Making Elections Smarter

Making Elections Smarter

[MUSIC PLAYING] SPEAKER: It’s my pleasure to
have Aaron from the center– the director of the Center
of Election Science, is that right? Executive director. Yeah, their organization
does a lot of great work. I’ve really enjoyed following
them and getting involved how I can, trying to make our
elections better, and more civil, and just make
smarter decisions about how we govern ourselves. With that, thanks for coming. AARON HAMLIN: Thanks man. So today I’d like to talk a
little bit about what exactly a voting method is,
and why that matters, and to give an overview of our
organization at the same time. So, our organization,
to put it simply, we study and advance
better voting methods. When looking back at the
origin of our organization, we were rather agnostic
about the voting method in terms of which one to push. And also even now the voting
method that makes sense is largely dependent
on the context that we’re working to
implement the method. So largely right now we also
look at [INAUDIBLE] kind of because it fits
this warm zone in terms of not being too complex,
while at the same time doing a good job in terms
of maximizing utility for the electorate. So how do we go about
doing what we do? We have three branches
within our organization. We conduct research. Often this is polling research. We did polling in
the 2016 election. We’ll also do that again
in the 2020 election. We take a different approach
for educating the public. This can be through our
website, through public talks, through the types
of advertising. And then we also coordinate
with other organizations to advance initiatives,
particularly ballot initiatives. So what is a voting method? A voting method
is the information that you put on
the ballot and what is done with that information
to compute a result. And then you have
a winner– could be one winner, could be
multiple winners depending on the particular voting method. Why do voting methods matter? Voting methods
matter because they determine who gets elected. They can determine who decides
to run in the first place. It can determine
what kind of support a particular candidate
is gauged at. And that, in turn, can
determine whether or not specific candidates get
airtime, whether their views are taken seriously. It can even determine
within the people who win whether they
take on certain causes that they may not
have otherwise taken. And of course, when we’re
looking at voting methods it determines who
sits in the seats that decides very important policy
issues whether we’re talking about environmental
issues, health care, and of course spending
vast amounts of money. So it’s important that
we use a very good voting method to decide who
sits in those seats to make those decisions. So when we’re looking
at a voting method, we have to look at all these
different types of criteria kind of like a voting
method has a job to do. So when we’re looking
at a voting method, a voting method has to do a good
job at electing a good winner. If you can, you want to minimize
complexity to the degree that that’s possible. And even– so a voting
method in itself, sometimes multiple
voting methods can choose the same winner. But it’s important that even
when a particular candidate is selected as a winner
that their support is gauged accurately. And that’s important so that
when new candidates come and bring ideas to the
table that they’re not marginalized unfairly. That even if they don’t win that
their support is accurate so when they do bring
these ideas you can see the public’s support
of those ideas, as well. We want a voting method
to encourage competition. So if a voting method
does something like tells people that they’re going
to spoil an election if they run that’s not a very good thing
because that can discourage new people from running. We want a voting method to
be as simple as possible. So when we’re looking at things
like precinct [INAUDIBLE],, we would like a
voting method to be able to tabulate its results
at the precinct level and then take those totals
and add them up at a group aggregate level
rather than, say, forcing all the ballot
tabulation to be done in a single location. To the degree that you don’t
have to add fancy software, that’s helpful. And when we’re talking about
different types of methods such as proportional
methods we want to make sure that representation
is there for everybody. So the bad news is that
the current method that we use for voting is terrible. This is our choose one voting
method called plurality, or first past the post. Has a whole laundry list of
things that are wrong with it. So because you can only
choose one candidate, the vote can divide
between similar candidates. There is an issue where it can
cause more extreme candidates to win by squeezing out
centrist candidates. You can’t always vote
for your favorite if that candidate is perceived
as not being very viable. You may instead choose
one of the front runners so you don’t waste your vote. And because candidates are
worried about being a spoiler, it can discourage some
candidates from running. And also when good
candidates do run, if they’re not perceived
as being viable even when they bring good
ideas to the table, they can get an
inaccurately low measure of support, which can in
turn have them marginalized by media and other outlets. And this is terrible for
third parties and independents because these are exactly
the types of folks who are considered spoilers that
can be dissuaded from running and marginalized. And to give more
credence for the way that our current
system is running we see that average citizens
play very little role– statistically nothing at all– in terms of being able
to have their preferences be predictive of
actual policies. So the tool that
we have is just not very effective for
the average voter. Also it’s clear that the
electorate as a whole has preferences in terms
of certain people having a shot at getting elected. So most people say that
a third party is needed and that plurality of
