The Future of Elections: Who Votes, How We Vote, and What Comes Next

The future of elections

0:00The mighty dollar. Once upon a time, it had a whole menu named after it. Now, no more. But one dollar a day lets you add Progressive renter's insurance to your auto insurance. Definitely something your younger self would roll their eyes at, but your current adulting self appreciates. Learn more at Progressive.com. Progressive casualty insurance company affiliates and other insurers. Average premium costs based on Progressive home data from 2025. Product and prices may vary by state.

0:30An election is one of those machines that looks simple only until you ask what, exactly, it's supposed to measure.

0:40Welcome to Science and Futurism with Isaac Arthur, where today we are looking at the future of elections, not who should win the next one, what an election has tried to accomplish, how many different ways there are to accomplish it, and what happens when new technologies let us do things that were previously impossible, while creating new problems our existing systems were never designed to handle. At the simplest level, we tend to imagine voting as asking everyone what they want, counting the answers, and giving the most popular answer the win. Or sometimes the least unpopular option, but elections actually contain several different problems stacked on top of one another.

1:16How do we verify that each eligible person votes once, preserve a secret ballot, count everything correctly, and then demonstrate that correctness convincingly enough that the people who lost believed the outcome was true and fair?

1:53And how do those answers change when computers can examine millions of possible district maps, digital identities can authenticate voters, cryptography can verify information in new ways, ballots might be cast from anywhere, and artificial intelligence can both help administer elections and create new ways to attack or manipulate them? Some technologies may let us discard compromises that existed only because better options were imprentical, others may force us to add safeguards nobody needed before, electronic voting and cyber security being a big one I spend a ton of my time on,

2:26and sometimes the solution to those sorts of problems may turn out to be a very old technology, like a reassuringly thick piece of paper. Improving one part of an election can easily make another harder, more proportional representation can be harder for voters to understand, remote voting might be wonderfully convenient while making secrecy and coercion harder to police, an algorithm can draw districts consistently while quietly embedding somebody's definition of what a fair district is, and a voting system can be extremely secure and still fail politically if voters can understand why it is secure.

3:01What we are mostly not doing today is redesigning democracy itself. When I first started writing this episode, it got long very quickly because voting methods led into direct democracy, liquid delegation, AI political agents, sortition, prediction markets, altered franchise, uploaded minds, and eventually the question of whether a copy of you gets another vote. At that point, I felt we really had two episodes. Those broader questions became the companion episode, The Strange Future of Democracy, What Technology Makes Possible,

3:34which will be out in a month or two. This one stays close to the machinery of elections themselves. The timing is deliberate too. This episode is coming out shortly before an election here in the United States, but again, I'm not going to tell you whom to vote for, what party to support, or which referendum deserves a yes or no. The candidates on states' ballots will eventually become historical footnotes. The mechanisms we use to choose among them are much older, and will probably still be evolving centuries from now. This is also one of those subjects on this show where I have another professional hat,

4:06besides the futurist one. I chair the Board of Elections for my area, and report to the Ohio Secretary of State. I also sit on the Legislative Committee of the Ohio Association of Election Officials, and have been working in election administration for about as long as I have been making this show, and election vote was before that, ever since getting out of the army in 2010. Part of the reason for starting the show, and keeping away from current events in politics, is that it is a nice escape from all the politics and bureaucracy I get inundated with, and I tend to feel a lot of the audience wants the same.

4:36So please respect that in the comments below. But I spend a surprising amount of my life thinking about ballot paper, precincts, voting machines, polling locations, district boundaries, voter registration, recounts, audits, and all the machinery most voters quite reasonably prefer not to think about unless something goes wrong. I also spend a lot of time studying how other states and countries run their elections, and what folks contemplate for future ways to run them. Ohio is often called a national gold standard for election administration.

5:07We tend to be the place others seek to beat, or brag about having one upped. But that does not mean I think Ohio has the one correct system, more that we are good at administering the system we inherited. Other states and countries have approaches and ideas generally worth considering. Needless to say, that experience makes me a little suspicious of any proposal that solves an election problem in one elegant sentence. Elections have to work mathematically, legally, physically, securely, and psychologically, usually all at the same time.

5:38This is also intended as a survey episode, so even though it's long enough you'll want a drink and a snack, we will only be giving quick details on most systems. You can doubtless find other sources detailing the history and pros and cons of any that catch your interest. We'll be skimming and trying not to oversimplify as we go. But through everything today, the useful question is not really what is the best election system. It is, what problem are we trying to solve? What trade-off are we willing to accept to solve it? And what technologies might alter the balance on deciding how we decide?

