I have my own material to add to the commentary from Dan Karlan, but mine
comes from a different direction. I did some analysis that sheds light on
some serious flaws with our convention voting. The flaws have nothing to
do with whether or not we use approval voting. In fact, the use of
approval voting made evident some issues that I suspected were happening,
but previously didn’t have granular level information to be able to see how
extensive a problem it was.
First I was curious about how many of the convention delegates took
advantage of the approval voting and cast votes for more people than the
number of seats to be filled. I'd suspected that delegates did take
advantage of the approval voting since we had so many candidates receive a
majority vote on a single ballot, and in the past very few of those elected
to the same positions had received majority votes. Facts are more useful
than hunches, though. David Blau sent the original ballots from Columbus
home with me, so I was able to answer the question.
For the at-large race, 299 delegates cast ballots, and 97 of them voted for
more than 5 candidates. That’s about 32%.
For the judicial committee, 206 delegates cast ballots, and 27 of them
voted for more than 7 candidates. That’s about 13%.
I'm not a mind-reader to be able to say with certainty why the percentage
is smaller for JC than for at-large, but one possible difference is that
there were more seats to fill and a lot of the judicial committee
candidates were not as well known as the at-large candidates.
As I did after the 2012 convention, I did a full audit of the 2014 ballots,
and that revealed some mistakes made on-site, *mistakes that in one race
this year impacted the outcome of the election.*
When I explain my findings below, I am not in any way criticizing those who
served as Secretary/tellers for the convention. Secretaries/tellers are
human, and mistakes happen. Manually tallying votes in our conventions is
significantly more difficult than managing an election to fill one LNC
vacancy with 18 people voting.
Whatever you imagine the job of convention secretary is like, it’s harder
than that. The convention elections are a big job anyway, and you want to
get it right so you put pressure on yourself, and there are time
constraints, and the delegates are perpetually restless because you’re not
done yet, and some state chairs won’t follow directions or they turn in
confusing or illegible tally sheets, or they turn in more votes than
they’re entitled to, and you are constantly interrupted by people asking
questions or turning in motions for later, and there is loud background
noise interfering with your ability to even hear the person next to you
reading numbers off the state tally sheets, and you have to do
error-checking, and some delegates keep insisting that we conduct other
business during the vote tally so as to further distract the Secretary from
the important job at hand, and you need to simultaneously wrap up the
previous round of votes and document what is happening on the convention
floor and generate the next round of ballots, etc. In 2012, because the
chair’s race took so long to complete, there were even overlapping votes,
such that delegates were voting on the next election before we finished
tallying the previous one. It is maddening torture for the Secretary.
My tallying team made mistakes in the 2012 convention. Luckily the only
one of them that would have impacted the outcome was caught onsite by doing
a state-by-state review of the spreadsheet on the screen in the convention
hall. The other mistakes that I found after-the-fact did not impact any of
the outcomes.
In Columbus this year, we didn’t do any state-by-state reviews of the
Secretary’s tally spreadsheet until we got to the vice-chair race, which
had a margin of only 3 votes between Mr. Vohra and Mr. Goldstein. The
secretary team double-checked those numbers, we did a state-by-state review
of the spreadsheet on-screen, and those numbers seem to have been correct.
However, we should have done state-by-state reviews sooner, even when the
results didn't seem as close.
Based on my audit of the ballots following the convention:
*LNC Chair Election – Round 1*
In the first round of the chair's race, Mr. Sarwark's votes were overstated
by 5 in the results shown on-screen, but that mistake didn't push him to a
majority, and Mr. Pojunis would still have been dropped for round two even
with the corrected totals.