voters are independent rather than
identifying explicitly for Democrats or Republicans. And yet when we look at these
lines that are practically flatlined on the
bottom of the screen, that’s looking at the actual
registration for third parties and looking at the
number of third parties who are getting elected. So we see this
enormous discrepancy, and this is because of our
choose one voting method. So the method that
we largely advance is a method called
approval rating. Approval voting allows you
to select as many candidates as you want. The candidate with
the most votes wins. So it’s very simple. And approval voting does a
lot of nice things very well. So for instance, attempts
to elect consensus winners, you can always vote for your
honest favorite, and this one, I think, is something where
we maybe take this for granted as a gimme, but it is
not a gimme at all. It’s actually very difficult
for a voting method to always allow you to give
your full expression of support for your favorite candidate. It actually turns out that’s
a very hard thing for a voting method to do, but approval
voting does that very well. In fact, it does it perfectly. And you don’t have
to worry about– approval voting does
an excellent job at mitigating vote splitting. So if there are multiple
candidates who are similar, you can support multiple
of those candidates rather than picking
only one so that it avoids that dividing of votes. And voters are able to
express a more fuller opinion within their ballot
rather than just choosing one candidate. And when you look at
candidates who don’t win, it’s important that they get
a more accurate reflection of support as well,
and approval voting does a great job at making sure
that that support is reflected. And the reason for that is
for a couple of reasons. One is because you can always
support your honest favorite no matter what so you’re
not afraid of expressing that information and
offering that information in the first place. And the other part is– and this is something that
can be easily overlooked– with approval voting you’re
just tabulating all the votes for all these candidates. And so you’ve given
all this information and the algorithm, which is
just adding up all these votes, is using all that
information simultaneously. So all the information
that you’re providing is being used at the same
time, and you see that all at once which is not necessarily
the case for other methods. So when we’re reaching out and
trying to spread better voting methods such as approval
voting, what we do is we look at particular
target cities. Oftentimes advocates from
within individual cities reach out to us. We work with them– if they’re not already– to become a 501(c)(4), which
is a certain type of nonprofit that can do more advocacy work. And we tend to prefer that
they have experience already just because it makes the
partnership a lot easier to work with. And then the Center
for Election Science runs education campaigns
and does outreach alongside the other
organization which focuses on the direct
advocacy, and then also making sure that they
have strong relationships with key stakeholders
within that community. And in terms of the outreach,
we do this in all kinds of ways through public forums, community
events, educational literature, radio, and digital advertising. In fact, we did
this within the city of Fargo, North Dakota,
which is where you’re seeing some of these pictures from. Fargo was a city
of 120,000 people. They had some issues
with their elections where there were five
or six people running and the candidate won with
something like 22% of the vote. It was just ridiculous. And even worse, that was not an
anomaly within their elections. They’d had this sort of
issue come up before, and so they created
a task force. The commission for their
city ignored the task force that they created when
their task force recommended approval voting. And in fact, they did
that for over a year which was even more embarrassing. And so one of the individuals
on that task force had reached out to us,
and we worked with them. They created an
organization local to Fargo, which was called Reform Fargo. And we worked with them in order
to help them run this campaign. And this is a quote from one
of our campaign coordinators who was on the ground in Fargo,
who admitted to being motivated to the fact that she would
be able to make history in having approval voting– have the city of Fargo to be the
first city in the United States to ever use approval voting. And we actually were
successful in that campaign. And Fargo, North Dakota
in 2018 last year– that’s this just past November–
became the first city ever to implement approval voting. And it will be used in the first
time in the general election in 2020. So other ways to learn
about what we do. We have our website, You can find information
from blog posts. We do some technical
analysis within those posts, oftentimes looking at
government elections, but collective
decisions take place in all kinds of
different contexts. So often we get
creative in terms of looking at different ways
that collective decisions take place, often
organizations with awards. And we’re also a good
resource for other items like electoral systems glossary. We’ve got a ton of terms
within voting methods, and so we try to make
that a bit easier and digestible for folks. And in terms of
how you can help, we have a toolkit that folks
can download off of our site. Different ways to get
involved such as volunteering. And a lot of this
information is something that is not common knowledge
for people, as well. So being able to follow
us on social media, we’re on just about
all of them, and being able to share our
work is very helpful. And of course, the
reason that we’re able to do the
things that we do is because of generous
supporters who are able to give and
support our cause. SPEAKER: Looked like
Sunnyvale had a few questions if you want to ask
