Re: Result for vote regarding new members for the board of directors

The voting period for the new board of director members ended yesterday
and the results are in.

A total of 136 people cast their vote, which is 55% of all eligible
voters. The three winners are:

Bruce Perens
John Goerzen
Benj. Mako Hill

The board of directors will vote on adding them to the board on its
next meeting.

The lists of voters, votes and results can be found online at the
following address:

http://www.spi-inc.org/secretary/votes/vote1/

Starting today there will be a seven day dispute-period during which
people can verify their votes using hash they received when they
voted. After that period the information that ties a hash to a
person will be removed from the voting database and the result will
be final.

Wichert.

--
Wichert Akkerman <wichert(at)wiggy(dot)net> http://www.wiggy.net/
A random hacker

Wichert Akkerman <wichert(at)wiggy(dot)net> writes:

> A total of 136 people cast their vote, which is 55% of all eligible
> voters. The three winners are:
>
> Bruce Perens
> John Goerzen
> Benj. Mako Hill
>
> The board of directors will vote on adding them to the board on its
> next meeting.
>
> The lists of voters, votes and results can be found online at the
> following address:
>
> http://www.spi-inc.org/secretary/votes/vote1/
>
> Starting today there will be a seven day dispute-period during which
> people can verify their votes using hash they received when they
> voted. After that period the information that ties a hash to a
> person will be removed from the voting database and the result will
> be final.

This is not a dispute but just a clarification request. I see some
votes which list only one option. As I understood the voting
guidelines, this effectively amounts to abstaining. On the other hand,
it is quite clear that the intent of those voters was different and
that they probably did not understand the details of the voting
procedure (or read the instructions...). I have no Condorcet voting
software at hand to run the results through but I know some people
have. Would it make a difference if those votes expressed a preference
over all other candidates and did not affect the races between the
others? Of course a similar question applies to all votes with less
than 8 options ranked. I don't think it would render this vote invalid
but nevertheless I consider it an interesting question.

Cheers,
Lukas

--
This is not a signature

Previously Lukas Geyer wrote:
> This is not a dispute but just a clarification request. I see some
> votes which list only one option. As I understood the voting
> guidelines, this effectively amounts to abstaining.

Correct.

> On the other hand, it is quite clear that the intent of those voters
> was different and that they probably did not understand the details of
> the voting procedure (or read the instructions...).

The instructions on the voting pages were pretty clear on that.

> I have no Condorcet voting software at hand to run the results through
> but I know some people have. Would it make a difference if those votes
> expressed a preference over all other candidates and did not affect
> the races between the others?

I suspect it will not make a difference, but I have not done a test
to confirm that. If you want to play with Condorcet voting you can use
the python script from http://www.electionmethods.org/ .

> Of course a similar question applies to all votes with less than 8
> options ranked. I don't think it would render this vote invalid but
> nevertheless I consider it an interesting question.

It might be interesting to see if that would change the result of the
vote; if so it might be educational for voters.

Wichert.

--
Wichert Akkerman <wichert(at)wiggy(dot)net> http://www.wiggy.net/
A random hacker

[CCing Wichert because I have a recommedation on ballot instructions.]

On Sat, Feb 22, 2003 at 02:11:41PM -0500, Lukas Geyer wrote:
> I see some votes which list only one option.

I see five: 33, 46, 67, 85, and 93[1].

> As I understood the voting guidelines, this effectively amounts to
> abstaining.

Yes.

> On the other hand,
> it is quite clear that the intent of those voters was different and
> that they probably did not understand the details of the voting
> procedure (or read the instructions...).

SPI members are not Floridians. We can expect basic literacy from them.
I suggest making future ballots more clear, perhaps with boldface text:

VOTING FOR ONE OPTION ONLY IS EQUIVALENT TO ABSTAINING. YOU MUST RANK
AT LEAST TWO OPTIONS FOR YOUR BALLOT TO AFFECT THE OUTCOME.

RANK ALL OPTIONS TO ENSURE THAT YOUR BALLOT HAS THE MOST EFFECT.

I strongly suggest that we do *not* try to interpret the five
aforementioned ballots under special rules that apply only to them. I
doubt that they'd affect the outcome, anyway.

> I have no Condorcet voting software at hand to run the results through
> but I know some people have. Would it make a difference if those votes
> expressed a preference over all other candidates and did not affect
> the races between the others? Of course a similar question applies to
> all votes with less than 8 options ranked. I don't think it would
> render this vote invalid but nevertheless I consider it an interesting
> question.

Wichert gave you a pointer to the software he used; if you'd do this
analysis I think it would be instructive. Considerably more ballots
(than five) failed to rank all options, and I'm not sure what impact
they had. If you can find this out and let the Board know before we
meet and act upon the results, it would be much appreciated.

I should re-iterate that under the SPI bylaws, this election does not in
and of itself appoint new people to the SPI Board; only a vote of the
Board itself can do this. This election basically serves as a
recommendation of the membership. Of course, the bylaws can be changed
to permit "direct election" of Board members by the organizational
membership, and if you feel that that should be the case, I suggest you
let the Bylaws Revision Committee know this:

spi-bylaws(at)lists(dot)spi-inc(dot)org

[1] http://www.spi-inc.org/secretary/votes/vote1/votes.xhtml

--
G. Branden Robinson, Treasurer
Software in the Public Interest, Inc.
treasurer(at)spi-inc(dot)org
http://www.spi-inc.org/

On Sat, Feb 22, 2003 at 08:21:40PM +0100, Wichert Akkerman wrote:
> Previously Lukas Geyer wrote:
> > This is not a dispute but just a clarification request. I see some
> > votes which list only one option. As I understood the voting
> > guidelines, this effectively amounts to abstaining.
>
> Correct.

