Election Data

Let me first thank you all for your input on this.

I can understand the sentiment that to give out this data would be somehow problematic, since it wasn't originally stated that it would given out.* However, if it is anonymized, how does that really violate the privacy of any voter?* In fact, you could even remove a random subset of the votes, so that even if a voter looked at what got published and saw a unique ballot identical to the one he cast, he would still have no certainty that it was his vote.* At that point we get into the philosophical realm of asking, how much do you have do dilute the data before it's no longer a concern?* In some sense the data for these elections (in which there was no warning of possible publication) is actually _more_ valuable, because if you were to add a notice that there would be anonymous usage analysis, fewer people might vote strategically.* (This kind of "irrational" behavior is the subject of a great Google lecture by Dan Ariely.)


Some people have countered that in public governmental elections, data is not might public.* This is incorrect.* Many cities publish anonymous ballot results online.* Here in my home of San Francisco, for instance. (And of course, just because the government does something in their election process, that doesn't mean it's right - just look at the terrible plurality voting system that almost all municipalities use.)


Of course we could also ask someone within Fedora to do the analysis, where we're primarily looking for things like the frequency of "polarized" ballots (all min and max scores), and other bits of data like the ones discussed here:

http://rangevoting.org/HaikuIcon.html

Whether or not this data can be made of any use for us, I'd still propose the notice in future elections that anonymous ballot data will be available after the election, either pre- or post-processing.* If that is what it would take to satisfy your ethical standards, I think there would be value in it.* In fact I think the voters themselves generally appreciate that kind of transparency more than they dislike it.


Thanks again for your time.

Clay

On Wed, Jul 23, 2008 at 07:53, Bill Nottingham <notting@redhat.com> wrote:

Paul W. Frields (stickster@gmail.com) said:

> > Debian also makes their election data public, though they use a worse

> > and much more complex Condorcet method, called "Shulze".

> > http://www.debian.org/vote/2003/leader2003_tally.txt

>

> I don't see a huge problem with this as long as the ballots are

> anonymized. *Vote data is often analyzed for trends and other purposes,

> and with Fedora being an open, transparent project overall, I think this

> request doesn't go counter to our goals. *But I think the Board should

> probably make this decision.



Considering that we didn't actually state before the election that

we would collect, anonymize, and mine the data, I don't think it's

a good idea to do that now.



Bill



_______________________________________________

fedora-advisory-board mailing list

fedora-advisory-board@redhat.com

http://www.redhat.com/mailman/listinfo/fedora-advisory-board



--
clay shentrup
phone: 206.801.0484

"Iraq? *No, YOU rock!"

_______________________________________________
fedora-advisory-board mailing list
fedora-advisory-board@redhat.com
http://www.redhat.com/mailman/listinfo/fedora-advisory-board

> Some people have countered that in public governmental elections, data is not might public.

Apparently I had beer for breakfast.* CORRECTION: "..not _made_ public."


On Wed, Jul 23, 2008 at 09:29, CLAY S <clay@brokenladder.com> wrote:

Let me first thank you all for your input on this.

I can understand the sentiment that to give out this data would be somehow problematic, since it wasn't originally stated that it would given out.* However, if it is anonymized, how does that really violate the privacy of any voter?* In fact, you could even remove a random subset of the votes, so that even if a voter looked at what got published and saw a unique ballot identical to the one he cast, he would still have no certainty that it was his vote.* At that point we get into the philosophical realm of asking, how much do you have do dilute the data before it's no longer a concern?* In some sense the data for these elections (in which there was no warning of possible publication) is actually _more_ valuable, because if you were to add a notice that there would be anonymous usage analysis, fewer people might vote strategically.* (This kind of "irrational" behavior is the subject of a great Google lecture by Dan Ariely.)



Some people have countered that in public governmental elections, data is not might public.* This is incorrect.* Many cities publish anonymous ballot results online.* Here in my home of San Francisco, for instance. (And of course, just because the government does something in their election process, that doesn't mean it's right - just look at the terrible plurality voting system that almost all municipalities use.)



Of course we could also ask someone within Fedora to do the analysis, where we're primarily looking for things like the frequency of "polarized" ballots (all min and max scores), and other bits of data like the ones discussed here:


http://rangevoting.org/HaikuIcon.html

Whether or not this data can be made of any use for us, I'd still propose the notice in future elections that anonymous ballot data will be available after the election, either pre- or post-processing.* If that is what it would take to satisfy your ethical standards, I think there would be value in it.* In fact I think the voters themselves generally appreciate that kind of transparency more than they dislike it.



Thanks again for your time.

Clay

On Wed, Jul 23, 2008 at 07:53, Bill Nottingham <notting@redhat.com> wrote:


Paul W. Frields (stickster@gmail.com) said:

> > Debian also makes their election data public, though they use a worse

> > and much more complex Condorcet method, called "Shulze".

> > http://www.debian.org/vote/2003/leader2003_tally.txt

>

> I don't see a huge problem with this as long as the ballots are

> anonymized. *Vote data is often analyzed for trends and other purposes,

> and with Fedora being an open, transparent project overall, I think this

> request doesn't go counter to our goals. *But I think the Board should

> probably make this decision.