Why would elections change

6:12Why would elections change? The first thing to understand about the future of elections is that election systems are unusually conservative technologies. I do not mean politically conservative. I mean that once a system is familiar, lawful, and broadly trusted, changing it carries a cost all by itself. That makes sense. If I replace the software on my phone and dislike it, I can complain for a week and eventually learn where they move the settings menu. If I change the rules that determine who wins public office, everybody has an incentive to ask whether the change was really about making elections better,

6:45or about making some particular side win more often. And often, that is the intent, or at least the icing on the cake. So election systems tend to be sticky. A new method does not get adopted merely because somebody proves that it has an elegant mathematical property. Usually, there has to be a problem people actually feel. Crowded candidate feels, impossibly long lines and waits to vote from overly long ballots, repeated spoiler concerns, expensive runoffs, districts people regard as unrepresentative,

7:16inaccessible polling places, cyber security threats, or simply old equipment reaching the end of its useful life. Pricey, too. We spend thousands of dollars on a single voting machine we usually only use twice a year and have to replace in a decade, and they still often look like a glorified paper shredder. Same for a tablet poll book that's obsolete in five years, after which time we've used it all of ten times. Technologies can create pressure of its own. Once voters can bank, work, sign contracts, and communicate from anywhere,

7:48being told they must appear at one particular school gymnasium on one particular Tuesday can feel increasingly arbitrary. Once computers can draw ten million legal district maps before lunch, old assumptions about how boundaries get drawn become harder to sustain. And some technologies make an old alternative newly practical without making it newly wise. Ranked ballots, proportional representation, and weighted voting are not inventions of the computer age. What computers, networks, digital identity, scanners,

8:18and eventually artificial intelligence do is change their cost, scale, speed, and sometimes the kinds of failures we have to worry about. At first glance, elections still seem simple. Ask everybody what they want, count the answers, and do whatever got the most votes. Ten people want pizza, and six want hamburgers, so we order pizza. Democracy accomplished. But was it the optimal outcome? Certainly, it would seem wrong to order hamburgers when most prefer pizza, but it would approach the problem by deciding on pizza or burgers,

8:50while modern tech can circumvent that issue by letting everyone order what they like. Of course, a country is rarely choosing between pizza and hamburgers, nor it would have both as an easy solution. Suppose 40% of voters prefer Candid Alex, 35% prefer Bob, and 25% prefer Carol. Under ordinary plurality voting, Alex wins. But suppose nearly every Bob voter prefers Carol to Alex, and nearly every Carol voter prefers Bob to Alex. A majority of the electorate would rather have either alternative

9:23than the candidate who actually won. So, did the election correctly measure what the electorate wanted? Maybe. Alex generally did receive more first choice votes than anyone else. The problem is that we have quietly changed the question. Do we want the candidate with the largest first choice following, the one acceptable to the greatest number of people, the candidate who would beat every opponent head to head, or a result that makes the eventual legislature resemble the electorate as closely as possible? Those are not always the same answer.

9:55Collective preferences can get stranger still. You can construct cases where a majority prefers Alex to Bob, another majority prefers Bob to Carol, and another majority prefers Carol to Alex. The electorate has collectively produced a circle. Results such as Arrow's impossibility theorem formalize the deeper problem. Once we have at least three alternatives, no method of aggregating everyone's ranked preferences can always give us a collective ranking, while satisfying every apparently reasonable condition we might demand from it.

10:25So, there is no mathematically perfect voting system hidden in a filing cabinet waiting for civilization to become sensible enough to adopt it. There is only a collection of trade-offs, and the first of those appears on the ballot itself. How much information do we actually want a voter to be able to give us?

When one mark isn't enough

10:45When one mark isn't enough the simplest common ballot asks you to choose one candidate, A, B, C, or D, whoever gets the largest pile of marks wins. That system has an enormous advantage that can sound almost embarrassingly unsophisticated. People understand it. That matters more than many election reform discussions acknowledge. I personally prefer to keep the amount of procedural knowledge a voter needs to cast a valid ballot as low as reasonably possible. When I talk about voter education in this context,

11:18I do not mean teaching people what to think about candidates or issues. I mean how much instruction somebody needs before they can successfully express their choice they already made. Simple plurality voting scores very well there, but it also deliberately throws away information. Maybe Bob is your favorite, Carol is nearly as good, Dan is tolerable, and Alex is the candidate you most want to keep out of office. The ballot records only Bob. If polling suggests Bob has almost no chance, you may abandon your favorite and vote for Carol

