Re: [WebForm] Computer General

[http://dummy.us.eu.org/robert (Robert)]

 > From: "Doug Wagoner douglaswagoner-at-gmail.com |201105blogsubmit douglaswagoner/Example Allow|" <http://dummy.us.eu.org/robert>
 > Date: Thu, 14 Apr 2011 11:47:54 -0400
 >

 > 
 > Thanks for your feedback.  I am not sure how the sorting algorithm would
 > change the process.  What you are probably thinking of as the tedious part
 > is not related to the sort, but rather to the process of doing a 1:1 compare
 > of every chosen value.  By eliminating shortcuts and actually voting for
 > each pair, you can uncover details of what you  value that are sometimes
 > surprising.

I see.  So, the process actually prefers a bubblesort.  I didn't realize
that.

 > And pre-culled "popular" values similarly demolishes the personal quality of
 > the process.

I'm assuming you mean that, by making it random, it enhances the
correctness of the final result.

 > But I would be interested in seeing your version should you complete it.

Well, allowing for the shortcomings of not going through every combination
and being bias towards the results of others' value orderings, quicksort
should have an average of O(n*log(n)) combinations and, using previous
results of others, I'm guessing it would typically be less than that.

My only motivation in my suggestion was cutting down on the number of
combinations which may make the test more appealing and less likely to
have people give up (I almost did).  But, of course, quantity is not
quality.

 > Again, thanks for writing.
 > 
 > == Doug
 > 
 > On Thu, Apr 14, 2011 at 10:46, Douglas Wagoner <
 > http://www.horologium.dreamhost.com/~djwagone> wrote:
 > > Name: Robert
 > > Email: http://dummy.us.eu.org/robert
 > > Message:
 > > I took your values test.
 > >
 > > Doing the part which sorts the values was a bit tedious.
 > >
 > > Being a geek, I thought about it, and realized that if you used a
 > > "quicksort" rather than a "bubblesort" (or equivalent), you could reduce the
 > > number of combinations required to be evaluated.
 > >
 > > In fact, if you wanted to get really fancy, you could use popular sorted
 > > values as the starting point and determining the pivot for the quicksort.
 > >  That could drastically reduce the number of combinations required.
 >