Cookies   I display ads to cover the expenses. See the privacy policy for more information. You can keep or reject the ads.

Video thumbnail
My name is Daniela Rus.
I'm the deputy dean of research in the Schwarzman
College of Computing and the director of CSAIL.
And I'm very excited today to welcome to our "Hot Topics"
Professor Charles Stewart.
Professor Stewart is the Kenan Sahin distinguished professor
of political science at MIT.
He received his undergraduate degree
from Emory University and his PhD
in political science from Stanford University.
Professor Stewart works in congressional politics,
elections, and American political development.
His current research about Congress touches
on the historical development of committees,
origins of partisan polarization,
and Senate elections.
In 2017, Professor Stewart established the MIT Election
Data and Science Lab, which applies
computational principles to how elections
are studied and administered.
In 2020, he partnered with Professor Nate Persily
of the Stanford Law School to establish the Stanford MIT
Healthy Elections Project.
And this is what we're here to talk about.
Now, Professor Stewart is a member of the American Academy
of Arts and Sciences and a beloved teacher of the MIT
students, recognized with a MacVicar Award,
a Baker Award for Excellence in Undergraduate Teaching,
and many other recognitions.
Now we have muted all participant microphones
in order to keep the speaker clear for those listening.
During the Q&A portion, please submit your questions
directly to Jill, Laura Lynn, or me, utilizing the chat feature.
To open your chat feature, just click the chat icon
at the bottom of your Zoom screen.
The chat will open on the right.
Charles, there's been so much happening these past few months
that is impacting voting.
And I'm so eager to hear from you.
Let's give professor Stewart a warm virtual College
of Computing welcome.
Charles, take it away.
Thanks, Daniela.
It's great to be here.
It's great to see so many people who I know.
I wish we all could be together.
One of the things I like about MIT
is that, though I am a political scientist,
in some ways on the edges of the empire,
I can look at my screens and see a number of people
that I've been co-conspirators with over the years,
in both voting and curriculum and policy
and all sorts of things around the Institute.
So it's great to see people today.
I am going to be talking about the coming election.
And I'm going to be giving a talk that
is a little different from what I would normally give
with an academic department.
And I hope you will indulge me.
I've been giving a lot--
this is the 10th talk I've given in the last week and the fourth
in the last two days on this topic.
And so I will be talking at a general level
about what I've been up to over the last seven months.
But I hope this will provide an opportunity
to key up discussions on any dimension you
would like to talk about the upcoming election.
So the title of the talk is, "Will the Election
Be Safe and Decisive?"
These are two questions that have
been at the top of people's minds-- safe,
since the pandemic arose back in March,
and decisive ever since people began to worry about
whether Donald Trump, were he to lose,
would actually go out willingly.
The quick answers are yes, it will be safe.
And only ironically it will, I say,
it will probably be decisive.
And I'll say some words about both of those.
But before that, I want to-- this is my agenda.
And before getting to the questions of safety
and decisiveness, I wanted to tell you
about the Healthy Elections Project, which,
as Daniela mentioned, I'm a co-directing with Nate
Persily, who's a law professor at Stanford.
So the Healthy Elections Project,
you can find out everything that we're
doing at
This is a project that started up in March, right
as the pandemic began to hit.
Nate and I both had been supported by the Hewlett
Foundation in our work for many, many years,
working in election administration.
Nate also directed the Cyber Policy Center at Stanford.
Nate and I began working together in 2013,
when Nate was the research director
of the presidential commission on election administration,
which President Obama appointed after the 2012 election
to deal with issues like long lines
and emergencies and the rest.
Nate asked me as one of the co-directors
of the Caltech-MIT voting technology project
to put together a team of experts
to give testimony and advice to that nonpartisan presidential
commission, which, by the way, as an aside, I would argue
is probably the greatest-- probably the best
example of non-partisan work in the election
space over the last couple of decades, the way
you should do election reform properly in this country.
So, I encourage
you to go to the website.
You can even go during the talk, although I'd prefer
if you went after the talk.
So here's the idea, though.
So Nate and I got experience working with the president's
commission, working with election officials,
about solving practical problems in elections
and became quite trusted within a community that
oftentimes distrusts academics, and used
to work in a non-partisan way in this space.
We were aware in March that there
was a lot of litigation coming down the pike related
to the pandemic and otherwise that
was going to be driving election officials to have
to change how they managed and administered elections.