live, otherwise I can read it off the dory. AUDIENCE: Should be good. AARON HAMLIN: Can you hear us? AUDIENCE: Yes. Hi. I’ll pull up my
list of questions. Hi, Aaron, it’s Greg. I’ll start with the fun one. So I’ve been an approval voter– [INAUDIBLE] AARON HAMLIN: Oh, hi Greg. AUDIENCE: Hey. So I’ve been an
approval voting fan since before CES existed, and
I remember when it was founded. And I have yet to have any swag. I have no t-shirt from which I
could advertise my affiliation with CES or approval voting. And I saw those
wonderful Fargo pictures that at least for that
campaign there was. So is there a section– I’d be quite happy
to buy this stuff, but is there a section
that CES has or is planning to have just to sell
some apparel so people who want to support this stuff
can let themselves be known? AARON HAMLIN: Yes,
we do need to– we bring things to
different events. For this one I don’t
have any t-shirts, I think I do have some chick
clips for folks that are here. But our operations
person is working on making it easier for folks
to be able to buy apparel. So we have heard you,
and we are working to make that easier
for folks to be able to show their support
in a very visible way. AUDIENCE: All right,
I’m looking to see what else I have in the dory. I guess I’m going to
monopolize a little bit. Do you have a wish
list of what you wish that either Googlers
who are sufficiently engaged or Google as a whole
company would do in order to help support better election
methods and decision making, especially if that looks
like a product feature? If there’s some gap in our
products where you’re like, if only there was a
voting system right here and it was good. AARON HAMLIN: I think
that being able to have like a really clear
app for group decisions is pretty helpful. There are some out on
the market already like– I’m trying to think on the
one that tends to let you pick a date in front
of– like on the– SPEAKER: Doodle, or– AARON HAMLIN: Doodle, yeah. Doodle polls are really good. But providing folks
with different ways to make this quick decisions– although there is
some of that in there. Really anytime
internally whenever you’re making collective
decisions don’t use plurality voting I think is a
good mantra, and being able to use alternatives
such as approval voting. When using small
groups, score voting when you score candidates
on a scale is a good idea. And one of the technical
reasons for that is when you have a
small sample size, you have more random
error, score voting has more sensitivity
on the scale and sensitivity with your
measurement instrument will reduce random error so you
get a more accurate result when you have a small sample size. And of course,
with Google having matching with gifts, that
always goes a long way in terms of making sure that
we can get work done, we can spread this
to other cities. AUDIENCE: So how
about a segue there? So I know you have a campaign
in Fargo that was successful. I haven’t heard about what
other cities you are– are prospects right now. Anything to tell us? AARON HAMLIN: Yeah,
so we’ve got– we’ve been working
with a group recently, and we’re just
firming up our MOU, and they’re applying
for a grant with us. We haven’t announced
it publicly, but the city is over 300.000
that we’re looking to hit next. And we’ll be making
that announcement within the next few weeks or so. So that’s your tease. SPEAKER: I see there’s
a question from Mountain View from the Livestream. So you mentioned
approval voting is one of the few voting
methods where it’s safe to vote for your favorite. How should a voter choose
which of their second favorites to also approve? Like what’s the
kind of algorithm to pick a good threshold
for approved or not? AARON HAMLIN: So
tactically there are a number of heuristics
that you could use. You could say among
the front runners who do you like as
your favorite candidate among the front runners? Support that candidate
and everyone else that you like more than
that candidate, which would include other people in
maybe the middle of the race who are moderately
competitive, and then folks who are not
competitive at all. That might be kind
of a quick heuristic. And then if your– another issue may be
if, say, maybe you’re a little bit on the
left, and there’s a candidate in the middle,
and a candidate on the right. And then it’s kind of
a tight three way race, if you would find that
person in the middle acceptable it may be
appropriate to support that candidate in
the middle just to hedge your bets
against the candidate who is on the far right. Or if you find yourself
maybe center right and there’s a candidate on
the center and the left, and you would find the candidate
in the center acceptable, and you really don’t
want the candidate more on the left winning, then it
would also make sense for you to support that
candidate in the center. AUDIENCE: In Fargo did
you have to educate the population on how to
use this type of voting? Like basically what
you just told us now. AARON HAMLIN: Yeah, so
with the city of Fargo we did a lot of media outreach. And then getting people to
come to our site to learn more about approval voting. So the education component was
a large part of that outreach. We had public meetings. There were a number of different
approaches to get out there. AUDIENCE: I know
gerrymandering is a big issue. Has this been studied
with that issue? Like is there any
research there saying this will help mitigate that, or– AARON HAMLIN: So
there’s no research looking at approval voting
relative to gerrymandering. Approval voting tends to elect
a more consensus style winner. So even if– so if
you were to, say, to take a single member
district and gerrymander it, you would still push even the
approval voting winner over to whatever, say, the median is
for that particular electorate. If you don’t like
gerrymandering, which I don’t know many people
that do other than people who were elected, then the way
to go and circumvent that is to use small
time no districts and to use a proportional
voting method. And there are proportional