Err, ...

i thought the voting system would rank the 'default' option as lower
than all the expressed votes and higher than all the non-expressed
votes. I even asked about this, and was told that it was so on this
list, if i remember well.

So a vote with only one option (one candidate) should be a vote where
this candidate rate higher than (defeats) all other candidates. How can
that be similar to abstention ?

So, from your response, i think that is not so, which feels strange as
the condorcet voting used here is different from the condorcet voting
used for debian leader election for example.

I personnally did rank all 8 candidates though.

Friendly,

Sven Luther

On Sat, Feb 22, 2003 at 02:56:42PM -0500, Branden Robinson / SPI Treasurer wrote:
> [CCing Wichert because I have a recommedation on ballot instructions.]
>
> On Sat, Feb 22, 2003 at 02:11:41PM -0500, Lukas Geyer wrote:
> > I see some votes which list only one option.
>
> I see five: 33, 46, 67, 85, and 93[1].
>
> > As I understood the voting guidelines, this effectively amounts to
> > abstaining.
>
> Yes.
>
> > On the other hand,
> > it is quite clear that the intent of those voters was different and
> > that they probably did not understand the details of the voting
> > procedure (or read the instructions...).
>
> SPI members are not Floridians. We can expect basic literacy from them.
> I suggest making future ballots more clear, perhaps with boldface text:
>
> VOTING FOR ONE OPTION ONLY IS EQUIVALENT TO ABSTAINING. YOU MUST RANK
> AT LEAST TWO OPTIONS FOR YOUR BALLOT TO AFFECT THE OUTCOME.
>
> RANK ALL OPTIONS TO ENSURE THAT YOUR BALLOT HAS THE MOST EFFECT.
>
> I strongly suggest that we do *not* try to interpret the five
> aforementioned ballots under special rules that apply only to them. I
> doubt that they'd affect the outcome, anyway.

Err,

Citing yourself on Thu, 06 Feb 2003 16:48:12 +0100 :
|
| On Thu, Feb 06, 2003 at 07:25:17AM +0100, Sven Luther wrote:
| > I suppose not voting for all 8 candidate (that is providing a string
| > with less than 8 characters) will result in "none of the above" to rank
| > higher than the candidates not figuring in the string, right ?
|
| It certainly should, or the vote tabulating software is horrendously
| buggy and should not be trusted to return valid results.

As i understand it, this is the complete oposite of what you are now
claiming, i may be wrong or have misunderstood something though.

Friendly,

Sven Luther

Previously Sven Luther wrote:
> So, from your response, i think that is not so, which feels strange as
> the condorcet voting used here is different from the condorcet voting
> used for debian leader election for example.

Debian indeed uses another form of Condorcet. The debian-vote
archives contain lots of discussions about various voting methods, and
based on those the decision was made to use Condorcet-SSD for SPI.

Wichert.

--
Wichert Akkerman <wichert(at)wiggy(dot)net> http://www.wiggy.net/
A random hacker

>>>>> In article <20030222195959(dot)GA2665(at)iliana>, Sven Luther
>>>>> <luther(at)dpt-info(dot)u-strasbg(dot)fr> writes:

> On Sat, Feb 22, 2003 at 08:21:40PM +0100, Wichert Akkerman wrote:
>> Previously Lukas Geyer wrote:
>> > This is not a dispute but just a clarification request. I see
>> > some votes which list only one option. As I understood the
>> > voting guidelines, this effectively amounts to abstaining.
>>
>> Correct.

> i thought the voting system would rank the 'default' option as
> lower than all the expressed votes and higher than all the
> non-expressed votes.

What default option?

> I even asked about this, and was told that it was so on this list,
> if i remember well.

Are you sure you did not do so on the debian voting list?

> So a vote with only one option (one candidate) should be a vote
> where this candidate rate higher than (defeats) all other
> candidates. How can that be similar to abstention ?

The way it works is that a vote for one candidate is ranks
that candidate higher than a non-existent default option, and other
candidates are unranked. So the vote did not cause the selected
candidate as winning over any other candidate.

I think the way it is set up is fine, since it allows people
to express I would prefer a over b, but I don't care whether the
others win or lose.

I would have thought the description of the voting mechanism
was clear about this, but obviously your mileage has varied.

manoj
--
Jim, it's Jack. I'm at the airport. I'm going to Tokyo and wanna pay
you the five-hundred I owe you. Catch you next year when I get back!
"The Rockford Files"
Manoj Srivastava <srivasta(at)acm(dot)org> <http://www.datasync.com/%7Esrivasta/>
1024R/C7261095 print CB D9 F4 12 68 07 E4 05 CC 2D 27 12 1D F5 E8 6E
1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C

>>>>> In article <20030222200329(dot)GB2665(at)iliana>, Sven Luther
>>>>> <luther(at)dpt-info(dot)u-strasbg(dot)fr> writes:

> On Sat, Feb 22, 2003 at 02:56:42PM -0500, Branden Robinson / SPI
> Treasurer wrote:

> Err,

> Citing yourself on Thu, 06 Feb 2003 16:48:12 +0100 :
>>
>> On Thu, Feb 06, 2003 at 07:25:17AM +0100, Sven Luther wrote:
>> > I suppose not voting for all 8 candidate (that is providing a
>> > string with less than 8 characters) will result in "none of the
>> > above" to rank higher than the candidates not figuring in the
>> > string, right ?
>>
>> It certainly should, or the vote tabulating software is
>> horrendously buggy and should not be trusted to return valid
>> results.

