Stan suggested:

> I agree...I'd rather let users vote as many points as they
> want across as many items as they want.  Then, we could
> use a program on the backend to normalize their voting.
> I.e., if I voted 100 points for item A, and 50 points for item
> B, and 1 point for item C, and nothing else, the program
> would normalize that to 6.6 votes for item A, and 3.3 votes
> for item B, and something like 0.06 votes for item C.
>
> As it is now, if 11 things are interesting out of the 28 items
> on the ballot, the voter is screwed...they can't indicate that
> interest!

John Burke already gave a pretty good answer for this year..
I will just add:

Generally I like Stan's idea;  although if we went back to 100
votes per ballot (that's what we used for quite a few years) as
a practical matter I think that would give all voters enough
granularity that they could always spare one or two "dollars"
for everything they think is "interesting".  While a program to
normalize everything would certainly not be hard to do, from
an operational process point of view staying with a fixed 100
votes (or 150 or whatever) is probably "cleaner".

One reason we dropped down to 10 from 100 is that a certain
HP manager with the initials JB (and I believe others) fairly
strongly encouraged us to do so;  with I believe the idea that
they wanted voters to focus on just those items that really were
most important to them;  as John already also mentioned.   ;-)

At least for SIGIMAGE, when we had 100 votes per ballot we
used to drop items that got zero votes in any given year.  We
stopped doing that when the number of votes went down to 10
for essentially the reason Stan mentioned:  If you only have
10 and there are more (in some cases many more) that 10
items on the ballot, you have no way to say "not a high-pri for
us, but it's not a bad idea;  don't discard it."

On balance, think I am leaning in favor of going back to 100
votes per ballot....   For next year, of course.... Regardless of
what makes it both to and through the final overall SIB vote,
SIGImage/SQL will not discard any items because they don't
get any or many votes this year:  There will be another
chance for all next time around...

Ken Sletten
SIGImage/SQL Chair