versions of approval voting that that can be used. So oftentimes people
look at it and say, well, maybe I’ll get this
computer algorithm to draw some lines using some
metric that’s not associated with partisanship, or we’ll
get this independent commission together and they
would draw the lines in some way that is nonpartisan. But there is an
issue when you take– it’s kind of like a really
bad sampling technique when you do winner take all. Because the surplus
doesn’t go to anything when you get more
votes than you need, and when you get
less votes– when you get fewer votes
than you need then those don’t do anything. And so you have all
this waste occurring whenever you’re doing these
winner take all elections. And in fact, Canada has
used independent commission since 1964, and two of its past
recent elections that had what are called false
majorities which is when a party can have
less than half of the vote and they get more than
half of the seats. And it went both ways. So in one election, it didn’t
work out for conservatives and it benefited the liberals. And then another election right
after that it was the opposite. And so you get these really
kind of blatantly unfair results from using these winner
take all approaches, and you get around that by using
a multi-winner proportional method where if you’re going
to do it still district based, you want those seats to be
of at least five or more. And as the number of seats
that you elect simultaneously increases, so does
that proportionality. AUDIENCE: I have a
live question here. So with choose one
voting, whether it’s like Washington’s top two to
be in a primary or parties doing national presidential
primaries and things like that, what role do you see
for a primary type thing combined with approval voting? Could you see throwing
just literally everyone into one giant bag
and then they’ll fight, or some kind of
weeding down process, or what do you think about that? AARON HAMLIN: That’s actually
an issue that we directly had to contend with. Because when we’re looking
at different jurisdictions to collaborate with and
implement approval voting, you’ve got issues like you
have to have ballot initiative, if you’re looking
at cities you have to have a [INAUDIBLE] state
so that the city doesn’t have to get explicit
permission from the state. And another issue
that we come up with is that the state itself will
require that all of its cities do something like use priority
voting for the office. So they’ll say that you
can’t choose more candidates than there are
seats to be filled, or they’ll say something like– they’ll make a
majority requirement, explicitly require a runoff. In Texas, they’re
so strict about that that they don’t even count
instant runoff voting as meeting the
majority requirement. So the way around
that for both of those is to have an open primary. It could be partisan where
people can throw the letter by their name but still have
it be open so voters can choose anyone that they want or
you can have a nonpartisan where there are no letters
by these candidates’ names. And then from there, if you’re
required to use plurality in the general
election or if you’re required to get a
majority, the way that you would satisfy
both those criteria would be to have just two people
go on to the general election. And so in the general
election you’re using plurality voting,
which is fine in this case because only a few
people are running. The voting method
really only matters when you have more than two choices. When there are only two choices,
plurality voting is fine. And so it addresses all
the vote splitting nonsense that would otherwise occur
under plurality voting for the primary. So you can use approval
voting for the primary so you’ve got this long
list of candidates. Approval voting goes
really well with that. It’s really easy, even
with a long candidate list, to use approval voting. And then the top two would go
over to the general election. SPEAKER: Looks like we had
similar questions in Sunnyvale. AUDIENCE: Yeah, it’s
supposed to touch on the– back on gerrymandering. So I agree with what Aaron said. I do want to point out that
it’s not perfect by far, but algorithmic redistricting
or redistricting commissions are better than
the lack of them. And there are lots and lots
of algorithmic proposals out there, and one of them
is by a [INAUDIBLE] named Brian Olsen who’s [INAUDIBLE]. And those tend to do pretty well
unless you very strongly care about the idea of having a
community cohese together in a district as opposed
to a district having somewhat arbitrary boundaries. AARON HAMLIN: So with
different kind of line drawing algorithms, so the idea is
that they are not really a solution because
of just like– they’re using this inherently
bad sampling method. But there’s certainly
worse, which is what we’re doing right now. So there is a terrible
approach, which is allowing a particular
party or group to leverage the greatest
review possible to give them as much power as possible. So that would be the worst end. But even when you
have, for instance, independent commissions
like in Canada, they’re the party that got
more than half of the seats. In that case, they were winning
that, but with just a sliver less than 40%. And so that’s still a
pretty egregious result, and that’s just–
inherently that’s due to this inherent
vulnerability when we’re using these
single member districts. So you can mitigate
that somewhat, but you’re still
in a bad situation. AUDIENCE: For those of us who do
live in plurality voting areas, do you have advice
for how voters can avoid the pitfalls
of plurality voting? AARON HAMLIN: Yeah. So move to Fargo. [LAUGHTER] It’s really
a bad situation. I mean, you can do different
types of protest votes, write letters to the editor. Protest votes like
vote for somebody that you really like even
though they’re not going to win, but you’re doing that at
the cost of not having a say in the outcome of the election. Yeah, it’s really a
bad situation to be in, and it’s one that