And while a lot of this litigation is important--
and it is, indeed, in many cases, appropriate--
Nate and I also knew that you cannot litigate yourself--
you cannot litigate your way to a well-run election,
that in order to make a pivot, for instance,
from a predominantly vote in-person system
to a vote-by-mail system, it's going to require a lot
of logistical--
it's going to bring up logistical challenges
that will require logistical responses,
and that there are tools that are available in the civic tech
space and universities kind of around
in this area that should be made available to election
And so we were approached by Hewlett--
come back to Hewlett--
to start a University-based program at the two universities
where we would basically assemble research
and to make it available or visible to election officials,
but also to the public, to legislators,
to the press, to help ease this movement into voting by mail.
And once we got going, we got approached
by wealthy individuals who suggested
that we expand out our vision.
And so we've actually spent the last seven months--
and I'll tell you a little bit about what
we've been doing that's reflected in the website.
We've been doing research, but we have also been supporting--
by supporting, I mean sometimes directing funds toward--
groups like the University of Rhode Island--
Gretchen Macht is an industrial engineer there
who designs, say, polling places--
to an effort at the University of Southern California
that's about placing drop boxes, folks in the Civic tech space
who've been designing ways of tracking ballots,
basically redesigning management systems--
doing whatever it takes to vote safely in this environment.
And so we've developed a number of partners who are
University-based in the NGO space, some LLCs--
a large collection of former election directors
who've been embedded in state and local election departments
to help them manage transitions, et cetera.
Now, its visible part of our work
is reflected in this website at,
where we have been posting up research that's
been primarily done by graduate students and law
students at Stanford and MIT.
And you can go there and you can check it out.
Some of that research focuses on the states themselves.
So if you are curious about how the states, and particularly
the battleground states, managed voting
during COVID and the rise of the pandemic,
we have a number of state-specific reports.
We also have a number of state-specific reports
that have addressed the particular problems that
are facing the states as they make these huge management
And so you can go there.
By the way, there's about 500 pages
of original research on the website
about these various topics.
We've focused particularly on vote by mail.
There's a number of reports written mostly
by our students about vote by mail topics.
This is the most recent one, which
is about issues related to the US Postal Service.
Working with the Stanford Design School,
we've helped to develop a number of resources,
videos, written materials, individuals who've
been trained in redesigning polling places to make them
safe to vote in given social distancing,
information about becoming a poll worker.
This is one of the areas in which we've not only
been academics but we've also been impresarios,
working to help support groups out recruiting poll workers.
Academics usually don't do this, but we've
been encouraging the recruitment of poll workers
through the partners that we've been working with.
We've been able to get about 600,000 people to volunteer
to become a poll worker.
For scale, there are about a million poll workers who
are deployed on election day.
So we've been very active on the poll worker recruitment side.
We've been very active in collecting together
in one place pointers to a variety of tools.
Some of them are web-based.
Some of them are software-based.
Some of them are practice-based.
They're related to managing mail ballots, designing polling
places, communicating with voters,
tools for analyzing data, et cetera.
Some of these applications were written here at MIT,
oftentimes by students of Steve Graves, who's
on the faculty in the Sloan School.
Other tools have been written by people at other places.
Could I ask a question?
Who is using the tools today?
Well, for instance, this morning I
was on a call with the Secretary of State of Georgia.
Georgia has been in the news.
They had long, long lines in their--
if you've been reading the news, they've
had long lines in their early voting period.
And so on that call, one of the things we talked about
was their utilization of an online resource
we have that's basically an application of queuing theory
to the allocation of resources at the polling places
and stuff like that.
And one of the students on my team
has been working with the state of Washington
to adapt similar tools to a problem they have,
which is they're going to be inundated with people
walking in and wanting to register
to vote on Election Day.
So mostly-- the ones that we know about are state officials
who are trying to kind of scramble to meet--
to basically match resources to demand for in-person services.
The ones that we don't observe directly
are when local governments use these tools.
But I'm happy to talk about that.
Bunch of other reports.
There's a litigation tracker.
If you're into--
I'm getting now to be close to 400 cases in the courts related
to voting according to COVID.
OK, so that's Healthy Elections.
I encourage you to take a look at the website.
Happy to talk about the project and what
we're learning from that.
And like I suggested, we're talking
to a lot of election officials every single day.
Is the election going to be safe?
Well, in reflecting on this question,
my mind is brought back to back last March.
Here's Los Angeles County, where I
was at the beginning of March, before COVID.
People were voting like we used to always vote,
packed close together at machines doing our thing.
But within weeks, COVID hit.
And this slide reminds us that within two weeks
of the middle part of March, although rates were low,
first and second derivatives were high.