The results reported on-site for LNC Chair Round 1 were:
Sarwark – 161 (40.66%)
Neale – 135 (34.09%)
Pojunis – 66 (16.67%)
NOTA – 31 (7.83%)
Starchild (write-in) – 3 (0.76%)
TOTAL: 396 ballots cast
However, the audited results from the state tally sheets show the results
should have been:
Sarwark - 156 (39.90%)
Neale - 135 (34.53%)
Pojunis - 66 (16.88%)
NOTA - 31 (7.93%)
Starchild write-in - 3 (0.77%)
TOTAL: 391 ballots cast
The mistake in that round seems to have come from the Massachusetts tally
sheet, which was reported in the draft minutes to have been 9 votes for
Sarwark and 3 votes for Neale. However, the tally sheet showed only 4
votes for Sarwark, and 3 votes for Neale. The state’s tally sheet
indicated they only had 7 delegates present at the time, not 12. This
mistake could potentially have been caught by noticing that the recorded
votes added up to more delegates than the tally sheet said were present, by
noticing the credentials system only had 7 Massachusetts delegates
credentialed at the time, or by an on-screen review of the state-by-state
totals.
Note that I have attached scans of the state tally sheets in question which
did not match the numbers reported in the draft minutes.
*LNC Chair Election – Round 2*
In the second round of the chair's race, the results reported on-screen
were:
Sarwark – 194 (51.32%)
Neale – 144 (38.10%)
NOTA – 40 (10.58%)
TOTAL: 378 ballots cast
However, the audited results from the state tally sheets show the results
should have been:
Sarwark – 194 (50.00%)
Neale – 153 (39.43%)
NOTA – 40 (10.31%)
Pojunis (write-in) – 1 (0.26%)
TOTAL: 388 ballots cast
Mr. Sarwark had exactly 50%, but not a majority which is necessary for
election. Had these been the displayed results on-site, under our rules,
Mr. Neale would have been dropped from the ballot, and there would have
been a run-off between Mr. Sarwark and NOTA. While it is quite possible
that Mr. Sarwark would have crossed the majority threshold in that
scenario, it is not a given. At one point in the 2012 convention, a chair
candidate had 50% of the vote, but he was not ultimately elected in the
subsequent voting.
The difference of 9 votes for Neale came from the Florida tally sheet,
which was reported in the draft minutes to have been 6 votes for Sarwark, 2
for Neale, and 3 for NOTA. However the tally sheet showed 6 votes for
Sarwark, 11 for Neale, and 3 for NOTA, with 20 delegates present and
voting. This mistake could potentially have been caught by an on-screen
review of the state-by-state totals.
The on-screen results in Columbus did not report the Pojunis write-in vote
that was written in the margin of the Indiana tally sheet, though write-ins
are allowed for any eligible person. The ballot is merely a listing of
those who have been nominated. Pojunis had been removed from the ballot
(un-nominated essentially), but that doesn't prohibit write-in votes for
him. See RONR (11th ed.) p. 430 line 17 – p. 431 line 8, which states:
“Strictly speaking, nominations are not necessary when an election is by
ballot or roll call, since each member is free to vote for any eligible
person, whether he has been nominated or not. In most societies, however,
it is impractical to proceed to an election without first making
nominations. While members are always free to "write in," on a ballot, the
name of an eligible person who has not been nominated, or to vote for an
eligible non-nominee during a roll-call vote, under normal conditions it is
likely that most members will confine their choice to the nominees. Without
nominations, voting might have to be repeated many times before a candidate
achieved the required majority.”
Also see RONR (11th ed.) p. 441 line 25 – p. 442 line 2, which states:
“If the bylaws require the election of officers to be by ballot and there
is only one nominee for an office, the ballot must nevertheless be taken
for that office unless the bylaws provide for an exception in such a case.
In the absence of the latter provision, members still have the right, on
the ballot, to cast "write-in votes" for other eligible persons.”
Perhaps the Indiana delegation chair wasn’t sure if he could count that
write-in vote or not, and he did not include it in his own total number of
votes cast, though it was clearly written in the margin of their tally
sheet so the tellers could decide what to do with it. This mistake would
not likely have been caught by an on-screen review of the state-by-state
totals if the delegation chair was under the impression that he couldn’t
count that vote, though if the results came down to a difference of 1 vote,
the Indiana delegation chair might have asked if that write-in should be
counted, or it might have prompted a thorough double-checking of all the
tally sheets which could have caught it.