> As i understand it, this is the complete oposite of what you are
> now claiming, i may be wrong or have misunderstood something
> though.

I do not think this is relevant, since there _was_ no none of
the above option in this vote. (I now think that the current Debian
voting draft is vindicated by this confusion, since that insists on
having a default option for most votes).

manoj
--
"Is this foreplay?" "No, this is Nuke Strike. Foreplay has lousy
graphics. Beat me again." Duckert, in "Bad Rubber," Albedo #0
(comics)
Manoj Srivastava <srivasta(at)acm(dot)org> <http://www.datasync.com/%7Esrivasta/>
1024R/C7261095 print CB D9 F4 12 68 07 E4 05 CC 2D 27 12 1D F5 E8 6E
1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C

On Sat, Feb 22, 2003 at 09:28:13PM +0100, Wichert Akkerman wrote:
> Previously Sven Luther wrote:
> > So, from your response, i think that is not so, which feels strange as
> > the condorcet voting used here is different from the condorcet voting
> > used for debian leader election for example.
>
> Debian indeed uses another form of Condorcet. The debian-vote
> archives contain lots of discussions about various voting methods, and
> based on those the decision was made to use Condorcet-SSD for SPI.

Mmm, ok, i didn't know about that.

Friendly,

Sven Luther

Branden Robinson / SPI Treasurer <branden+spi-treasurer(at)deadbeast(dot)net> writes:

> Wichert gave you a pointer to the software he used; if you'd do this
> analysis I think it would be instructive. Considerably more ballots
> (than five) failed to rank all options, and I'm not sure what impact
> they had. If you can find this out and let the Board know before we
> meet and act upon the results, it would be much appreciated.

OK, as I don't know a thing about python and this was an opportunity
to learn more about Condorcet, I wrote my own program which basically
only calculates the matrix of defeats. Fortunately there are no
ambiguities to resolve, so I did not have to implement that SSD
stuff. I calculated the matrix for the SPI method and for the other
canonical method where all ranked options are preferred over all
non-ranked options. The result changes slightly: The two winners G and
B (Bruce Perens and John Goerzen) do not change but there is a tie for
the third place with the other method, between E and H (Benjamin Mako
Hill and Craig Small). Note that this is not a cyclic ambiguity, as
the direct vote between them is tied (65 votes each) and both lose
only to G and B. I have no idea how such a tie would have been
resolved in reality, but probably the point is moot this time. (Unless
Craig wants to start fighting...) Here is the program I wrote, the
textfile of raw votes and the two results:

http://people.debian.org/~lukas/spi/

> I should re-iterate that under the SPI bylaws, this election does not in
> and of itself appoint new people to the SPI Board; only a vote of the
> Board itself can do this. This election basically serves as a
> recommendation of the membership. Of course, the bylaws can be changed
> to permit "direct election" of Board members by the organizational
> membership, and if you feel that that should be the case, I suggest you
> let the Bylaws Revision Committee know this:
>
> spi-bylaws(at)lists(dot)spi-inc(dot)org

Yes, I would favor direct elections, therefore I Cc this message to
that list.

Best, Lukas

--
This is not a signature

On Sat, Feb 22, 2003 at 08:21:40PM +0100, Wichert Akkerman wrote:
> It might be interesting to see if that would change the result of the
> vote; if so it might be educational for voters.

By my count the actual results are:

Bruce, John, Benj, Craig, Scott, David G. tied with Jimmy, David H.

(Bruce beats John 74:47, John beats Benj 61:51, Benj beats Craig 61:50,
Craig beats Scott 65:52, Scott beats David G. and Jimmy 70:38/73:39,
David G. and Jimmy (tied on 54:54) beat David H. 62:45/56:52)

By my count, the results had the marked options been interpreted as being
preferred to unmarked options, the results would've been:

Bruce, John, Benj tied with Craig, Scott, Jimmy, David G., David H.

(84:46, 72:54/77:53, 66:56/74:54, 80:41, 60:57, 66:52)

21 people didn't include Benj in their ballot, 10 people didn't include
Craig.

For entertainment's sake, a probably buggy implementation of Single
Transferable Vote for three seats also seems to give Bruce, John and
Benj as the winners, with Craig just missing out.

Cheers,
aj

--
Anthony Towns <aj(at)humbug(dot)org(dot)au> <http://azure.humbug.org.au/~aj/>
I don't speak for anyone save myself. GPG signed mail preferred.

``Dear Anthony Towns: [...] Congratulations --
you are now certified as a Red Hat Certified Engineer!''

On Sat, Feb 22, 2003 at 02:56:42PM -0500, Branden Robinson / SPI Treasurer wrote:
> SPI members are not Floridians. We can expect basic literacy from them.
> I suggest making future ballots more clear, perhaps with boldface text:
> VOTING FOR ONE OPTION ONLY IS EQUIVALENT TO ABSTAINING. YOU MUST RANK
> AT LEAST TWO OPTIONS FOR YOUR BALLOT TO AFFECT THE OUTCOME.

Documenting a buggy interface isn't fixing the problem, and the obvious
interpretation of the vote "------1-" or "G" is to rank Bruce above
everyone else. For comparison, the latest drafts of the Debian voting
mechanism say:

A.6 Vote Counting
1. Each voter's ballot ranks the options being voted on. Not all
options need be ranked. Ranked options are considered
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
preferred to all unranked options. Voters may rank options
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
equally. Unranked options are considered to be ranked equally
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
with one another. Details of how ballots may be filled out
^^^^^^^^^^^^^^^^
will be included in the Call For Votes.