Rates of infection were beginning to skyrocket.
And we were beginning to wonder whether or not--
some people were beginning to wonder whether or not there was
going to be an election at all.
At the time, I reminded people that because
of statute and Constitution--
our Constitution-- even if asteroids were raining down
on us and the zombies were wandering the streets,
we would be having an election on November 3.
And the design constraint was, how do we hold that election
in a way that respects the public health
crisis that's facing us?
So Ohio ended up postponing their primary.
And virtually every state from mid-March through April
postpone their primaries and held them in June--
when, by the way, the rates of infection
were orders of magnitude greater than they were back in March.
Wisconsin, we will recall, did not postpone, famously so.
It went to the US Supreme Court.
They ended up holding it in the middle of April.
A lot of confusion.
I'll talk about some particular things
that happened during that primary in just a second.
But you know, things were looking pretty bad back then.
But now we're voting.
And by the end of the summer, voting began to look like this.
This is my home precinct, socially distanced.
You know, you can see masks and shields and indicia
of social distancing.
We also see things like drop boxes.
This is Boston City Hall.
In short, we've adapted.
Now, what have we done?
The imperative back in March was to de-densify in-person polling
places for public health reasons.
This de-densification had two prongs to it.
First of all, we had to move people away
from voting in person, for obvious reasons, which
meant more mail balloting.
And that's been kind of the big management challenge.
But we've also had to maintain polling places.
While we've been de-densifying, we've
had to actually maintain those polling places.
Now, why?
It's been important to maintain the presence of those polling
places for a number of reasons.
First of all, some people--
despite the need to de-densify--
are going to need to vote in person.
Many of those are going to have language difficulties.
Some will be disabled.
Many people, as we've discovered during the fall
and as we're discovering right now,
will not get the mail ballot that they've requested.
And so in-person voting is a fail safe.
Many people distrust the mails.
There will be late deciders.
Basically, people who are deciding now--
for the next two weeks, if you decide now that you're going
to vote, you should not be voting by mail,
because the Postal Service cannot guarantee it's going
to get you the ballot in time.
So late deciders are going to need to vote in person.
There's also another thing that needs to be remembered,
that even if we were to squeeze or just
get 80% of voters out of polling places into the mails,
the remaining 20% would still clog
the capacity of the existing in-person polling
places because of the need to socially distance
those polling places.
So we would still need to maintain almost as many people
to staff those polls, and certainly as many facilities.
The challenge of reaching the imperative was two-pronged.
On the mail side, we had a number of challenges.
First of all, most of the nation lacked experience voting
by mail.
This graph illustrates one of the challenges
at the national level.
And that is that Americans basically vote in person.
We can see in 2016, for instance, roughly 60%
of Americans voted in person.
Now, if we go back to 2000 or 1996 where the graph starts,
90% of Americans voted in person.
And we've seen gradually over the last 20 years
a gradual decline of voting in person on Election Day,
while there's been a gradual increase in voting by mail
and also voting in-person before Election Day.
But the levels of voting by mail that
are needed for this election are such that when
I do this graph with the 2020 data,
it's going to look like this.
So it's going to be a radical discontinuity in how we vote.
And it's going to be a discontinuity not just
at the national level, but it's going to hit hard
in particular states.
This growth in mail voting that we've seen over the last 20
years, almost all of it has occurred in about six states
in the far West.
And so if we break down how states vote, we get ourselves
to this ternary plot that, by the way, as an aside,
this is kind of always-- whenever I give talks
to the public, I trot this out and say
some joke about an MIT professor has to have something
really geeky like this.
This is one of the most famous and beloved plots now
in the election administration space.
They love it because it describes very well the world
that election officials have lived over the last 20 years.
So the way you read this plot--
this is 1996-- the states at the top of the triangle
are all voting in person on Election Day.
The states down at the bottom left are all voting by mail.
And states at the bottom right are all voting early.
You will see that almost every state is up at the very, very
There's a few states, Washington and Oregon,
that are beginning to ooze down the left, beginning
to experiment with voting by mail.
There's the states of Tennessee and Texas
that are beginning to experiment with voting in-person
before Election Day.
So that was 1996.
This is 2016.
OK. '96, 2016.
We can see here that Washington and Oregon are all
voting by mail.
Colorado is not too far behind.
We have about a half dozen states over here
predominantly voting by mail, about another half dozen
that are predominantly voting early in-person.
But most states are still predominantly voting--
predominately voting in-person on Election Day,
so that the trick is going to be to get that big discontinuity
in people voting by mail, we're going
to have to move these states that
are in the upper right hand part of this triangle
down into the left.