The natural question at this point is: What do we do now that we know that
a chair candidate was declared elected with only 50%, rather than a
majority? After some research, I’ve decided that the answer is: *nothing*.
RONR (11th ed) p. 446 has this to say about the question:
"Because the voting body itself is the ultimate judge of election disputes,
only that body has the authority to resolve them in the absence of a bylaw
or special rule of order that specifically grants another body that
authority. Thus, for example, when an election has been conducted at a
membership meeting or in a convention of delegates, an executive board,
even one that is given full power and authority over the society's affairs
between meetings of the body that conducted the election, may not entertain
a point of order challenging, or direct a recount concerning, the announced
election result. While an election dispute is immediately pending before
the voting body, however, it may vote to refer the dispute to a committee
or board to which it delegates power to resolve the dispute."
And this is another instance of a point I made during the first LNC meeting
of the term, from RONR (11th ed.) p. 483, which states:
“In any event, no action of the board can alter or conflict with any
decision made by the assembly of the society, and any such action of the
board is null and void (see p. 577, ll. 23–33).”
If the convention decided a person was elected, the board cannot take up a
subsequent dispute about it, even with evidence that a mistake was made.
On-site error-checking is pretty important because once the convention has
adjourned, we have no other body that can correct an erroneous election
result. I am aware of one instance in the 2010 convention where the votes
for two candidates were accidentally being reversed as they were entered
into the tally sheet, but it was caught about half-way through the tally
process. If that hadn’t been caught until after adjournment, there would
have been no remedy for the wrong person having been declared elected.
*Other Officer Elections*
My audited results matched the draft minutes for the Vice Chair and
Secretary elections. The Treasurer’s race only had one candidate, so it
was elected by acclamation rather than by ballot.
*At-Large Elections*
For the officer elections, the only thing turned in to the Secretary team
for tabulation is the state tally sheet. It is not possible in those
elections to demonstrate whether the delegation chairs correctly totaled
their delegation’s votes, though there is plenty of room for error there as
well. Delegation chairs are humans, too, and they can just as easily make
mistakes as anyone else.
Because the approval voting system used for the At-Large and Judicial
Committee elections required that the individual delegate ballots be turned
in along with the state tally sheets, that has helped demonstrate that
delegation chairs do make non-trivial numbers of mistakes when they total
their state results as well.
In the At-Large race this year, 5 of the 40 reporting states (12.5%) had
incorrect totals on the sheet they gave the Secretary. In the Judicial
Committee race, 5 of the 33 reporting states (15.15%) had incorrect totals
on the sheet they gave the Secretary.
I presume that a similar number of mistakes occurred on the state tally
sheets for the officer elections as well, but because we don't have the
individual delegate ballots from those races, that cannot be verified.
I will go out of my way here to commend the Texas delegation for how
meticulous they were about their voting process. It was evident from the
markings on their ballots. They even caught and disallowed it when a
delegate voted twice for the same candidate in an approval vote.
The results reported on-site for LNC At-Large were:
Redpath – 225 (75.25%)
Goldstein – 191 (63.88%)
Evan McMahon – 165 (55.18%)
Johnson – 161 (53.85%)
Craig – 136 (45.48%)
McLendon – 136 (45.48%)
Pojunis – 134 (44.82%)
Starchild – 124 (41.47%)
Pickens – 97 (32.44%)
McVay – 58 (19.40%)
Fulner – 41 (13.71%)
Pedersen – 36 (12.04%)
NOTA – 0 (0.00%)
TOTAL: 299 ballots cast
However, the audited results from the state tally sheets show the results
as:
Redpath - 215 (71.91%)
Goldstein - 191 (63.88%)
Evan McMahon - 165 (55.18%)
Johnson - 161 (53.85%)
Craig - 136 (45.49%)
McLendon - 136 (45.49%)
Pojunis - 134 (44.82%)
Starchild - 124 (41.47%)
Pickens - 97 (32.44%)
McVay - 58 (19.40%)
Fulner - 41 (13.71%)
Pedersen - 36 (12.04%)
NOTA - 0 (0.00%)
Brett Bittner write-in - 1 (0.33%)
Chuck Moulton write-in - 1 (0.33%)
Lou Jasikoff write-in - 1 (0.33%)
TOTAL: 299 ballots cast
The differences between the on-site results and the audit of the state
tally sheets were:
- The draft minutes show that Indiana cast 34 votes for Redpath, but
their tally sheet shows only 24 votes for Redpath.
