[Wikipedia-l] Logo results as tallied pairwise (Condorcet)

Wed Oct 1 17:21:23 UTC 2003

Andre Engels wrote:

>On 30 Sep 2003, Erik Moeller wrote:
>
>>In any case, most invalid ballots have been marked as
>>invalid long before the vote ended, and all invalid ones are listed on
>>[[International logo vote/Invalid ballots]]. The rules for what type of
>>ballot is acceptable and what type is not have been made very clear
>>repeatedly, and, to my knowledge, nobody has voted for Pat Buchanan.
>>
>
>No, but some have voted in the reasonable expectation that their vote was
>valid, whereas it was not. The rules were laid out to ensure that people
>were a valid user (i.e. with at least 10 contributions) on some Wikipedia.
>They were not laid out to make voting more difficult by creating
>technicalities.
>
I think most people opted out of the voting method discussion because it 
became far too arcane.  When the time came for the 2nd round vote on the 
logo they just gave it their best shot, trusting that it would be dealt 
with fairly.  As a system becomes more complicated there are more things 
that can go wrong.  I suspect (but can't be sure of) where my vote went 
wrong.  Somehow, my Meta user page did not link to another project, but 
I do have at least 10 edits in Meta alone so it was not needed, and 
nobody ever said that Meta was an excluded project.

>People have linked directly to their elsewhere user page rather than through
>their meta page, people have had links on their meta page to Wikipedia pages
>on www.wikipedia.com because they were so long users, people have given their
>nick and said "I am from the English Wikipedia". Those have all not been
>counted. Why? Because there is the possibility they cheated? No. Because
>they did not go with some random rule. Or in some cases, like the one we
>are describing now, because they simply had technical problems.
>
This was bound to have problems, and was compounded by the slow server. 
 When I was voting I twice received an edit conflict message.  A vote 
with such a wide range of options where all the votes are put into the 
same "article" is bound to have slow loading problems.

>>Note that some people refused to accept the first round of voting because
>>the rules were too lax and cheating was too easy. If some people refuse to
>>accept the second round of voting because the rules are too strict and
>>cheating is too difficult, then I think these complaints can be fairly
>>considered to negate each other.
>>
>In other words, that people may have cheated on the first round does not
>matter, because on the second round there were people who did not cheat
>who were counted as cheaters? And that in the second round there are people
>whose vote did not count whereas some (many?) think it should does not matter
>because in the first round there are people whose vote did count whereas
>some (many?) think it should not?
>
>That's a strange definition of complaints negating, in my opinion. If you
>had said, there are people who think the _second round_ rules were not strict
>enough as well as people who said they were too strict, then you would have
>had a point. But not this way. It's like saying, "Wikipedia has too much on
>A and too little on B. But that cancels out, so we're not going to change
>either article."
>
I never considered cheating to be a major issue.  The purpose of the 
first round was to reduce the number of candidates to a reasonable size 
by eliminating those that didn't have a chance of winning.  Ten winners 
in that round was an arbitrary number, and adding the eleventh was a 
perfectly sensible act.  The most that a cheater could have accomplished 
in that round is have his pet design added to the top ten to the 
exclusion of another design, but if this ten is taken as an arbitrary 
and flexible amount, and the voting director has the right to include 
borderline losers on the list there would be no problem.

One very good way to create confusion in a voting system is to develop 
rules for problems that aren't there.  Confusion has been a far bigger 
problem than cheating.  The number of people who would cheat in this 
kind of situation is small, and the number whose cheating would be on a 
large scale is a small proportion of all the cheaters.  Catching and 
invalidating the efforts of the large scale cheaters makes sense, but 
when it comes to the small scale cheating that might only affect one 
ballot we would do better to develop situations where that cheating 
doesn't matter.

Ec