Cheers,
aj

--
Anthony Towns <aj(at)humbug(dot)org(dot)au> <http://azure.humbug.org.au/~aj/>
I don't speak for anyone save myself. GPG signed mail preferred.

``Dear Anthony Towns: [...] Congratulations --
you are now certified as a Red Hat Certified Engineer!''

Previously Anthony Towns wrote:
> A.6 Vote Counting
> 1. Each voter's ballot ranks the options being voted on. Not all
> options need be ranked. Ranked options are considered
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> preferred to all unranked options. Voters may rank options
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> equally. Unranked options are considered to be ranked equally
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> with one another. Details of how ballots may be filled out
> ^^^^^^^^^^^^^^^^
> will be included in the Call For Votes.

Sounds like you are trying to introduce the concept of 'everyone else
equally' into Condorcet, which feels a bit awkward. If you start doing
that you might also want to consider adding an 'anyone but X' option.

Wichert.

--
Wichert Akkerman <wichert(at)wiggy(dot)net> http://www.wiggy.net/
A random hacker

On Sun, Feb 23, 2003 at 07:49:28AM +1000, Anthony Towns wrote:
> Documenting a buggy interface isn't fixing the problem, and the obvious
> interpretation of the vote "------1-" or "G" is to rank Bruce above
> everyone else. For comparison, the latest drafts of the Debian voting
> mechanism say:

This omits a useful option. For instance, what if I do not feel that I know
enough about a given candidate and wish to express no opinion whatsoever
about that candidate?

This may be what happened here (I don't know). It can also happen in
Debian, but the draft excludes that possibility. Even if a voter is allowed
to rank all "unknown" candidates the same, there is still some preference
expressed between the "known" candidates and the unknown ones.

-- John

On 20030222T145642-0500, Branden Robinson / SPI Treasurer wrote:
> I should re-iterate that under the SPI bylaws, this election does not in
> and of itself appoint new people to the SPI Board; only a vote of the
> Board itself can do this. This election basically serves as a
> recommendation of the membership. Of course, the bylaws can be changed
> to permit "direct election" of Board members by the organizational
> membership, and if you feel that that should be the case, I suggest you
> let the Bylaws Revision Committee know this:
>
> spi-bylaws(at)lists(dot)spi-inc(dot)org

Where I live, in an organization such as what SPI would be if it would
be under local laws, it would be outright illegal for a board to decide
on its members (the decision must be by the membership, or by their
elected representatives). I was quite astonished to find out that this
is not the case in the USA, and I do consider it a flaw that the SPI
bylaws do not make it so.

The board *must* be under the membership's control, in all respects.
Therefore I am in favour of direct election (not just permitted, but
required).

--
%%% Antti-Juhani Kaijanaho % gaia(at)iki(dot)fi % http://www.iki.fi/gaia/ %%%

Please send any work-related email to <antkaij(at)mit(dot)jyu(dot)fi>

On Sat, 22 Feb 2003, John Goerzen wrote:

> On Sun, Feb 23, 2003 at 07:49:28AM +1000, Anthony Towns wrote:
> > Documenting a buggy interface isn't fixing the problem, and the obvious
> > interpretation of the vote "------1-" or "G" is to rank Bruce above
> > everyone else. For comparison, the latest drafts of the Debian voting
> > mechanism say:
>
> This omits a useful option. For instance, what if I do not feel that I know
> enough about a given candidate and wish to express no opinion whatsoever
> about that candidate?

FWIW I've been asked how this very thing could be done during the vote
by at least one voter. So to assume that everyone who voted "XYZ"
prefers them to everybody else might not be correct.

Peter
--
PGP signed and encrypted | .''`. ** Debian GNU/Linux **
messages preferred. | : :' : The universal
| `. `' Operating System
http://www.palfrader.org/ | `- http://www.debian.org/

Peter Palfrader <weasel(at)spi-inc(dot)org> writes:

> On Sat, 22 Feb 2003, John Goerzen wrote:
>
> > On Sun, Feb 23, 2003 at 07:49:28AM +1000, Anthony Towns wrote:
> > > Documenting a buggy interface isn't fixing the problem, and the
> > > obvious interpretation of the vote "------1-" or "G" is to rank
> > > Bruce above everyone else. For comparison, the latest drafts of
> > > the Debian voting mechanism say:
> >
> > This omits a useful option. For instance, what if I do not feel
> > that I know enough about a given candidate and wish to express no
> > opinion whatsoever about that candidate?
>
> FWIW I've been asked how this very thing could be done during the
> vote by at least one voter. So to assume that everyone who voted
> "XYZ" prefers them to everybody else might not be correct.

That might well be true but I fail to see how this applies to the
ballots ranking only one option. I do see that the SPI voting method
was correctly documented but I would like to have more flexibility,
maybe to be able to either prefer "no preference" for some candidate
and "equal preference" for some set of candidates. As a sidenote, the
explanation of Condorcet on

http://www.electionmethods.org/CondorcetEx.htm

does it in the (IMHO) intuitive way, that is non-ranked options are
implicitly ranked below ranked ones, and no preference is given
between them.

Lukas

--
This is not a signature

On Sun, Feb 23, 2003 at 06:25:58AM +0100, Peter Palfrader wrote:

> > This omits a useful option. For instance, what if I do not feel that I know
> > enough about a given candidate and wish to express no opinion whatsoever
> > about that candidate?
>
> FWIW I've been asked how this very thing could be done during the vote
> by at least one voter. So to assume that everyone who voted "XYZ"
> prefers them to everybody else might not be correct.