For instance, Massachusetts actually
is kind of a prototypical state in this regard.
And it's going to take a lot of work.
So lack of experience getting a lot of
states to change how they're doing.
Lack of capacity-- when you don't
have a lot of mail ballots, you can do a lot of things by hand.
When you have a lot of mail ballots, you have to automate.
And you have to buy machines like this
from a company called Runback.
This is the machine that's used by--
the most common machine used to ingest mail ballots,
to scan them, sort them so that they
can be counted to take a digital image of the incoming envelope
so that signature matching can be done.
This is the sort of machine that you
would need if you were the city of Chicago.
The city of Chicago has three of these
to manage their mail ballots.
Each one cost $100,000.
And so if we're going to be moving
to this large capacity of voting by mail,
your basically-- a city is going to have
to be buying a fire truck or two or three having not known
that they had to buy a fire truck a couple of months ago.
There's also the problem of restrictive laws.
A lot of states that had previously not been voting
by mail had made it very hard to vote by mail,
and frequently would reject ballots that
would come through the door.
New York reminded us of this during the summer
when, during their primary, they ended up
rejecting 10% of the mail ballots received
in that primary.
So it's one thing to be rejecting
10% of your mail ballots if only 5% of your voters
are voting by mail.
But if 60%, 70%, 80% of your voters are voting by mail,
then that's going to be an entire other story.
And then there's going to be just the need
to develop procedures and to procure just a lot of stuff.
Here's a graph.
I'm not going to go into the details.
But Nate and I developed this graph back in March
to delineate all of the procedures
that an election official would have to worry about almost
writing from scratch if they were to move into a regime
where they did a lot of voting by mail.
A lot of manage--
a lot of standard operating procedures
need to be developed from a standing start.
Protecting polling places, in some ways,
is easier to describe, because the issue here is maintaining
protecting people and places.
People, because the problem is these million poll workers,
60% of whom are over the age of 60, and a quarter of whom
are over the age of 70.
The problems of the pandemic during the primaries
were such that these poll workers
were quitting in droves.
In addition, many of the traditional places
that had polling places--
like churches, firehouses, schools--
were closing.
So where are we going to vote?
And who is going to staff those polling places?
The problem in maintaining in-person polling places,
getting the people and the places together,
was illustrated in Wisconsin, and particularly,
and most sadly, in the city of Milwaukee during the pandemic.
This illustrates what happens when you can't plan
and you're facing voting in a pandemic.
This graph shows all the polling places in the city of Milwaukee
they planned to open.
That's all the blue dots.
This is all the polling places they actually opened.
They were only able to open 5% of the polling
places they'd planned.
And that gave rise to those long lines we saw.
So how did we end up doing?
Well, it turns out after the initial problems
in March and April, things actually ended up fine.
I'm not going to talk about this graph in detail,
just using it to prompt me to note
that turnout in the primaries in 2016 was very robust.
And after the dip in April, it was actually higher
than it was--
slightly higher than it was, actually, in 2016.
It's also the case that Americans figured out
how to vote by mail.
This graph shows at a state-by-state level
the percentage of people who voted by mail.
The blue dots in all the states in the primaries, the blue dots
are on numbers in 2016.
The red dots are numbers in 2020.
These are the states that voted before the pandemic rolled in.
These are the states that voted after the pandemic rolled in.
Very quickly, states got mail ballots to voters
and they voted them very, very quickly.
What we are seeing now during this period of voting
is that mail ballots are going out.
They're coming back quickly.
And poll worker recruitment is going surprisingly well,
to the point now that when we talk to local officials,
they're recruiting for poll--
for elections in 2021.
What am I worried about?
I'm worried about partisanship endangering people.
I'm worried about the fact that there
has become a partisan divide over whether people
are going to be willing to vote by mail or not.
And then particularly, Republicans
are listening to Donald Trump and are
deciding that they're not going to be voting by mail, even
if it's going to be good for them.
Democrats and Republicans right now
are reporting kind of differences of 40 percentage
points in terms of the likelihood
that they will vote by mail.
And this is true throughout the country.
And it's being reflected in the administrative records
that we're seeing.
Mail balloting, what am I worried about?
I'm worried about voters waiting too late
to get their ballots back in.
I'm worried about voters not following instructions, not
signing the ballots, leaving signatures off, et cetera.
So I'm worried about a few things.
I think administratively, we've come a long way.
But on the whole, I think we're doing pretty well,
with a few worries to be said.
