[Haskell-cafe] How do we decide on the new logo?
Fritz Ruehr
decide* on the new logo for the Haskell site.
I'm not sure what the protocol and procedure for voting and tallying
should be, but see below for a suggestion.
(* a higher-order decision, very appropriate)
If you haven't seen the page on the Wiki with proposed new logos, be
sure to go there and take a look
(but recall that submissions are now closed):
<http://haskell.org/haskellwiki/Haskell_logos/New_logo_ideas>
I want to thank Don for proposing the contest and everyone who
contributed logo designs or modifications.
As the designer of the last "official" logo, I think this new crop is
terrific, with many suggestions that are more
professional-looking and sleeker than the old one: I will have a hard
time deciding which one to vote for.
Without starting a war on the theory of voting systems, perhaps we
should use a system which allows for
a certain amount of secondary (etc.) preference to be expressed?
(Uh-oh, here come Control.Monad.Voting.HareSTV and
Control.Monad.Voting.BordaCount and
a hundred other variations, complete with back-tracking and
trampolined continuations and ... .)
Once we have a winning design, we could perhaps award the designer(s)
with a T-shirt or some similar item.
We have a small amount of "CafeBucks" (or whatever) accrued in the
CafePress account, despite all efforts to avoid profit
(rather like avoiding success at all costs). In the past, these funds
have been used to purchase courtesy shirts for a few
Haskell luminaries, at the discretion of the store proprietor.
(Currently I think there are about $60 available.)
-- Fritz (Ruehr)
_______________________________________________
Haskell-Cafe mailing list
Haskel...@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
Achim Schneider
> Now that the new year is upon us, I suppose we must decide how to
> decide* on the new logo for the Haskell site.
> I'm not sure what the protocol and procedure for voting and tallying
> should be, but see below for a suggestion.
> (* a higher-order decision, very appropriate)
>
> If you haven't seen the page on the Wiki with proposed new logos, be
> sure to go there and take a look
> (but recall that submissions are now closed):
>
> <http://haskell.org/haskellwiki/Haskell_logos/New_logo_ideas>
>
> I want to thank Don for proposing the contest and everyone who
> contributed logo designs or modifications.
> As the designer of the last "official" logo, I think this new crop
> is terrific, with many suggestions that are more
> professional-looking and sleeker than the old one: I will have a
> hard time deciding which one to vote for.
>
> Without starting a war on the theory of voting systems, perhaps we
> should use a system which allows for
> a certain amount of secondary (etc.) preference to be expressed?
>
>
Step 1: Crunch down the size of proposals by factoring out common themes
(eg. all the >\= logos count as one)
Step 2: Determine the winner by polling preferences, same-level
preference (ambivalence) allowed
(eg. place 1 for logos C and D, place 2 for A and place 3 for B)
Step 3: Re-open contest, accepting submissions _using_ the winning
logo, in the categories a) colour schemes[1] b), official shapes[2] c),
font[3] to go to b), d) layouts of b) + c)
Step 4: Repeat step 2 for every category of step 3
Step 5: Announce the new buzzword-compliant branding and hand over a
t-shirt to the one who wrote code to apply colour schemes to logos for
displaying step 3.
Obviously, we need to know the winner of step 2 to completely define
step 3.
Did I miss anything?
[1] coloured, monochrome and b/w
[2] shape details[4] vs. (coloured and monochrome vs. b/w)
[3] not forgetting its licence
[4] like whether or not to completely connect that > and \ to a lambda
--
(c) this sig last receiving data processing entity. Inspect headers
for copyright history. All rights reserved. Copying, hiring, renting,
performance and/or quoting of this signature prohibited.
Henning Thielemann
> Without starting a war on the theory of voting systems, perhaps we
> should use a system which allows for
> a certain amount of secondary (etc.) preference to be expressed?
Give everyone 10 points and let every voter assign these points to his
favorite logos, where it is possible to give more than one of his points
to the same logo. (Or give every user one point and let him choose how
to divide this into fractions which can be assigned to logos. Or give
every voter any number of points he want and scale them to 1 afterwards.)
Another question is how to handle logos with variations. I think all
logos of one idea should be grouped and considered one object and the
favorite variant can be voted on later.
Henning Thielemann
> Step 2: Determine the winner by polling preferences, same-level
> preference (ambivalence) allowed
> (eg. place 1 for logos C and D, place 2 for A and place 3 for B)
We recently had to vote for the new design of our university's website.
This was done by asking every voter for an order of preference, with no
equal preferences allowed. However, when the maintainer of the voting
system was asked, how these answers are processed, he didn't know an
answer. I think he finally converted positions to scores and added them.