It appears to me that all that is really needed is a way of saying "I rank
this group of candidates equally"; whether this be because you don't know
enough about them to express a preference or whatever the reason may be.

It would also allow other candidates to be ranked as more or less desirable
than the group (or groups, even).

I think that this, possibly coupled with a "none of the above", would cover
everything -- if there is a way of accounting for this in a satisfactory
manner when it comes to working out the outcome.

Cheers,

Nick

--
Nick Phillips -- nwp(at)lemon-computing(dot)com
Bonus fortune cookie: buy 10 get 1 FREE!

On Sat, Feb 22, 2003 at 09:03:29PM +0100, Sven Luther wrote:
> Err,
>
> Citing yourself on Thu, 06 Feb 2003 16:48:12 +0100 :
> |
> | On Thu, Feb 06, 2003 at 07:25:17AM +0100, Sven Luther wrote:
> | > I suppose not voting for all 8 candidate (that is providing a string
> | > with less than 8 characters) will result in "none of the above" to rank
> | > higher than the candidates not figuring in the string, right ?
> |
> | It certainly should, or the vote tabulating software is horrendously
> | buggy and should not be trusted to return valid results.
>
> As i understand it, this is the complete oposite of what you are now
> claiming, i may be wrong or have misunderstood something though.

You appear to have ignored or overlooked:

Message-ID: <20030207050432(dot)GD17341(at)deadbeast(dot)net>

--
G. Branden Robinson, Treasurer
Software in the Public Interest, Inc.
treasurer(at)spi-inc(dot)org
http://www.spi-inc.org/

[why did you CC Manoj?]

On Sun, Feb 23, 2003 at 07:49:28AM +1000, Anthony Towns wrote:
> Documenting a buggy interface isn't fixing the problem,

I think whether it's a feature or a bug is a matter of opinion.

> and the obvious interpretation of the vote "------1-" or "G" is to
> rank Bruce above everyone else.

It's only "obvious" to people unaccustomed to preferential voting
systems.

> For comparison, the latest drafts of the Debian voting
> mechanism say:
>
> A.6 Vote Counting
> 1. Each voter's ballot ranks the options being voted on. Not all
> options need be ranked. Ranked options are considered
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> preferred to all unranked options. Voters may rank options
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> equally. Unranked options are considered to be ranked equally
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> with one another. Details of how ballots may be filled out
> ^^^^^^^^^^^^^^^^
> will be included in the Call For Votes.

Debian's voting system is required to have properties that take it
afield from the strict Condorcet Cloneproof/SSD method. Quorums,
default options, and supermajorities are all mandated properties under
the Debian Consitution, for instance.

I don't see the harm in letting SPI experiment with slightly simpler
methods.

--
G. Branden Robinson, Treasurer
Software in the Public Interest, Inc.
treasurer(at)spi-inc(dot)org
http://www.spi-inc.org/

On Sun, 23 Feb 2003, Nick Phillips wrote:

> On Sun, Feb 23, 2003 at 06:25:58AM +0100, Peter Palfrader wrote:
>
> > > This omits a useful option. For instance, what if I do not feel that I know
> > > enough about a given candidate and wish to express no opinion whatsoever
> > > about that candidate?
> >
> > FWIW I've been asked how this very thing could be done during the vote
> > by at least one voter. So to assume that everyone who voted "XYZ"
> > prefers them to everybody else might not be correct.
>
>
> It appears to me that all that is really needed is a way of saying "I rank
> this group of candidates equally"; whether this be because you don't know
> enough about them to express a preference or whatever the reason may be.

This came up before the election started. It would not be hard to allow
for that. Are you sure it doesn't mess up condorcet in such a way that
it deadlocks/gives no valid result/whatever?

Peter
--
PGP signed and encrypted | .''`. ** Debian GNU/Linux **
messages preferred. | : :' : The universal
| `. `' Operating System
http://www.palfrader.org/ | `- http://www.debian.org/

On Sun, Feb 23, 2003 at 01:26:20AM -0500, Branden Robinson / SPI Treasurer wrote:
> [why did you CC Manoj?]

From: Branden Robinson / SPI Treasurer <branden+spi-treasurer(at)deadbeast(dot)net>
Message-ID: <20030222195642(dot)GD11626(at)deadbeast(dot)net>
Mail-Followup-To: spi-general(at)lists(dot)spi-inc(dot)org,
secretary(at)spi-inc(dot)org

> On Sun, Feb 23, 2003 at 07:49:28AM +1000, Anthony Towns wrote:
> > and the obvious interpretation of the vote "------1-" or "G" is to
> > rank Bruce above everyone else.
> It's only "obvious" to people unaccustomed to preferential voting
> systems.

You're invited to survey everyone who voted that way as to whether it was
their intention to abstain.

> I don't see the harm in letting SPI experiment with slightly simpler
> methods.

We've already experimented with this method in past Debian elections;
people find it confusing and are mislead into rendering their votes
useless.

Cheers,
aj

--
Anthony Towns <aj(at)humbug(dot)org(dot)au> <http://azure.humbug.org.au/~aj/>
I don't speak for anyone save myself. GPG signed mail preferred.

``Dear Anthony Towns: [...] Congratulations --
you are now certified as a Red Hat Certified Engineer!''

>>>>> In article <20030222222429(dot)GA74462(at)gesundheit(dot)complete(dot)org>, John Goerzen <jgoerzen(at)complete(dot)org> writes:

> This omits a useful option. For instance, what if I do not feel
> that I know enough about a given candidate and wish to express no
> opinion whatsoever about that candidate?