- California reported 1 write-in vote for Brett Bittner; Florida
reported 1 write-in vote for Chuck Moulton; Kentucky reported 1 write-in
vote for Lou Jasikoff; none of the write-in votes were reported on-site.
These mistakes could potentially have been caught by an on-screen review of
the state-by-state totals.
However, an audit of the individual delegate ballots revealed 13 errors
made by 5 state delegation chairs when completing the state tally sheets.
Results from the individual delegate ballots were found to be:
Redpath - 217 (72.58%)
Goldstein - 191 (63.88%)
Evan McMahon - 165 (55.18%)
Johnson - 161 (53.85%)
Craig - 137 (45.82%)
McLendon - 137 (45.82%)
Pojunis - 133 (44.48%)
Starchild - 123 (41.14%)
Pickens - 98 (32.78%)
McVay - 58 (19.40%)
Fulner - 42 (14.05%)
Pedersen - 35 (11.71%)
NOTA - 0 (0.00%)
Brett Bittner write-in - 1 (0.33%)
Chuck Moulton write-in - 1 (0.33%)
Lou Jasikoff write-in - 1 (0.33%)
TOTAL: 299 ballots cast
The differences between the state tally sheets and the audit of the
individual delegate ballots were (only the erroneous totals are listed, and
other reported votes for other candidates were correct):
- The Colorado tally sheet showed Pojunis (8), Redpath (2), and Pickens
(6), but the delegate ballots had votes for Pojunis (7), Redpath (3), and
Pickens (7).
- The North Carolina tally sheet showed Pojunis (3), but the delegate
ballots had votes for Pojunis (2).
- The Pennsylvania tally sheet showed Redpath (3), Craig (0), Starchild
(6), Pickens (4), McLendon (1), and Fulner (0), but the delegate ballots
had votes for Redpath (4), Craig (1), Starchild (5), Pickens (3), McLendon
(2), and Fulner (1).
- The South Carolina tally sheet showed Pickens (0), but the delegate
ballots had votes for Pickens (1).
- The Tennessee tally sheet showed Pojunis (1) and Pedersen (1), but the
delegate ballots had votes for Pojunis (2) and Pedersen (0).
*Judicial Committee Elections*
Repeating myself from above, in the Judicial Committee race, 5 of the 33
reporting states (15.15%) had incorrect totals on the sheet they gave the
secretary.
The results reported on-site for Judicial Committee were:
Linnabary – 117 (56.80%)
Carling – 116 (56.31%)
Wolf – 112 (54.37%)
Sink-Burris – 110 (53.40%)
Power – 107 (51.94%)
Latham – 104 (50.49%)
Visek – 104 (50.49%)
Fockler – 100 (48.54%)
Youngs – 82 (39.81%)
Albin – 77 (37.38%)
Starchild – 76 (36.89%)
Carol McMahon – 60 (29.13%)
NOTA – 3 (1.46%)
TOTAL: 206 ballots cast
However, the audited results from the state tally sheets show the results
as:
Carling - 124 (60.19%)
Linnabary - 117 (56.80%)
Wolf - 112 (54.37%)
Sink-Burris - 110 (53.40%)
Power - 107 (51.94%)
Latham - 104 (50.49%)
Visek - 104 (50.49%)
Fockler - 100 (48.54%)
Youngs - 83 (40.29%)
Albin - 77 (37.38%)
Starchild - 76 (36.89%)
Carol McMahon - 60 (29.13%)
NOTA - 3 (1.46%)
TOTAL: 206 ballots cast
The differences between the on-site results and the audit of the state
tally sheets were:
- The draft minutes show that Ohio cast 30 votes for Carling, but the
state tally sheet shows 38 votes for Carling.