11:49because avoiding Alex matters more to you. That is not necessarily irrational behavior, it is a response to information the ballot allows you to express. A similar issue appears with ballot questions. A tax levy may ask yes or no, even though a voter might support half the proposed amount, or support the tax only if the money were divided differently. Fun fact, the first structured video I ever put on this YouTube account, before it became a science-focused channel, was me going over roughly 50 local tax levies

12:20on the upcoming ballot, and where each applied. Thrilling cinema. I remember someone asking why voters could not have more choices, perhaps approving half the requested levy, or ranking spending priorities. In principle, you can design a ballot to ask much richer questions. In practice, every added option complicates the law, ballot design, voter instructions, tabulation, recounts, audits, and the meaning of whatever result comes back. One traditional answer to a crowded field is to have large parties with primary elections of their own,

12:53and another is a runoff, or another election between the leading candidates if nobody gets the required threshold. That is easy to explain, but it means actually running another election. With another round of staffing, ballots, equipment, campaigning, and voters having to show up again, depending on turnout and location in the US, you are looking at roughly $10 to $20 per ballot cast, before anything spent by candidates or voters themselves. And those ballots themselves, typically only cost about $0.20 to $0.30 a piece. My usual comment on election budgets

13:24is that if you spend hundreds of billions of dollars defending democracy, spending a few billion practicing it, is pretty reasonable, but it's still a costly process, and the cost does not rise much when you add more candidates or issues to one ballot, which is one reason frequent special elections are frowned on. Instant runoff voting tries to compress that second election into their first ballot. Voters rank candidates, the lowest candidate is eliminated if nobody has enough support, typically half plus one, though it doesn't have to be, and those ballots transfer the next available choice

13:55into someone meets the rule for winning. The idea does not require advanced computers. People have counted ranked ballots by hand for a long time, and ranked voting is fairly common worldwide. Modern scanners and software simply make complicated transfers faster and easier to administer at scale, and make it easier to simulate how different county rules behave before adopting them. The attraction is clear enough. A voter can support a less popular first choice without necessarily surrendering any say over the contest between the leading candidates,

14:26and a jurisdiction may avoid needing a separate runoff. The cost is also real. The result is less immediately visible from the first pile of ballots. Voters can rank only some candidates and eventually have no remaining choice to transfer. Ballot design and instructions matter more. Auditors have to verify not merely totals, but the transfer process. And if people do not understand why the candidate who led on the first count did not win, the mathematically correct explanation does not automatically produce political trust. For a lot of folks, seeing Alex at 38% to Bob's 27% on round one,

14:59then becoming Alex's 47% to Bob's 53% in the final round, as he picked up more second or third choices, just seems unfair and confusing. For my part, I am personally more skeptical of ranked choice systems than many reform advocates are, largely because I put a very high value on simplicity and transparency at the voter-facing end. That said, ranked systems do have their uses. For the National Space Society Awards, for instance, I implemented a border-style ranked scoring system.

15:30Developed back in the 18th century in France and used in modern Slovenia, it has voters assigned higher point values to nominees they preferred, and the total produced the final ordering. And in this case, they could only use each number once, except zero, blank. It worked well when all the nominees were already viewed as worthy, no one was denouncing nominees as crazy, just debating which was most awesome. And we needed an ordered list, as we were giving out more than one award, and also opting to have finalists never respond or decline. But it was still a bit tricky to explain,

16:01even to an organization that tends to be high on math and logic skills. So when I found a good online voting app for it for the second attempt, things went way smoother. Shorter term, I could also see ranked systems working better in multi-seat, non-partisan races, such as school boards or township trustees, where voters may already be choosing three candidates out of ten or so. Frequently, the candidates are mostly unknown to most voters, and while I've seen exceptions, they tend not to be as heated as federal or state races. I would not automatically assume

16:32success in those lower-profile contests means it scales cleanly to a major state or federal race, but it's a good place to experiment. Approval voting makes a different trade. Instead of ranking candidates, mark every candidate you consider acceptable. Score voting might let you rate each from 0 to 5 or 0 to 10, kind of like restaurant or venue reviews. Nothing stops you saying multiple options are 5 stars or 1. I sometimes think of approval voting as the anyone-but-that-guy method,

17:02as it can favor people with a low disapproval rating more than a high approval rating, which can lead to lesser-known nice-guy candidates getting in, for better or worse. Those methods invite strategy of their own. An approval voter may approve only a favorite rather than every acceptable candidate because they fear helping a rival. A score voter may learn that using only the highest and lowest scores gives them more influence than honestly describing every shade of preference. Every restaurant has either zero stars or five stars.