> This may be what happened here (I don't know). It can also happen
> in Debian, but the draft excludes that possibility. Even if a
> voter is allowed to rank all "unknown" candidates the same, there
> is still some preference expressed between the "known" candidates
> and the unknown ones.

Hmm. In Debian, I suppose you can have two sets of candidates,
One set, all ranked equally, but above the default option (by
giving all these candidates an equal rank), and another, of all
unranked candidates (or a set of candidates all ranked equally but
below the default option).

So yes, you may express no preference amongst groups of
candidates, but you do have to express an opinion about whether you
find them acceptable options or not.

I do not find the inability to express no opinion whatsoever
much of a loss, personally; if I care so little, I would probably
mind them not being chosen anyway.

manoj
--
Never underestimate the power of stupid people in large groups. George
Carlin
Manoj Srivastava <srivasta(at)acm(dot)org> <http://www.datasync.com/%7Esrivasta/>
1024R/C7261095 print CB D9 F4 12 68 07 E4 05 CC 2D 27 12 1D F5 E8 6E
1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C

>>>>> In article <20030223062620(dot)GH11626(at)deadbeast(dot)net>, Branden
>>>>> Robinson / SPI Treasurer <branden+spi-treasurer(at)deadbeast(dot)net>
>>>>> writes:

> [why did you CC Manoj?]

I don't mind. My mail handling system is competent enough to
deal with duplicates.

> Debian's voting system is required to have properties that take it
> afield from the strict Condorcet Cloneproof/SSD method. Quorums,
> default options, and supermajorities are all mandated properties
> under the Debian Consitution, for instance.

> I don't see the harm in letting SPI experiment with slightly
> simpler methods.

I think I want to express the opinion that in this case,
having a default option would have better allowed me to express my
preferences, and would have been less surprising to people
unaccustomed to condorcet voting.

manoj
--
Even if you aren't in doubt, consider the mental welfare of the person
who has to maintain the code after you, and who will probably put
parens in the wrong place. -- Larry Wall in the perl man page
Manoj Srivastava <srivasta(at)acm(dot)org> <http://www.datasync.com/%7Esrivasta/>
1024R/C7261095 print CB D9 F4 12 68 07 E4 05 CC 2D 27 12 1D F5 E8 6E
1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C

>>>>> In article <20030222215239(dot)GC25781(at)wiggy(dot)net>, Wichert Akkerman <wichert(at)wiggy(dot)net> writes:

> Sounds like you are trying to introduce the concept of 'everyone
> else equally' into Condorcet, which feels a bit awkward.

Umm, could you explain why you think it is awkward?

> If you start doing that you might also want to consider adding an
> 'anyone but X' option.

Debian already has this; rank the one person unacceptable to
you below the default option, rank everyone else equally above the
default option.

manoj
--
Fortune's Real-Life Courtroom Quote #19: Q: Doctor, how many autopsies
have you performed on dead people? A: All my autopsies have been
performed on dead people.
Manoj Srivastava <srivasta(at)acm(dot)org> <http://www.datasync.com/%7Esrivasta/>
1024R/C7261095 print CB D9 F4 12 68 07 E4 05 CC 2D 27 12 1D F5 E8 6E
1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C

On Sun, Feb 23, 2003 at 01:21:49AM -0500, Branden Robinson / SPI Treasurer wrote:
> On Sat, Feb 22, 2003 at 09:03:29PM +0100, Sven Luther wrote:
> > Err,
> >
> > Citing yourself on Thu, 06 Feb 2003 16:48:12 +0100 :
> > |
> > | On Thu, Feb 06, 2003 at 07:25:17AM +0100, Sven Luther wrote:
> > | > I suppose not voting for all 8 candidate (that is providing a string
> > | > with less than 8 characters) will result in "none of the above" to rank
> > | > higher than the candidates not figuring in the string, right ?
> > |
> > | It certainly should, or the vote tabulating software is horrendously
> > | buggy and should not be trusted to return valid results.
> >
> > As i understand it, this is the complete oposite of what you are now
> > claiming, i may be wrong or have misunderstood something though.
>
> You appear to have ignored or overlooked:
>
> Message-ID: <20030207050432(dot)GD17341(at)deadbeast(dot)net>

Well, maybe, but does this mean you are no more saying what you did say
in the cited mail ?

Anyway, it does not matter much for me, since i ranked everyone, but
when i voted, i had doubts about that, didn't find documentation or
explanation easily, and was maybe a bit fooled by my knowledge of the
debian voting system and all this long discution in debian-vote.

Just let it be clearer for next time.

Friendly,

Sven Luther

On Sat, Feb 22, 2003 at 04:24:29PM -0600, John Goerzen wrote:
> On Sun, Feb 23, 2003 at 07:49:28AM +1000, Anthony Towns wrote:
> > Documenting a buggy interface isn't fixing the problem, and the obvious
> > interpretation of the vote "------1-" or "G" is to rank Bruce above
> > everyone else. For comparison, the latest drafts of the Debian voting
> > mechanism say:
>
> This omits a useful option. For instance, what if I do not feel that I know
> enough about a given candidate and wish to express no opinion whatsoever
> about that candidate?
>
> This may be what happened here (I don't know). It can also happen in
> Debian, but the draft excludes that possibility. Even if a voter is allowed
> to rank all "unknown" candidates the same, there is still some preference
> expressed between the "known" candidates and the unknown ones.

You rank the unknown ones above the 'none of the above', and the known
ones you absolutely don't want below the 'none of the above'.

Would this not solve the problem ?

Friendly,

Sven Luther

Previously Manoj Srivastava wrote:
> Debian already has this; rank the one person unacceptable to
> you below the default option, rank everyone else equally above the
> default option.