Considering that we didn't actually state before the election that

we would collect, anonymize, and mine the data, I don't think it's

a good idea to do that now.



Bill



_______________________________________________

fedora-advisory-board mailing list

fedora-advisory-board@redhat.com

http://www.redhat.com/mailman/listinfo/fedora-advisory-board



--
clay shentrup
phone: 206.801.0484

"Iraq? *No, YOU rock!"



--
clay shentrup
phone: 206.801.0484

"Iraq? *No, YOU rock!"

_______________________________________________
fedora-advisory-board mailing list
fedora-advisory-board@redhat.com
http://www.redhat.com/mailman/listinfo/fedora-advisory-board

CLAY S wrote:

Of course we could also ask someone within Fedora to do the analysis,
where we're primarily looking for things like the frequency of
"polarized" ballots (all min and max scores), and other bits of data
like the ones discussed here:

http://rangevoting.org/HaikuIcon.html

This is something that could be made a useful part of the voting
application itself. If the aggregate results are available after the
fact (I believe this is set per election now) then other analysis of the
aggregate results also seems fair. But it does require that someone
wants to code this. So there's two parts to this question:


1) Does the Board agree that this processing and dissemenation of
aggregate data is fine?


2) Does Nigel want to add this to the election software or allow someone
else to add it?


-Toshio

_______________________________________________
fedora-advisory-board mailing list
fedora-advisory-board@redhat.com
http://www.redhat.com/mailman/listinfo/fedora-advisory-board

I am not sure if it does range voting, but selectricity.org does a
fair amount of good data presentation.



On 7/23/08, Toshio Kuratomi <a.badger@gmail.com> wrote:
> CLAY S wrote:
>
>> Of course we could also ask someone within Fedora to do the analysis,
>> where we're primarily looking for things like the frequency of
>> "polarized" ballots (all min and max scores), and other bits of data
>> like the ones discussed here:
>> http://rangevoting.org/HaikuIcon.html
>>
> This is something that could be made a useful part of the voting
> application itself. If the aggregate results are available after the
> fact (I believe this is set per election now) then other analysis of the
> aggregate results also seems fair. But it does require that someone
> wants to code this. So there's two parts to this question:
>
> 1) Does the Board agree that this processing and dissemenation of
> aggregate data is fine?
>
> 2) Does Nigel want to add this to the election software or allow someone
> else to add it?
>
> -Toshio
>
>

--
Sent from Gmail for mobile | mobile.google.com

_______________________________________________
fedora-advisory-board mailing list
fedora-advisory-board@redhat.com
http://www.redhat.com/mailman/listinfo/fedora-advisory-board

On Wed, Jul 23, 2008 at 11:29 AM, CLAY S <clay@brokenladder.com> wrote:
> Let me first thank you all for your input on this.
>
> I can understand the sentiment that to give out this data would be somehow
> problematic, since it wasn't originally stated that it would given out.
> However, if it is anonymized, how does that really violate the privacy of
> any voter? In fact, you could even remove a random subset of the votes, so
> that even if a voter looked at what got published and saw a unique ballot
> identical to the one he cast, he would still have no certainty that it was
> his vote. At that point we get into the philosophical realm of asking, how
> much do you have do dilute the data before it's no longer a concern? In
> some sense the data for these elections (in which there was no warning of
> possible publication) is actually _more_ valuable, because if you were to
> add a notice that there would be anonymous usage analysis, fewer people
> might vote strategically. (This kind of "irrational" behavior is the
> subject of a great Google lecture by Dan Ariely.)

Why should Fedora make any of it public knowing that doing so will
affect the way people vote?

John

_______________________________________________
fedora-advisory-board mailing list
fedora-advisory-board@redhat.com
http://www.redhat.com/mailman/listinfo/fedora-advisory-board

On Wed, Jul 23, 2008 at 10:37, inode0 <inode0@gmail.com> wrote:

Why should Fedora make any of it public knowing that doing so will

affect the way people vote?



John
i'm saying it will affect it _for the better_.* which is "bad" from a research point of view, but good for fedora.

--
clay shentrup
phone: 206.801.0484

"Iraq? No, YOU rock!"

_______________________________________________
fedora-advisory-board mailing list
fedora-advisory-board@redhat.com
http://www.redhat.com/mailman/listinfo/fedora-advisory-board

On Wed, Jul 23, 2008 at 1:02 PM, CLAY S <clay@brokenladder.com> wrote:
> On Wed, Jul 23, 2008 at 10:37, inode0 <inode0@gmail.com> wrote:
>>
>> Why should Fedora make any of it public knowing that doing so will
>> affect the way people vote?
>
> i'm saying it will affect it _for the better_. which is "bad" from a
> research point of view, but good for fedora.

I misunderstood strategic voting to be a supposed benefit of this.

What is meant by "strategic" voting and why is it bad?