17:32Computers can collect vastly more information from a ballot than one mark beside one name. The scarce resource increasingly becomes human attention and public confidence, rather than arithmetic. So I expect some future contests become more expressive because technology makes them cheap enough to administer, while many others remain almost aggressively simple, because simplicity itself is a feature, not a bug.

When votes fail to represent

17:57When votes do not produce the representation people expect. Changing the ballot is only one way elections can change. We can keep exactly the same ballot and change what the votes produce. Much of the United States uses single-member districts. One geographic district elects one representative. If candidate A gets 51% and candidate B gets 49%, A represents the whole district. Repeat that across hundreds of districts, and the composition of the legislature can differ substantially from the overall vote.

18:28One party can win many districts narrowly, while another wins fewer districts by enormous margins. Political groups can also become geographically concentrated, so their share of seats depends partly on where their supporters happen to live. And by default, historically, your voters will break into one of two alignments. It's urban-centered versus rural, or region A versus region B, north versus south, for instance. You also don't automatically get ideal representation this way. If you've got a region that is 51-49 orange party

19:00versus purple party, with two representatives, they elect one of each, and that's fine. But a place that's 60-40 purple versus orange is still generally electing one of each party, that gives you a lot of motivation when districting areas to cut them up into a lot of districts where you have 40% of the vote and will get one rep out of the two, and the other is where you have 70%, and that's where we'll reliably grab both seats. And if you looked at your two districts there, you had 55% of the vote and just grabbed 75% of the seats. You can partially get around that

19:31with bigger districts and more seats. If your objective is to ensure that a smaller but significant gray party with about 11% of voters gets representation, then a 10-seat district using proportional allocation makes that much easier than 10 separate win-or-take-all districts. These are not futuristic inventions. New Zealand, for example, replaced first-past-the-post with mixed-member proportional representation after voters approved the change in a 1993 referendum. First-past-the-post is simply where you win

20:01if you have more votes than anyone else, rather than specifically getting a majority. And we use a closely-related plurality system in multi-seat contests, where the top three vote-getters might take three school board seats, even if they themselves only got about 10% of the vote each. Technology can make proportional systems easier to administer and explain with fast tapulation, interactive ballot instructions, public simulations, and audit software, but does not create the political reason to adopt them. That reason is usually some version of we do not think the seats

20:32reflect the electorate the way they should, and the counter-argument is some version of the alternative weakens something else we value. A single-member district gives voters a particular person tied to a particular place. Larger proportional districts may reflect overall political opinion more closely, but weaken that local relationship. Party lists can give parties more control over who reaches office. Multi-member districts can give voters more options, while producing longer ballots and more complicated counting. Proportional systems also tend to make coalition governments more common,

21:04which some societies regard as a healthy compromise, and others regard as a reliable method for ensuring nobody quite got what they voted for, or no major changes ever happen, which you might regard as a bug or feature, depending on perspective. That is why I would not predict future democracies all converge on one voting method. I would instead predict more divergence and experimentation where the costs of experimentation fall, especially in cities, states, new constitutions, and hopefully eventually new space settlements away from Earth.

21:34Those places may get to start with a clean slate, though history has plenty of examples reminding us that a clean slate does not automatically produce a better system.

Drawing lines through people

21:46Drying Lines Through People If we keep geographic districts, another old problem becomes increasingly technological. Every ten years in the United States, we conduct a census and account for people being born, dying, or moving around. Some areas grow while others shrink, and if boundaries remain fixed forever, representatives from fast-growing regions could eventually represent several times as many people as representatives from declining ones, so districts get withdrawn. I have been through that process repeatedly from an election administration side.

22:18It is not just a state-level matter either. Municipalities with district representation, like city wards, had to redraw their maps too, and it is always a mess. Equal population is only one thing people tend to want. We may also want contiguous and reasonably compact districts, fewer splits of counties, cities, townships, and neighborhoods, compliance with legal requirements involving minority representation, competitive elections, and preservation of communities whose residents actually share local interests.

22:48Those goals can conflict almost immediately. A wonderfully compact district may split a city in half, keeping counties intact to produce larger population differences, following a river valley, industrial corridor, or transport network can create an odd-looking district that represents a real community much better than a mathematically tidy square. And then there is gerrymandering, deliberately shaping boundaries to influence political outcomes. An ugly-looking district is not automatically a gerrymander though. A city that is 20% too small

23:19to form a district by itself might reasonably be combined with commuter communities along the highways rather than a slightly closer farming town. The approach can be debated but the intent is not necessarily to rig an election. This is one place technology has changed the game dramatically. Modern geographic information systems and powerful computer maps can evaluate enormous numbers of maps, combine detailed demographic and electoral data, and estimate the consequences of moving a boundary one precinct at a time.