Yes, the default option it the trick Debian uses to accomplish that.
However it can also cause an election result that is undesired: the
'further discussion' outcome.

Think of what would happen when you vote for a new government and the
result is 'further discussion'. The old government (which possibly
nobody wants) stays on and all progress is halted.

It might be doable to change the handling of further discussion so that
you can rank it, but it is ignored in the final result since it is not
a real valid option.

Wichert.

--
Wichert Akkerman <wichert(at)wiggy(dot)net> http://www.wiggy.net/
A random hacker

On Sun, 23 Feb 2003, Sven Luther wrote:

> On Sun, Feb 23, 2003 at 01:21:49AM -0500, Branden Robinson / SPI Treasurer wrote:
> > On Sat, Feb 22, 2003 at 09:03:29PM +0100, Sven Luther wrote:
> > > Err,
> > >
> > > Citing yourself on Thu, 06 Feb 2003 16:48:12 +0100 :
> > > |
> > > | On Thu, Feb 06, 2003 at 07:25:17AM +0100, Sven Luther wrote:
> > > | > I suppose not voting for all 8 candidate (that is providing a string
> > > | > with less than 8 characters) will result in "none of the above" to rank
> > > | > higher than the candidates not figuring in the string, right ?
> > > |
> > > | It certainly should, or the vote tabulating software is horrendously
> > > | buggy and should not be trusted to return valid results.
> > >
> > > As i understand it, this is the complete oposite of what you are now
> > > claiming, i may be wrong or have misunderstood something though.
> >
> > You appear to have ignored or overlooked:
> >
> > Message-ID: <20030207050432(dot)GD17341(at)deadbeast(dot)net>
>
> Well, maybe, but does this mean you are no more saying what you did say
> in the cited mail ?
>
> Anyway, it does not matter much for me, since i ranked everyone, but
> when i voted, i had doubts about that, didn't find documentation or
> explanation easily, and was maybe a bit fooled by my knowledge of the
> debian voting system and all this long discution in debian-vote.

The voting page itself said:

| Voting "XY" states that you prefer candidate X to Y. However, it does
| not express any preference relationships for any of the other
| candidates.

Peter
--
PGP signed and encrypted | .''`. ** Debian GNU/Linux **
messages preferred. | : :' : The universal
| `. `' Operating System
http://www.palfrader.org/ | `- http://www.debian.org/

>>>>> In article <20030223092531(dot)GA14177(at)wiggy(dot)net>, Wichert Akkerman <wichert(at)wiggy(dot)net> writes:

> Previously Manoj Srivastava wrote:
>> Debian already has this; rank the one person unacceptable to you
>> below the default option, rank everyone else equally above the
>> default option.

> Yes, the default option it the trick Debian uses to accomplish
> that. However it can also cause an election result that is
> undesired: the 'further discussion' outcome.

Why is that undesired in Debian? Indeed, since you cannot
force anyone in Debisan to stop discussing anything anyway, how is
the reject this proposition any different whatsoever than further
discussion?

> It might be doable to change the handling of further discussion so
> that you can rank it, but it is ignored in the final result since
> it is not a real valid option.

I posit that if you had a defeat this proposition option, it
could be turned into a further discussion option by anyone
with a MUA and a will to type.

> Think of what would happen when you vote for a new government and
> the result is 'further discussion'. The old government (which
> possibly nobody wants) stays on and all progress is halted.

But neither Debian nor SPI is a government, and,
furthermore, I think that, say, if we can't select a new DPL (since
a significant set of debian developers found all of them
unacceptable, having the old DPL around would be nice, since he was
selected by some folks a year ago.

manoj
--
Darth Vader! Only you would be so bold! Princess Leia Organa
Manoj Srivastava <srivasta(at)acm(dot)org> <http://www.datasync.com/%7Esrivasta/>
1024R/C7261095 print CB D9 F4 12 68 07 E4 05 CC 2D 27 12 1D F5 E8 6E
1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C

Previously Manoj Srivastava wrote:
> Why is that undesired in Debian? Indeed, since you cannot
> force anyone in Debisan to stop discussing anything anyway, how is
> the reject this proposition any different whatsoever than further
> discussion?

This is not Debian; I'm not really interested in how relevant this
might be to Debian.

Wichert.

--
Wichert Akkerman <wichert(at)wiggy(dot)net> http://www.wiggy.net/
A random hacker

>>>>> In article <20030223111503(dot)GC14177(at)wiggy(dot)net>, Wichert Akkerman
>>>>> <wichert(at)wiggy(dot)net> writes:

> Previously Manoj Srivastava wrote:
>> Why is that undesired in Debian? Indeed, since you cannot force
>> anyone in Debisan to stop discussing anything anyway, how is the
>> reject this proposition any different whatsoever than further
>> discussion?

> This is not Debian; I'm not really interested in how relevant this
> might be to Debian.

Cute. You cut away the context that shows that this is indeed
about debian voting. Let us see what the exchange was, really, with
context:

>>>>> Previously Anthony Towns wrote:
>>>>>> A.6 Vote Counting
>>>>>> 1. Each voter's ballot ranks the options being voted on. Not all
>>>>>> options need be ranked. Ranked options are considered
>>>>>> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>>>>>> preferred to all unranked options. Voters may rank options
>>>>>> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>>>>>> equally. Unranked options are considered to be ranked equally
>>>>>> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>>>>>> with one another. Details of how ballots may be filled out
>>>>>> ^^^^^^^^^^^^^^^^
>>>>>> will be included in the Call For Votes.