I want to end up and take questions.
And so I'm just going to very quickly talk
about kind of the decisiveness and give this kind
of weaselly answer, "probably."
And just note on the one hand, if this were a normal president
in a normal year, we would say this
is going to be a kind of a comfortable election.
If we look at the polls right now,
I took this from The Economist, whose models I prefer,
because they have actually--
they actually publish their model.
They have a GitHub with it and you can download it.
Biden is leading comfortably, 54-46
in an average of all the polls.
This would be similar to the margin
that George Bush had over Michael Dukakis in 1988.
If we look at the state level, which determines the Electoral
College, we see that the current predictions
are that Biden would win 346 to 192,
which is roughly the Electoral College victory that Obama
had over Romney.
But this might not be the right measure of decisiveness.
And here's a number of terms that
need to be considered as we move into the vote counting that
may be disruptive.
Something called blue shift and red mirage,
I'll say a little bit about that,
and the concerns about disruption of the count
and transfer of power.
The blue shift-- the blue shift is something
that I've written about with a colleague, Ned Foley,
at Ohio State Law School.
And this just simply refers to the tendency
of the post-election vote count to be
more Democratic than the votes counted on Election Day.
This is a phenomenon that we've observed since 2000.
It accelerated in 2012 and 2016.
And we think that it's going to accelerate even further in 2020
for reasons related, actually, to the rise of absentee ballots
and the fact that Democrats are voting
absentee ballots at a 40% higher rate than Republicans are.
There's related to this red mirage
that you may have read about, which
is kind of a cartoon version of the blue shift.
Under the red mirage, all of those Election Day
votes for Trump are counted first
and then the mail ballots are counted second.
Donald Trump declares victory based
on the votes on Election Day.
And there ends up being social unrest as a consequence.
Nate and I are not impressed with this argument
about the red mirage.
We have an op ed in the Wall Street Journal
from a couple of days ago.
I think more likely that things we have to worry about
are these--
first of all, a disruption of the count.
I think it's reasonable to believe that in some states,
like the state of Pennsylvania where it's going
to take a few days to vote--
to count the mail ballots, that the places where
the ballots are being counted are going
to be targets for disruptions.
And we've already had a history of this.
You'll recall one of my "favorite," in quotes,
episodes during the recount in Florida in 2000
was the so-called Brooks Brothers riot,
when a bunch of lawyers so disrupted the activities
in the Miami-Dade election department
that they actually put a halt to the recount in that county.
So we can imagine that things like this
may very well happen during the recount.
And then of course--
during the counting in 2020.
Then, of course, we have problems
related to Donald Trump declining
to say that he will peacefully transfer
power, which makes us worried--
again, not so much that the election officials won't
do their jobs, but that there will
be an even larger army of people willing
to engage in things like the Brooks Brothers riot
and be even more disruptive in the count in the days
after the election.
So there you go.
That's really quick.
That's a lot of stuff.
I mean, the bottom line here is this.
We've come through a lot as a nation.
Election officials have done quite a lot of work
to pivot on how they run elections.
American voters are figuring it out.
Now, if this ends up being not a close election,
our sense is that there will be a lot
of tweeting after Election Day.
There will be a lot of action in the courts.
There may very well be social unrest.
But at the end of the day, we will
have that peaceful transition of power
that President Trump has not been promising.
However, there are a non-zero probability paths
that can take us in other directions.
And I won't spend time delineating those now.
But I'd be more than happy to talk about those in the Q&A.
And so with that, I'll put a stop to it
and be happy to take any questions that people
might want to throw my way.
So thanks for your time.
Daniela, you are muted.
Thank you, Charles, for giving us this great tour of what
we can expect from elections.
I have a question about your study of the 2016 elections,
where you found that 4% of mail in ballots were not cast.
And then MIT and Caltech researchers previously
found that in the 2000 election, roughly 4% to 6%
of all votes cast were unrecorded.
So given that in elections sometimes
the decision is by thousands of votes,
what do you think when you look at these kinds of numbers?
Well, the first thing it reminds me of
is what's known as the election administrator's prayer, which
is please God, don't let this election be close,
because it's in the close elections where these issues
and what we call lost votes really come into the fore.
And by the way, the lost votes, I mean, just
to make very clear about what we're
talking about in that kind of 4% to 6% of lost votes,
these are oftentimes not votes that are actually
sitting in an election official's office waiting
to be counted.
In the case of, for instance, of in-person voting,
these may be votes--
voters who actually walk away because they're
in a long line and they give up.