- The draft minutes show that Texas cast 12 votes for Youngs, but the
state tally sheet shows 13 votes for Youngs.
These mistakes could potentially have been caught by an on-screen review of
the state-by-state totals.
However, an audit of the individual delegate ballots revealed 5 errors made
by 5 state delegation chairs when completing the state tally sheets.
Results from the individual delegate ballots were found to be:
Carling - 124 (60.19%)
Linnabary - 118 (57.28%)
Wolf - 113 (54.85%)
Sink-Burris – 111 (53.88%)
Power - 107 (51.94%)
Latham - 105 (50.97%)
Visek - 104 (50.49%)
Fockler - 100 (48.54%)
Youngs - 83 (40.29%)
Albin - 77 (37.38%)
Starchild - 77 (37.38%)
McMahon - 60 (29.13%)
NOTA - 3 (1.46%)
The differences between the state tally sheets and the audit of the
individual delegate ballots were (only the erroneous totals are listed, and
other reported votes for other candidates were correct):
- The Colorado tally sheet showed 3 votes for Wolf, but the delegate
ballots showed 4 votes for Wolf.
- The Illinois tally sheet showed 0 votes for Starchild, but the
delegate ballots showed 1 vote for Starchild.
- The New York tally sheet showed 4 votes for Sink-Burris, but the
delegate ballots showed 5 votes for Sink-Burris.
- The Ohio tally sheet showed 46 votes for Linnabary, but the delegate
ballots showed 47 votes for Linnabary.
- The Pennsylvania tally sheet showed 0 votes for Latham, but the
delegate ballots showed 1 vote for Latham.
*Conclusion*
What is evident is that our process of manual tallying of delegate ballots
has not just the potential for a lot of errors, but there actually are a
lot of errors. In at least one instance this year, it impacted the outcome
of an election, with someone being declared elected who did not actually
have a majority.
Some of the errors from the teller team seem to have been introduced by the
process of hand-copying numbers from the state tally sheets to an interim
handwritten table, and then using the handwritten table as the input for
the totals spreadsheet. It seems there were mistakes with interpreting
handwriting on that interim table. The interim table had no convenient
place to note write-in votes. In one instance, it seems likely that the
Arabic numerals representing an eleven were misread to be two tick marks,
so a 2 was recorded instead of an 11, though there were clearly 11 tick
marks grouped beside that Arabic number. These are just the general nature
of human mistakes. It’s what happens in high pressure, fast-paced
situations.
I have some ideas about how to improve the accuracy of our elections, and
I’ll certainly be discussing that with the next Bylaws and Rules Committee.
Electronic voting would be my preference, as it would improve accuracy and
the speed of the tabulations. If we can't get that, there should at
MINIMUM be a MANDATORY on-screen review of the state-by-state results for
each set of election results.
Perhaps we should also require that each delegation use ballots for their
internal voting, and that each delegation's tallies then have to be
double-checked by a delegation chair from another state before they are
turned in. My other ideas are still conceptual, and they will need some
details fleshed out at some point.
Again, I am not pointing out these problems to criticize the good-faith
efforts of any individual to do their job well while tallying the results.
My tallying team made mistakes in 2012 also. It’s much easier to find the
mistakes after everyone has gone home when there are no more time
pressures. Knowing that the problems exist is just a necessary first step
to actually finding solutions for the problems for our future conventions.
Alicia Mattson
LNC Secretary