almost everyone is in. SPEAKER: Related question on the
dory from Joseph in New York. Changing the voting system is
kind of against the interests of the people
currently in power. So how did something like
the Fargo reform happen? AARON HAMLIN: So
the way that we did that is we didn’t ask
the people in power, we just ran a ballot initiative. So you circumvent those people
that could say no to you. How do you take care of that? Although that strategy
may change over time. For instance,
another voting method is the runoff voting, which
we’re not as excited about. With that one, there does seem
to be more legislative track there. And part of that is it
having more of a history. So right now, the door
to approval voting being implemented, there
really isn’t a very good door in terms of legislators saying,
OK, this is a good idea. You should be using
this, we’ll vote on it, and then all of my colleagues
will also vote on it. That’s something that’s
really an option right now. But over time, that
may be something that opens up,
particularly as you start to get closer
to a tipping point with more people
implementing approval voting. But we’ve got a lot of
work to do before we’re close to that point. SPEAKER: Another
question from New York. Why approval voting
instead of something with a more expressive– like score voting where
you pick a number of stars? AARON HAMLIN: So one
thing to keep in mind when thinking about a voting
method is that it has multiple components. So the expression
element is one component, but there’s also the
calculation element. Score voting, so– and there
are other voting methods that allow you to
provide more information. So ranking methods let you
provide more information, score voting lets you
provide more information. The issue with– and I’ll
go to score in a moment– the issue with ranking methods
is the algorithm that’s used for a particular method. So for instance, with
instant runoff voting even though you’re explicitly
providing more information the algorithm
itself may actually ignore some of that information
during its calculation and can allow a number
of anomalies to occur. So we can’t ignore
the algorithm and just look at the expression
element in isolation. With score voting,
score voting is, I would say, in terms of both
expression and the utility that it offers, is better
than approval voting in terms of the outcome that you get. The reason that we push approval
voting over score voting, at least right now, is
because it is just so easy. It is really, really easy. You just change the
directions from choose one to choose all the
candidates that you want and even the dumbest
of voting machines can figure this out. So it works out very well. And so when looking at– one of the jobs
of a voting method is pick a good winner
to offer high utility to the electorate on average. And so you do get a
bit of a utility gain from going from approval
voting to score voting, but there’s also
that complexity cost. And the perceived
complexity cost can be high given that
most people don’t even know what a voting method
is, that it’s not even a concept on their radar. And so going from this choose
one method to score voting is a high complexity
cost for those folks. So the range of acceptance
can be more challenging there. And then also you have
logistics for voting machines in some cases. And so the difference
in utility that you go from approval voting to score
voting, that’s not nothing, but it’s not a huge amount. What is a huge amount is
moving from plurality voting, or this choose one
method that we have now, going from there to
what we have now. That is a large amount. And so we’re focused on
getting that enormous gain before thinking about squeezing
out the last bit of utility that we can. Because we can take
this even further. We can say, OK, well,
there are other variants of voting methods that
give even more utility. But the question is,
how much complexity cost are we willing to pay for that? And right now we pay really a
token complexity cost in order to get an extremely high utility
gain from the current method that we’re using now
to approval voting. AUDIENCE: I had a question
that follows on from that. That was one of my questions
is the use at adapting existing systems, whether they be
ballot marking devices, or paper forms, et cetera,
to this new system, it seems like a very
small logistical change. Is that actually
borne out in practice? AARON HAMLIN: So before I
can say that explicitly, I think we’d have to wait
until next year in Fargo. But I don’t expect
there to be any issues. When talking with
folks on the ground there who also communicated
with the people who are actually running the elections
in Fargo and they don’t expect to
issues with this method. And to kind of pull
that out a bit more, there are voting methods that we
use right now that allow people to pick multiple candidates. You look at things like
block priority voting when they’re selecting
a council, for instance, when you could pick
as many candidates as there are seats to be filled. Some cities use
cumulative voting, which is a semi-proportional method. There you have as many pips or
votes as there are candidates, and you can stack votes
on individual candidates. And so those are our
methods that we already see in practice. And of course, you look at
cities that use [INAUDIBLE] runoff voting, and that’s
a way more complex method. AUDIENCE: Yeah, it
does appear to be more complex at first glance,
the instruments in runoff, and cities are using that. So that’s attractive. It would be great to see what
the experience is and have that as information that people
can use when they’re trying to persuade their locality. AARON HAMLIN: I’ll be right
there with you watching. AUDIENCE: One thing that comes
to mind I was thinking of, there seems to be– it’s almost with
the approval voting that voting is almost more
diffuse in that you would have kind of the left wingers will
vote for the left candidates, the right wingers will vote