Or these could be, as we saw in 2000, hanging
chad and pregnant chad and those sorts of things where
votes actually aren't recorded.
So it's a kind of a dog stew of things
that are out there that could have conceivably have
been a vote if the technology had been better,
the procedures had been better, et cetera.
Now to speak a little bit more concretely about this election,
if this election is like others, we can expect between 2% and 4%
of mail ballots to arrive and be rejected.
And that means about 1% of all ballots--
1% to 2% of all ballots in this election
are going to be rejected.
Now, one consequence of this is--
so I'm about to make myself a liar
by saying that many-- when I said many of these lost votes
are actually not votes sitting around in an election
official's office, well, a rejected absentee ballot
is a vote that's sitting around in an election official's
So if we have a recount in some place like--
if it's really close and comes down
to Pennsylvania or Michigan or Wisconsin,
recounts are litigated over rejected ballots.
And so we could very well have a lot of litigation
around these ballots.
That would have substantive consequences.
We know that these ballots are likely to be more
Democratic than Republican.
So they're going to have a substantive meaning as more
of them get included in the count.
But they're also going to have a procedural consequence.
With potentially 1% of all ballots in election
administrator offices to be litigated over,
there's going to be a lot of arguing.
That's going to be a lot of delay.
And it's going to make it harder for states to actually get
their count done before the meeting of the Electoral
College in the middle part of December.
So some of the kind of worst case scenarios
that involve kind of an election process going off the rails
involve fighting over these absentee ballots,
and the fighting being so delayed
that states risk not having their votes
count in the Electoral College, in which case, we could then
have state legislatures potentially attempting
to jump in and elect electors.
That's a fraught area.
I won't go in--
I won't go down that rabbit hole even further.
But the reason why I say it's off the rails
is you can imagine, if we get to that point,
we're talking about all sorts of constitutional questions that
have not been raised in about 150 years.
And it scares the bejesus out of a lot of people
if we were to go down those paths.
I'll stop there.
Charles, related to that, what would compel the Supreme
Court to get involved, and what would preclude that?
That's an interesting question.
My sense is this--
there's a lot of activity in the federal courts right now
around election--
election rules.
And my sense of it is that the Supreme Court actually
would like to stay out of this election after Election Day
as much as possible.
And that's one of the reasons why,
right now the court has taken--
I mean, taken for consideration a large number of cases
on appeal related specifically to the late counting
of absentee ballots.
Lower federal courts have, in many, many states,
agreed to extend the deadlines for the receipt of mail ballots
past Election Day, so long as those ballots are postmarked.
It's pretty clear that conservative appellate courts
don't like these decisions and would like to overturn them.
But the courts don't want to be the reason why Donald Trump
or Joe Biden was won.
And they don't want to be the controversy.
They don't want to be the story after Election Day.
So my sense right now is that they're working really hard
to get rid of these cases.
They announced one yesterday by basically punting.
They're probably going to announce the others
in the next few days.
And then they're going to do the best they
can to get out of their way.
My sense in talking to my kind of election law colleagues
is the following.
The most likely path is what follows.
If the rules are locked down before Election Day--
and keep in mind that the rules for counting and adjudicating
ballots have become very much more clarified than they
were in 2000.
So if election officials are in a situation where
they can apply pretty straightforward rules
about the ballots that are counted
and the processes for adjudicating disputes,
then the courts are going to let the election
officials do their jobs and count their votes--
which, by the way, I think there's a very good reason
to believe they're going to be counted
much faster than many people believe--
and try to hew to a very strict reading of the election
laws in states.
Because I'm certain that Justice Roberts and the other members
of the Supreme Court recognize that if they have
to answer certain questions in the middle part of December,
it will be really bad for the Court
and it would be very bad for the country.
And so they're doing everything they can to keep things out
of the courts--
keep weird decisions out of the courts.
You don't know what's going to happen on Election Day,
and you know something strange is going
to happen on Election Day.
The final thing-- and so I could say more about that, but just
maybe to kind of tee up something else,
my sense is that we are, in fact, going to know on election
night what the narrative is going to be,
what the path to 270 is going to be.
And if we do know that, then the incentives
for trying to get the courts to pull one out
of the hat for Donald Trump will be greatly reduced.
And so there's also, I think, a wide understanding,
even among many Republicans and many
of the states, that it's imperative
that election officials be well on their way to counting
on all the ballots by Tuesday night at midnight or Wednesday.
So things I think will move much faster than we sometimes
are worried about.
Charles, can you say something about numbers--
for instance, how many have machines they need?