However, I suspect in chosing the scores for each position, he had an
essential influence of the outcome of the election.
roco...@theorem.ca
> Step 2: Determine the winner by polling preferences, same-level
> preference (ambivalence) allowed
> (eg. place 1 for logos C and D, place 2 for A and place 3 for B)
The only reasonable method of voting using this ranking data is one of the
Condorcet methods. How you break ties doesnt matter much to me.
Wikimedia, Debian, Gentoo, and Software in the Public Intrest all use
Schulze method for what that is worth.
However, I'm more concerned about who gets to vote and how many times do
they get to vote.
--
Russell O'Connor <http://r6.ca/>
``All talk about `theft,''' the general counsel of the American Graphophone
Company wrote, ``is the merest claptrap, for there exists no property in
ideas musical, literary or artistic, except as defined by statute.''
Achim Schneider
> Achim Schneider schrieb:
>
> > Step 2: Determine the winner by polling preferences, same-level
> > preference (ambivalence) allowed
> > (eg. place 1 for logos C and D, place 2 for A and place 3 for B)
>
> We recently had to vote for the new design of our university's
> website. This was done by asking every voter for an order of
> preference, with no equal preferences allowed. However, when the
> maintainer of the voting system was asked, how these answers are
> processed, he didn't know an answer. I think he finally converted
> positions to scores and added them. However, I suspect in chosing the
> scores for each position, he had an essential influence of the
> outcome of the election.
>
I intended to sum up all scaled preferences for every logo, and take the
smallest one as the winner. "Scaled" meaning that every voter, in the
end, had the same total points to distribute... like in your system. I
like mine because it shifts the calculation burden from the voter to
the program doing the processing.
A vote of all 10's would be equivalent to a vote of all ones, and a
vote of just one one would be equivalent to a vote of one one and the
rest two's.
--
(c) this sig last receiving data processing entity. Inspect headers
for copyright history. All rights reserved. Copying, hiring, renting,
performance and/or quoting of this signature prohibited.
Richard Kelsall
..
> Step 3: Re-open contest, accepting submissions _using_ the winning
> logo, in the categories a) colour schemes[1] b), official shapes[2] c),
> font[3] to go to b), d) layouts of b) + c)
This is a good suggestion. I would like small adjustments to the logo
to be possible before it's frozen. Some kind of Step 3 will result
in a much better logo. For example, I really like the pyramid from
above / square containing three triangles that Lenny222 submitted, but
I wouldn't choose this precise colour scheme and form. I didn't have
time to enter an alternative. When the field has been significantly
reduced I think people will be willing to expend effort improving the
remaining entries.
Richard.
Sebastian Sylvan
> On Sat, 3 Jan 2009, Achim Schneider wrote:
>
> Step 2: Determine the winner by polling preferences, same-level
>> preference (ambivalence) allowed
>> (eg. place 1 for logos C and D, place 2 for A and place 3 for B)
>>
>
> The only reasonable method of voting using this ranking data is one of the
> Condorcet methods. How you break ties doesnt matter much to me. Wikimedia,
> Debian, Gentoo, and Software in the Public Intrest all use Schulze method
> for what that is worth.
>
Yes. Condorcet voting picks the best compromise and is IMO the way to do
this - we won't all agree on the best logo, but at least we can pick the
least disliked one. It doesn't need to be super sophisticated, just a box
next to each logo where you can enter a rank in any range you like (1 being
most preferred, empty boxing being equivalent to +Inf),
allowing multiple entries to share the same rank.
--
Sebastian Sylvan
+44(0)7857-300802
UIN: 44640862
Eelco Lempsink
Since there already is a condorcet voting package on Hackage, I made a
simple (HAppS powered) web app where you can drag-n-drop your
preferences. See http://github.com/eelco/voting/tree/master for the
code (contributions more than welcome! Note, there's also a jQuery
branch which has a bit different drag-n-drop behaviour) and http://code.tupil.com/voting/
for a live demo.
It needs a bit more work but mainly a whole bulk of decisions, like
* Limit voting, if so how? Email confirmation, IP based, vote once,
once per day?
* Maybe don't show the results until the contest is over?
I, for one, very much welcome any benevolent dictator to make these
decisions, because we can probably argue about pros and cons for
months. Since Don started the contest (http://www.haskell.org/pipermail/haskell-cafe/2008-December/051836.html
) and also seems to have some ideas about the voting process (http://www.haskell.org/pipermail/haskell-cafe/2008-December/052257.html
), I hereby officially appoint him to lead to masses. (Does it work
like that? ;)
--
Regards,
Eelco Lempsink