approval or disapproval within a range indicating
strength.

Enumerating all the possibilities needed for proper binary reduction is
often tedious and difficult to follow but not always.


[Remainder of reply inline:]


2.  APPLICATION BEŸOND THE IMMEDIATE QUESTION.

Original Subject:  Re: [Koha] Koha license upgrade voting method

On Mon, January 10, 2011 01:15, Reed Wade wrote:


> Feels like we're thinking way too hard about this. It might simplify
> things to look directly at our situation instead of getting too
> clever.

Whether the issue is about what ballot design or what voting method best
maximises the preferences of voters is an important one which goes well
beyond this particular case.  Reaching some helpful conclusions about both
ballot design and voting methods will be helpful for guiding the
procedures for any future votes in which the voting method may effect the
result.

On the issue of voting methods, MJ Ray rightly identifies that even when
votes cast appear to be overwhelming in favour of one particular ballot
option, we do not know what votes would have been cast under a different
voting method.  A similar case could be made about ballot design.

We may have a significant period of time to consider ballot design and
voting methods.  I hope that people at Georgia Public Library Service
(GPLS) will quickly decide to update the OpenNCIP copyright license
invocation statement from GPL 2 to GPL 2, with an or later version option,
consistent with Georgia PINES and Evergreen license invocations.  The
inconsistency seems to be merely an obvious mistake.  However, GPLS may
take the long period to time which library organisations too often take to
come to a decision despite the apparently obvious mistake and the
simplicity of correcting it.  Dropping widely used OpenNCIP support is not
an option.  I do not know of any volunteers to completely rewrite that
code.  Waiting a reasonable period for GPLS to resolve the OpenNCIP
license invocation issue seems the best course.  While the OpenNCIP
copyright license invocation issue is unresolved, holding a vote on 
upgrading the copyright license for Koha would be moot and leave more time
to consider how best to conduct such a vote.


3.  AMBIGUOUS MISTAKEN BINARY QUESTION BALLOT DESIGN.

>
> We're at GPL2. Options are:
>
>   - stay there
>   - move to GPL3
>   - move to AGPL3
>   - some other scenarios that I don't think have been proposed by
> anyone other than as a "we could..."
>
> From what I can tell, moving to GPL3 is not contentious.
>
> From what I can tell, moving to AGPL3 is mostly not contentious but
> has some strong minority opposition.
>
> So, would a meaningful vote be:
>
>   - yes/no I approve of moving to GPL3
>   - yes/no I approve of moving to AGPL3

Despite Reed's clever thinking about an alternative, he has skipped
several steps.

If a ballot with approval options expressed independently from one another
would be constructed as described by Reed, what would the following
overall result mean?

I approve of upgrading the Koha copyright license to:

GPL 3, invoked with an or later version option.  Yes wins.

AGPL 3, invoked with an or later version option.  Yes wins.

Would we have a disjunctive licensing for Koha such that Koha would be
available as both GPL 3 with an or later version option, and AGPL 3, with
an or later version option.  There had been an earlier thread with a
subject heading which mistakenly implied such a possibility as meaningful.
 Using both GPL 3 and AGPL 3 for the same Koha code designed to be used
over a remote network connection would nullify the effectiveness of the
only difference between AGPL 3 and GPL 3 which is over network services. 
If both GPL 3 and AGPL 3 licenses would be available for Koha, then
everyone would be free to ignore AGPL 3 and treat the license as GPL 3.


4.  UNAMBIGUOUS BINARY QUESTION BALLOT DESIGN.

The binary reduction principle could be applied to reduce an appropriate
set of options to be considered to a binary form.

I start with the presumption that the question with a few options which
would need to be divided for a binary ballot would be something like the
following.

What should been done about the Koha copyright license?

Retain GPL 2, invoked with an or later version option.
Upgrade to GPL 3, invoked with an or later version option.
Upgrade to AGPL 3, invoked with an or later version option.
Postpone for further discussion.

The following questions could have ranges added to express strength of
approval or disapproval but that is a somewhat different discussion about
voting methods.

4.1.  POSSIBLE UNAMBIGUOUS BALLOT.

Should the question of upgrading the copyright license for Koha be voted
upon now or postponed for further discussion?  Choose one option.

Vote now.
Postpone for further discussion.

Should Koha retain GPL 2, invoked with an or later version option, or
should the copyright license be upgraded?  Choose one option.

Retain GPL 2, invoked with an or later version option.
Upgrade the license.

If the Koha copyright license is upgraded which upgrade choice should be
taken?  Choose one option.

Upgrade to GPL 3, invoked with an or later version option.
Upgrade to AGPL 3, invoked with an or later version option.


4.2.  CONSIDERATION OF POSSIBLE UNAMBIGUOUS BALLOT.

Is a ballot design with one question and four options divided into three
binary questions sufficiently easy to follow?

The logical relations between options in this case allows avoiding a
degree of exhaustive enumeration with questions comparing one option to
another individually until all possible comparisons are given.

Do the logical relations which allow avoiding exhaustive enumeration or
the degree of enumeration necessary bias the ballot in favour or against a
particular result?

Perhaps the degree of enumeration needed makes the issue seem unduly
complex for which postponement for further discussion may be favoured.

Perhaps choosing options in which later questions would not be applicable
would be favoured.  Alternatively, perhaps choosing options in which all
questions would be applicable would be favoured.

[...]


Thomas Dukleth
Agogme
109 E 9th Street, 3D
New York, NY  10003
USA
http://www.agogme.com
+1 212-674-3783