Do we have an idea of how many have enough poll
workers, how many states have signature
checking on mailing ballots?
What does that look like?
On the resources-- on poll workers,
election officials now are pretty much
saying they have the poll workers
that they need in general.
The places where they are needy right
now are-- it's actually in terms of language,
you know, bilingual interpreters, people like that.
And so actually anybody on the call,
if you have a language specialty or skill, especially Spanish,
in this area you can still be used,
I'm sure, in Massachusetts.
That's where that need is.
In terms of hardware, by which I mean
those sorters and those scanners,
they appear to have what they need.
Now, they may or may not be able to deploy them
where they need them.
And so to give you an example, again,
from this call with Georgia today, they
were talking about-- so part of the discussion
was about how to deal with the fact
that many counties have obviously under provisioned
their early voting sites.
And so they need to roll out a whole bunch more voting
machines and a whole bunch more electronic poll books.
And it turns out that the voting machines are generic
and they can go anywhere and they're fine that way.
The poll books are dedicated to a particular polling place.
And they're harder to redeploy.
And so at this moment, I think they're better situated
for things that are easily deployable,
including voting machines, voting booths,
those sorts of things.
It might be a little too early to see if they're well
positioned in terms of having things like poll books that
can be easily deployed.
The other thing we won't know about
is, until some of these systems are tested,
whether they have enough bandwidth
to handle things like big hits to their voter registration
databases as they get surges, say, of early voters.
So there are some parts of the system that
maybe haven't been tested yet.
But on the surface, at least, most people
think that they're well positioned.
The final thing I'll say is that on counting ballots,
there are some places-- and I think Milwaukee and Detroit may
be a couple of these places--
where maybe they've not been able to automate as much as
they may have wanted to.
But they've been able to recruit enough election
workers to do manual counting and processing in bulk.
And so you know, we've been reading a lot
about the NBA opening up arenas for people to vote in.
In many of these cities, they've opened--
they're actually going to be doing absentee ballot
counting in these arenas.
And so they're going to be full of election workers counting
And so at the end of the day, some of this capacity
deficiency can be dealt with by doing more manual work.
And right now, they're recruiting people
to do more manual work in a few large cities.
So just a quick follow-up on that--
Pennsylvania, Wisconsin, and Michigan
will wait until election day to count votes.
How do you think about this?
Do you see it as an obstacle?
So those three cities on those three states specifically--
let me tell--
I mean, those are three states that we worry about a lot.
And we've talked to election officials a lot.
We talked with them offline.
We've talked to the press.
We've talked to lawyers.
This is what I've concluded, and I might be wrong.
This is what I've concluded.
Wisconsin is actually sort of like Massachusetts
in the following sense.
Except for Milwaukee, they count their absentee ballots
in the polling places.
They process them in the polling places.
And they do it on Election Day.
And so there's going to be a big deficit of people showing up
to those polling places on Election Day to vote in person.
So what those poll workers are going to be doing instead
is feeding absentee ballots into scanners all day
long, processing and feeding.
It's going to take a few more hours
to get all that processed.
But the folks I talked to in Wisconsin
are pretty confident that most jurisdictions will
be done counting by midnight.
Milwaukee will be delayed by a matter of hours
rather than a matter of days.
Michigan, they've gotten the state legislature
to agree to allow them to begin processing
ballots before Election Day.
They get 10 hours ahead of time.
By the way, the evidence that convinced
the Republican leadership to do that
was actually produced by one of Steve Graves' students.
And so that's a win for MIT.
So people, again, my sources are convinced that even Detroit may
be done by Wednesday at noon.
It's not going to be a big deal.
I'm more worried about Pennsylvania.
And I'm not worried about Pennsylvania
because they have a large number of ballots
and because they're going to be counted and processed
on Election Day.
The thing that worries me about Pennsylvania
is that they have a highly partisan system of election
administration where there is a D and an R
at every level of election administration.
So it's partisanship all the way down, turtles all the way down.
There's also a history in Pennsylvania
of frivolous challenges to absentee ballots.
And so the thing I worry about in Pennsylvania--
and there's some evidence that this is going to happen--
is that Republican poll watchers-- in Philadelphia,
will challenge every single absentee ballot one at a time
as they go to count those in Philadelphia.
And that's where I think-- and so that's
the thing I worry the most about and where
I think I am hoping that the judges in Pennsylvania
will not put up with these sorts of-- with those sorts
of frivolous challenges.
But that will be one of the areas in which the court
system will be tested to see whether they will batten down
obvious attempts to try to rope-a-dope the count.