John

_______________________________________________
fedora-advisory-board mailing list
fedora-advisory-board@redhat.com
http://www.redhat.com/mailman/listinfo/fedora-advisory-board

On Wed, 2008-07-23 at 11:26 -0700, CLAY S wrote:
> strategic voting is voting in a less expressive way that increases
> your personal expected utility. for instance, the DH3 pathology with
> condorcet voting (used by debian, unfortunately) is where a candidate
> wins whom virtually everyone agrees is the worst, because they all use
> a strategy that is _beneficial_ at the individual level. it's like
> the prisoners' dilemma. everyone is worse off because they each use
> an advisable strategy.
>
> an example of strategic voting was when 90% of the nader-supporting
> voters voted for someone else (namely gore) back in 2000, because that
> had a higher expected value (where expected value is the chance your
> vote changes the election outcome times the difference in utility that
> causes for you).
>
> in score voting, you might actually feel that X=10, Y=7, and Z=0 --
> but if Y and Z are the clear front-runners, you can increase your
> expected value by voting X=10, Y=10, Z=0. that's more strategic.
>
> the nice thing about score voting is that it behaves quite nicely even
> when people are strategic. here are some links which discuss this.
>

With no disrespect: Can y'all take this offlist? It's really not on
topic to what this list is for.

Thanks,
-sv



_______________________________________________
fedora-advisory-board mailing list
fedora-advisory-board@redhat.com
http://www.redhat.com/mailman/listinfo/fedora-advisory-board

strategic voting is voting in a less expressive way that increases your personal expected utility.* for instance, the DH3 pathology with condorcet voting (used by debian, unfortunately) is where a candidate wins whom virtually everyone agrees is the worst, because they all use a strategy that is _beneficial_ at the individual level.* it's like the prisoners' dilemma.* everyone is worse off because they each use an advisable strategy.


an example of strategic voting was when 90% of the nader-supporting voters voted for someone else (namely gore) back in 2000, because that had a higher expected value (where expected value is the chance your vote changes the election outcome times the difference in utility that causes for you).


in score voting, you might actually feel that X=10, Y=7, and Z=0 -- but if Y and Z are the clear front-runners, you can increase your expected value by voting X=10, Y=10, Z=0.* that's more strategic.

the nice thing about score voting is that it behaves quite nicely even when people are strategic.* here are some links which discuss this.


http://rangevoting.org/StratHonMix.html
http://rangevoting.org/ShExpRes.html
http://rangevoting.org/PleasantSurprise.html

http://rangevoting.org/RVstrat1.html
http://rangevoting.org/RVstrat2.html
http://rangevoting.org/RVstrat3.html

http://rangevoting.org/RVstrat4.html
http://rangevoting.org/RVstrat5.html
http://rangevoting.org/RVstrat6.html

http://rangevoting.org/CondBurial.html

regards,
clay

On Wed, Jul 23, 2008 at 11:08, inode0 <inode0@gmail.com> wrote:

On Wed, Jul 23, 2008 at 1:02 PM, CLAY S <clay@brokenladder.com> wrote:


> On Wed, Jul 23, 2008 at 10:37, inode0 <inode0@gmail.com> wrote:

>>

>> Why should Fedora make any of it public knowing that doing so will

>> affect the way people vote?

>

> i'm saying it will affect it _for the better_. *which is "bad" from a

> research point of view, but good for fedora.



I misunderstood strategic voting to be a supposed benefit of this.



What is meant by "strategic" voting and why is it bad?



John



_______________________________________________

fedora-advisory-board mailing list

fedora-advisory-board@redhat.com

http://www.redhat.com/mailman/listinfo/fedora-advisory-board



--
clay shentrup
phone: 206.801.0484

"Iraq? No, YOU rock!"

_______________________________________________
fedora-advisory-board mailing list
fedora-advisory-board@redhat.com
http://www.redhat.com/mailman/listinfo/fedora-advisory-board

On Wed, Jul 23, 2008 at 10:29 AM, CLAY S <clay@brokenladder.com> wrote:
> Let me first thank you all for your input on this.
>

> Some people have countered that in public governmental elections, data is
> not might public. This is incorrect. Many cities publish anonymous ballot
> results online. Here in my home of San Francisco, for instance. (And of
> course, just because the government does something in their election
> process, that doesn't mean it's right - just look at the terrible plurality
> voting system that almost all municipalities use.)
>

The data is only made anonymous for elections after it has been
advertised via legislation that it will be made so. I would prefer
that if research is going to be done on voting.. that it is announced
before the election and what kinds of research and how the data would
be anonymized is listed and approved before it is done so. This way
people are aware of the data being used for research AND that the
method for anonymizing has been publically looked at to make sure it
really anonymizes the data versus rot12's the ip addresses and account
names .



--
Stephen J Smoogen. -- BSD/GNU/Linux
How far that little candle throws his beams! So shines a good deed
in a naughty world. = Shakespeare. "The Merchant of Venice"

_______________________________________________
fedora-advisory-board mailing list
fedora-advisory-board@redhat.com
http://www.redhat.com/mailman/listinfo/fedora-advisory-board