for the right candidates, and those sorts of things. But does that open
up an opportunity for some special interest
groups, like say a single digit percent of the
population that wants to vote in a radical
candidate, so they only vote for this one candidate. And then their votes are very
targeted whereas everyone else’s votes are
somewhat diffuse? Do you see something
like that happening? AARON HAMLIN: Not unless
the radical candidate has some consensus issues that
they’re able to rally around. So this particular
voting method really does tend to elect more
consensus style winners. And data that we’ve seen,
like for instance, there’s one French election here
if you look at the graphic to the right, Bayrou was more of
a centrist moderate candidate. And here in this
election that candidate was the one that was preferred
over both the candidate that was more on the left
and the other one that was more on the right. And you can look at the third
party candidates in here, too, and there you’re going to
get more fringe opinions. And while they got more support
than they otherwise would, they still weren’t
in the position to take over the lead
relevant to other candidates. And this is also a really
nice visual in general. You can look at
the German election over to the left of that, too. Here you see that even when
the winner doesn’t change, that reflection of
support that you get for third party and
independent candidates is really really
quite significant. Here’s another peek
from the 2016 election. So this is looking
at a poll that we had done looking at approval
voting and plurality voting. You see enormous differences in
terms of the amount of support that candidates
got under approval versus plurality voting, even
when the winner doesn’t change. AUDIENCE: Does it appear
that approval voting or any other of these non pick
ones has any impact on turnout? AARON HAMLIN: It’s
a little unclear. I’m not sure with the
IRB election so far. There does seem to be a
role with the voting method and turnout with
countries that use proportional representation. But it’s maybe a bit
unclear why that is. It could just be that their vote
is more effective, that they’re more likely to get
someone elected, that their vote is
more meaningful. And to the degree that
that’s salient to voters under approval
voting election, it’s quite possible that that
would increase voter turnout. The voter turnout is kind of– that’s like a really kind
of popular issue or metric to point to. Perhaps because it’s
like a salient thing, it’s very easy to measure. When looking at voter turnout,
I look at it a couple of ways. Like one, maybe it’s
kind of a proxy measure for voter engagement that
they believe in the system. But in terms of it being
important for the outcome I look at it a bit more
from survey sampling. So when we’re interested in
what a population thinks, we don’t need to
ask every individual within that population so long
as the sample that we take is representative of the
population as a whole. And so to that extent I’m not
quite as worried about voter turnout, although
what does worry me is when the sample
that we’re taking or the people who
do go out and vote aren’t representative of
the population as a whole. When you see certain segments
either not going out to vote or being blockaded from
being able to vote, those are real issues. AUDIENCE: So for instance,
polling likely voters versus adults. That would be like your
discrepancy in sampling. AARON HAMLIN:
Correct, that’s right. You want those groups
to be the same, but that’s not always the case. AUDIENCE: Would you accept some
speculative commentary on that? Because I’ve spent a lot of
time thinking about this. This is going to be my
perspective, all grains of salt. But if you look at
the system right now in the US, and we have the two
dominant parties, Democrats and Republicans. They’re basically the only
ones that win elections, at least in large scale. Maybe you get an occasional
local candidate or something, but it’s the two party
system– it was called that. We all know that it’s
a two party system. That comes from the
plurality voting method that helps to reinforce this. Like you look at the chart up
there and you see third parties at 1% and 2% when
really they’re– even for this one
specific poll where we’re talking about 20
points or something. And I think this gives you
an iterative effect that squashes those numbers down even
lower than they would otherwise be because that 2% is the 2%
of people who voted anyway despite the fact that everyone
knows they’re not credible. If you started showing people
that those 12% and 20%, I think those would
amplify over time to something a lot less pressed. And I think one of the reasons
we have really low turnout in the US is because
our government is really ineffective, and it really
doesn’t implement things that have wide public support. And anything that we can do to
improve the quality of voting method a little bit is going
to make it much more likely that we’ll have a more
effective government that actually does what people want. And I think we’re in
a suppressed state where it’s just bad enough
that people rationally, to some extent, see that there’s
no reason for them to engage because they’re going to get one
or two choices instead of what they actually want. And I can see voter turnout
dramatically improving if we do even just a little
bit to break a two party domination, which is one of
the reasons I got excited about the cardinal methods
like score and approval voting. Because instant runoff
voting doesn’t seem to break two party systems. You can look at that in
the Australian government. They’ve been using instant
runoff for over 100 years and they’ve had domination
by two major parties in almost all seats
most of the time, except for a couple of years
when they were swapping parties earlier in the century. SPEAKER: Another
question from the dory. Basically, are there
easier ways to experiment with voting methods? Are there jurisdictions where
you don’t need a whole ballot initiative to change it? AARON HAMLIN: There