Charles, going to another question from the audience--
are there any changes in the way that the press
reports on election returns on election night and afterwards?
And are there best practices they
should follow to set calm public expectations and preempt
candidates from making unjustified declarations
of victory or otherwise?
Yeah, that's a really great question.
And I think it's a subtle one.
And in the interest of full disclosure,
I should note that I'm actually on the CBS call desk, decision
desk, although nothing that I say
will be based on any special knowledge from that.
But here's some things.
Thing number one is that the decision desks
more than anyone else have understood
the precarious position they're in in terms
of partial counts on election night.
I mean, this is their life.
I mean, this is what they do.
And they've known better than anybody else the stream
of information coming in on a typical night,
how variable it is, the fact that one part of a state
comes in first then another part of the state comes in first.
And you can look-- you can have a big Republican surge and then
a big Democratic surge.
They've known that for years.
And in fact, the modeling they do
is all about keeping track of where the ballots are coming
in, characterizing the ballots that haven't come in,
estimating the likelihood that the trailing candidate can
overtake the leading candidate given
the distribution of the ballots that haven't come in.
And they're very aware of this.
So here's the conundrum, that a free press is not
going to suppress facts.
And a partial vote count is a fact.
So I mean, we just have to put that out there.
If there really is a red mirage, that is going to be a fact.
And nothing is going to stop the AP or CBS or anybody
from just not reporting the votes that
have been posted up on a door post outside of a precinct.
That is going to happen.
So that having been said, one change
that's going to be different this time
is that every outlet except for the AP--
and the AP is not as important as it used to be--
every outlet except for the AP has
models of anticipated turnout and is
going to be reporting partial election
returns in terms of percentage of expected turnout.
New York Times is already doing this, by the way,
in their web pages.
So that's one way of inoculating against kind of partial claims.
So that's one thing.
So there is going to be reporting
trying to be just much more--
just much more transparent about where the ballots are
coming from and what they mean.
Having said that, so here's a couple of things
I've just put in as--
to kind of temper what I just said.
The first thing is that I think we
have to be very careful about characterizing
partial information, because we want to be patient.
But don't want to mistake that patience
from believing like we know nothing, that we only
trust the complete count, official count, which will come
in two weeks, because we know that if we don't report
anything, the candidates are going to make claims anyway.
And so we're going to have to be more
sophisticated in our consumption of the numbers on election
And that's why I've been touting this story,
is that although, thing number one, the election is not
official until the states certify the count,
and that's not going to be, in most states,
for a couple of weeks, we are going
to be able to make inferences, informed inferences
by the pattern of the votes as they come out
on election night.
And this is what I think the most likely narrative is going
to be, path of information is going to be,
and it's forming my-- and I see we have one minute, so I'll
be fast, but here's the deal.
I think this is important.
Florida-- Florida will report 70% of its votes
within half an hour of polls closing.
By 11 o'clock, Florida will have reported everything.
95% of the ballots in Florida will have been counted.
We will know by 10 or 11 o'clock if there has been basically
the five percentage point shift away from Donald Trump
that the polls are suggesting.
We will also have complete returns from Kentucky,
from Virginia, mostly from North Carolina,
from a large number of states on the East coast.
We will know by 11 o'clock if this big national shift
has occurred.
If that big national shift has occurred, we will know that--
if Florida has gone to Trump, he cannot win Pennsylvania.
He cannot win Wisconsin.
He cannot win Michigan.
And the question at that point is going to be,
is he going to lose Georgia, too?
That's going to-- those are inferences
that we can draw from the early patterns,
because election returns are very regular.
And so that's going to set an expectation.
It's not going to be a guarantee,
but it will set an expectation and a narrative.
And so I think on election night what
we're going to see is a lot more discussion of the shift
since 2016 and where the votes are shifting, where they've
come in, and a lot more explanation from the decision
desks when they do, quote unquote,
"declare a state" why they have declared that state.
And I think the networks are going
to take longer to declare states because of the uncertainty
and the desire--
and the problem of the numbers coming in.
Charles, thank you so much.
We are at the top of the hour and we've
had such an interesting conversation with you.
We did not get to some important topics,
like what computer scientists could do to contribute.
And perhaps we can come back to have a conversation about that
after the elections.
Also, I did not get to ask you my favorite question, which
is your opinion about the difference
between an electoral college approach to voting
and something like the Australian system,
for instance.
So we have much more to talk about
to continue the conversation.
And I look forward to that.
Thank you so much.
Thank you.