>>>> In article <20030222215239(dot)GC25781(at)wiggy(dot)net>, Wichert Akkerman
>>>> <wichert(at)wiggy(dot)net> writes:

>>>>> Sounds like you are trying to introduce the concept of
>>>>> 'everyone else equally' into Condorcet, which feels a bit
>>>>> awkward. If you start doing that you might also want to
>>>>> consider adding an 'anyone but X' option.

>>> Previously Manoj Srivastava wrote:
>>>> Debian already has this; rank the one person unacceptable to you
>>>> below the default option, rank everyone else equally above the
>>>> default option.

>>> Yes, the default option it the trick Debian uses to accomplish
>>> that. However it can also cause an election result that is
>>> undesired: the 'further discussion' outcome.

>> Why is that undesired in Debian? Indeed, since you cannot force
>> anyone in Debisan to stop discussing anything anyway, how is the
>> reject this proposition any different whatsoever than further
>> discussion?

> This is not Debian; I'm not really interested in how relevant this
> might be to Debian.

Heh. After 6 email exchanges where we are talking about the
Debian draft, you suddenly want to say this conversation was not
about Debian's voting system?

manoj
--
"It's OK to do the right thing... as long as you don't get caught."
The Lone Contractor
Manoj Srivastava <srivasta(at)acm(dot)org> <http://www.datasync.com/%7Esrivasta/>
1024R/C7261095 print CB D9 F4 12 68 07 E4 05 CC 2D 27 12 1D F5 E8 6E
1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C

Previously Manoj Srivastava wrote:
> Heh. After 6 email exchanges where we are talking about the
> Debian draft, you suddenly want to say this conversation was not
> about Debian's voting system?

The Debian draft was only brought up to show how Debian is considering
changing its voting system to introduce equal-ordering group into
Condorcet.

Wichert.

--
Wichert Akkerman <wichert(at)wiggy(dot)net> http://www.wiggy.net/
A random hacker

On Sun, Feb 23, 2003 at 10:21:18AM +0100, Sven Luther wrote:
> On Sat, Feb 22, 2003 at 04:24:29PM -0600, John Goerzen wrote:
> > This omits a useful option. For instance, what if I do not feel that I know
> > enough about a given candidate and wish to express no opinion whatsoever
> > about that candidate?
>
> You rank the unknown ones above the 'none of the above', and the known
> ones you absolutely don't want below the 'none of the above'.

No, because it would be implying a preference of the known ones over the
unknown ones.

On Sun, Feb 23, 2003 at 07:28:29AM +0100, Peter Palfrader wrote:

> for that. Are you sure it doesn't mess up condorcet in such a way that
> it deadlocks/gives no valid result/whatever?

No, hence the last sentence in my previous mail: "...if there is a way of
accounting for this in a satisfactory manner when it comes to working out
the outcome."

But if it is a valid preference to express, and the voting system isn't able
to account for it, then the voting system is inadequate. This may be due to
a desire to "always have a clear outcome", which is IMHO not to be preferred
over "allowing the voters to fully and clearly express their preferences."

Cheers,

Nick
--
Nick Phillips -- nwp(at)lemon-computing(dot)com
Live in a world of your own, but always welcome visitors.

>>>>> In article <20030223181209(dot)GA9735(at)gesundheit(dot)complete(dot)org>, John Goerzen <jgoerzen(at)complete(dot)org> writes:

> On Sun, Feb 23, 2003 at 10:21:18AM +0100, Sven Luther wrote:
>> On Sat, Feb 22, 2003 at 04:24:29PM -0600, John Goerzen wrote:
>> > This omits a useful option. For instance, what if I do not feel
>> > that I know enough about a given candidate and wish to express
>> > no opinion whatsoever about that candidate?
>>
>> You rank the unknown ones above the 'none of the above', and the
>> known ones you absolutely don't want below the 'none of the
>> above'.

> No, because it would be implying a preference of the known ones
> over the unknown ones.

If you truly have no preference, why not lump all these
options at the same level above the default option? All are now
eligible, but you have expressed no preference, apart from saying I
find them acceptable.

If you don't even have an opinion on whether the options are
acceptable or not, I suggest you educate yourself; I would not want
to pander lazy exercise of ones franchise.

manoj
--
Disclaimer: "These opinions are my own, though for a small fee they be
yours too." Dave Haynie
Manoj Srivastava <srivasta(at)acm(dot)org> <http://www.datasync.com/%7Esrivasta/>
1024R/C7261095 print CB D9 F4 12 68 07 E4 05 CC 2D 27 12 1D F5 E8 6E
1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C

Antti-Juhani Kaijanaho writes ("Re: Result for vote regarding new members for the board of directors"):
> The board *must* be under the membership's control, in all respects.
> Therefore I am in favour of direct election (not just permitted, but
> required).

Quite so.

Ian.

On Sun, Feb 23, 2003 at 07:04:38PM +1000, Anthony Towns wrote:
> You're invited to survey everyone who voted that way as to whether it was
> their intention to abstain.
[...]
> We've already experimented with this method in past Debian elections;
> people find it confusing and are mislead into rendering their votes
> useless.

This doesn't persuade me that education can't solve the issue.

Heck, we can even have the voting software send a distinct
vote acknowledgement message when people submit a ballot with only one
option on it, explaining that the ballot is valid but meaningless.

One nice thing about the mechanics of both SPI's and Debian's voting
system is that they allow for revised ballots if the voter makes an
error (or changes his or her mind).

--
G. Branden Robinson, Treasurer
Software in the Public Interest, Inc.
treasurer(at)spi-inc(dot)org
http://www.spi-inc.org/