are a few ways to do it for
government elections. One of them is
ballot initiative, which means you got to
go out there and get a bunch of signatures. You can do a referendum,
which you somehow convince the council
or the legislature to say that’s a good
idea, and then they put it for the voters, and
then voters vote on it. And then sometimes
they get really lucky, they can just pass
a law outright to change the voting method. But you’ve got to have the sun
really shining on you that day. If you like, I can address the
issue of third parties actually gaining seats. So with Michael’s
commentary, you can expect kind of a feedback
loop where support can grow. One interesting thing
about this figure that’s on the screen now looking
at the third party support, what that– there was a Gallup poll
earlier in the year that said that 2/3
of people didn’t even know who Stein or Johnson were. So imagine an election
where they at least get that much support. Like 12%, I think there are
different ways of cutting that data because
of the weighting, but it may be as low as 9%. But still, quite a lot
more than the 1% initially. You can imagine
the media not being able to marginalize
the same way, and then having to cover
them or getting into debates and then seeing
that support grow. And then the other
issue in terms of third parties
getting actually elected using alternative
voting methods. There is one approach
to looking at that, or framework of looking
at that [INAUDIBLE] law. And [INAUDIBLE] law looks
at a couple of factors. One is the threshold of support
needed to get someone elected, which under a single
winner voting method it’s always going
to be high, which is more support than
any other candidate. And even approval voting or
any kind of cardinal method– any method, at all, whether
it’s a single winner method is going to have issues there. The other factor is
being able to support your honest favorite
candidate no matter what. And so that’s really where
within the single winner framework that approval voting
can have a shot at getting third parties elected. And what I don’t expect to
see even with approval voting is the extent of third
party representation that you see in
countries like France or other countries using
proportional representation. So you would get more
than we have now, particularly for [INAUDIBLE]
would be with executive seats. But in terms of getting
the type of representation for third parties that you’re
seeing in other countries, if you really want
that then you need to use a proportional method. This will only get
you partly there for electing third
parties and independents. But if you want to
see a lot of that, like four or more active
parties within the government, then you need that
proportional method. SPEAKER: Yeah, another
question from Doug in New York. Are we going to
end up with ballots that are hundreds
of candidates long, or is there a way
that makes sense to limit to a reasonable
number of candidates? AARON HAMLIN: So that’s more
of a ballot access issue. And right now the US has some
of the most draconian ballot access laws in the entire world. But say part of the
reason for that is that– so we know that
the voting method doesn’t matter so much when
you only have two candidates. And from the people
who are elected, the way they look at it
is, well, we could either use a sensible voting method
or we could just make sure that nobody else runs. And they’ve chosen the latter,
which is why our ballot access laws are the way that they are. So we really got
that as an issue. Like the ballot
access laws are so high that even under
approval voting right now, it would take a
concerted effort in order to get people on the ballot
in some jurisdictions. But that may wait a bit in
terms of the that hurdle. It’s likely to go down as
the spoiler effect is not so much of an issue or
an incentive for people who are elected to create
these crazy ballot access laws. But over time, it
may be that there are a ton of people on the ballot. Like maybe you don’t want
50 people on the ballot. Maybe that’s too
many And you have to figure out what types
of fair barriers that allow a reasonable number
to be on that ballot. But even if you have a bunch
of people on the ballot, one of the nice
things about approval voting really more than about
any other voting method is, because the
expression element is so easy in terms of picking
it’s going to be candidates that you want, it
really lends itself well to long ballot lists. Which even if you have really
generous ballot access, it would work well
in that context. And you can always
narrow it down. So you can use, say,
an open primary. And then you don’t have
to do it to a top two depending on the
jurisdiction that you’re in or whatever the state laws are. You could have, say,
a top four, or top six in the general election. So you can have that
crowded field in the primary and then take it down
to a number that’s more manageable in
the general election. SPEAKER: Well, we’re
about out of time. Do you have any closing
thoughts for us? AARON HAMLIN: Thanks for
offering this opportunity, and I would encourage
folks to go to our website at We have a very non-pushy
newsletter that we don’t– that we are mindful of our
subscribers’ time with. And we put out lots
of great stuff there, and we encourage you all
to provide us with support by sharing it with
your friends and family in terms of this information
because this is something that most people
are not aware of. And I also believe
that Google has matching for charitable gifts. I would encourage that, as well. Help us keep up the
good work and help you to be able to see the
outcomes that you want to see. SPEAKER: Thanks for coming. I can plug the newsletter, too. I really enjoy those
emails, and it’s always great content that’s really
well thought out and very often quantitative like this
to show how would things be with a better system. Thanks for coming to Seattle
and giving this talk. AARON HAMLIN: Thanks. [APPLAUSE]

