Double Bubble, LA's Toil & Trouble

February 27, 2008

Three weeks after Super Tuesday,
thousands of Los Angeles votes may
finally be counted

On February 5, I was in Los Angeles, helping with a nonpartisan
election protection
effort.  We received calls reporting plenty
of problems
, some familiar, some new. 
But early in the day, it was clear that there was a serious issue confronting
independent voters.

In California, voters unaffiliated with any party could vote
in the primary for the Democratic party or the American Independent party (the
Republican primary was open to registered Republicans only).  Theoretically.

Problem one: bad information.  Some pollworkers didn't offer the choice of a
partisan primary ballot.  Worse, others
did affirmative harm, incorrectly telling unaffiliated voters that they
couldn't vote in a partisan primary.

Dem Ballot Problem two: bad ballots. 
In Los Angeles county, the unaffiliated ballots baffled thousands of
voters.  The ballots themselves were just
numbers and bubbles, like the answer sheet for a standardized test.  The candidates were listed in a separate
booklet, like the test's exam questions. 
Democratic candidates, rotating in position from voter to voter, were
assigned to bubbles #8-15; American Independent candidates, also rotating, were
assigned bubbles #8-10.  So bubbles #8-10
could have been votes for either party.  To
tell one from the other, voters were supposed to fill in a different bubble—bubble # 5 or 6—with their party choice. 

AI Ballot
The "double
bubble
" system was sufficiently confusing that about 49,500
voters
got it wrong.  Officials are
now trying to determine whether they can reliably count some of the ballots:
for example, voters who filled in bubbles #11-15
(which should be unambiguous Democratic votes), or voters in precincts where
all of the unaffiliated voters who wanted partisan ballots checked
in
for the same party's primary.  It
looks like the Los Angeles county clerk may now try
to count both categories
.  But that
will still leave the many ambiguous ballots uncounted, and can't possibly help
the people who were turned away without casting a ballot at all.

In the continuing brouhaha, one element has been left out of
the conversation: properly programmed touchscreen systems might have avoided the
toil and trouble of both problems.

To be crystal clear: this is not an endorsement of touchscreens over optical scans, or
punchcards, or rock-paper-scissors. 
There are plenty of concerns
with most, if not all, of the existing touchscreen systems.  This is also not a declaration that all paper ballots are flawed: the design
that confused thousands of Angelenos could have, and should have, been avoided.

Still, both the pollworker problem and the ballot problem
were big issues caused by small lapses in a series of rules.  We've seen both types before, and we'll see them
again, and it's useful to keep our own limitations in mind.  Most election procedures are sets of "if,
then" instructions: if unaffiliated, then ask about a partisan ballot . . . if
partisan ballot, then ask which party.  There
are an awful lot of "if, then" rules for people to keep track of on election
day.  And one thing computers can be good
at is walking people through those rules, step by step.