Danny Hutson

10 thoughts on “Making Elections Smarter

  1. As long as they don’t support these electronic voting machines without paper ballot accountability!

  2. Reset PW
    List + Id confermation

  3. I thought the title was "Making Electrons Smarter". I thought this was gonna be some deep talk about how to push back against entropic forces. Let me know when you Google people figure out how to make electrons smarter and I will be your biggest fan.

  4. DO NOT DO ELECTRONIC VOTING! PLEASE NO!!! There are so many potential problems, with the primary one being the point of failure being centralized!! Electronic voting is absolutely a no-no!!!!!!!

  5. Up until you mentioned "electing a 'good' winner". What is YOUR idea of a good winner; I bet it is not mine. You lost me early with this bullet point. If you have a solution offer it up at election time as a straw vote (just a vote, no impact) to test your system. I'll bet it will match exactly because you or someone like you will be running the straw vote.

    Please return to your academic haven and SHUT UP.

  6. Electonic referenda are necessary. First for local issues. Paul Cockshott has already developed the technology

  7. Google should stay out of elections and be a platform not a publisher if they don't the should be regulated like a publisher, stop meddling in our election.

  8. Thank you Aaron for coming out and giving this talk.

    Note to the other commenters: Tech talks are organized by employees, not endorsed by Google, the company.

Leave a Reply

Your email address will not be published. Required fields are marked *