23:49Needless to say, that can make gerrymandering much more precise. It can also make detect yet much more sophisticated, instead of arguing only about whether one map looks suspicious, Atlas can generate thousands or millions of legally-valued comparison maps and ask whether the adopted map is an extreme outlier. Artificial intelligence may push both directions further. An AI could search gigantic design spaces for maps that satisfy a long list of public criteria, or quietly search for the legal map that gives one faction

24:20the largest possible advantage. As usual, technology remains admirably neutral about which side of an arms race it helps. One future response is more algorithmically redistricting, but an incorruptible algorithm still needs somebody to define fair, should it maximize compactness, minimize county splits, maximize competitive districts, preserve communities of interest, or make the statewide seat total resemble the statewide vote. Those can all sound reasonable and still produce very different maps,

24:50and while I hear a lot of talks from all sides about developing ideal fair map methods, I haven't seen one yet, and I don't have a clue how you would objectively do that either. Any version that significantly alters the current balance of power is basically a non-starter, and any that don't tend to have a lot of dubious compromise tweaks in them designed to explicitly gerrymander the map, just with the intent of status quo rather than adding one side. Another response is to make individual boundaries matter less. Larger multi-member districts,

25:21or proportional systems, can reduce the importance of moving one street from one district to another. We also might ask why districts contain almost exactly the same population if representatives themselves no longer need identical voting weights. Suppose a future legislature aims for districts of about a million residents, but allows coherent communities to remain intact anywhere from perhaps half a million to two million. A city of 1.3 million could stay together, an interconnected rural region of 780,000 could stay together,

25:51their representatives simply cast different amounts of voting weights in the legislature, 1.3 votes and 0.78 votes, or we multiply everything by a thousand if fractional votes offend the dignity of parliament. Weighted voting is not technologically difficult. If the power goes off and you absolutely have to hold a vote right then, and nobody has ever heard of batteries, you can still run it with paper tables, calculators, and a whiteboard kept in a box by the podium. If your government cannot find a few

26:21members or aides capable of multi-digit or decimal addition, I would question the competency of said government. So it's already possible, computers merely make it trivial. It would require a very different constitutional structure from the current US Congress, and it creates problems of its own. For example, a representative serving two million people has a much larger constituency service burden than one serving half a million, so you might need bigger offices, let representatives have deputies of their choosing who can act as true proxies

26:52even when voting for bills or on committees, or permit larger districts to opt for proportional voting or some other multi-winner system. As a side note, while I expect it will be slow, I also expect we'll see more and more bodies meeting virtually for committees and voting. Besides how expensive gathering everyone in a brick and mortar building is, even just to testify on some minor matter, you circumvent the whole issue with legislators having to spend tons of time outside their district in the capital instead. I think this is more likely to pop up

27:23to the state level first though, as travel is still very inconvenient at this level and funding more limited, while local governments don't really benefit from it as much. Anyway, it illustrates the kind of change technology can make easier to contemplate. Instead of meeting in person or repeatedly moving boundaries around to keep representatives equal, a future system might let the boundaries move less and adjust the legislative weight instead. You would probably still need trigger conditions for changing districts, which we currently do every ten years more or less all at once,

27:53but I suspect the once-a-decade census is another tradition on its way out, obviously complicated by being written into the constitution. The notion was to get a decent headcount as often as was practical with options available a couple of centuries ago. We can already maintain far more frequent population estimates and administrative counts than that, and I would expect those to become much more accurate and closer to real-time. Governments have plenty of motivation to want them. Once you are at that point, especially if machines are doing most of the work of redistricting anyway,

28:24you could shift toward redistricting every election or every other year, with each significant deviation from ideal population inviting another legal challenge, or frustrated voters and legislators might prefer more static boundaries and simply tweak each district's voting power. I can say from personal experience that most legislators hate redistricting, because relationships they worked hard to forge with communities get tossed aside when huge chunks of their districts shift. So I think at some point we do see a push toward more static districts again,

28:55but the important prediction is not that this particular idea wins. Rather, that computation gives future constitution writers many more ways to separate geography from political weight than were convenient when a legislative vote literally meant counting raised hands.

Who actually gets a ballot

29:09Who Actually Gets a Ballot Before an election can record anybody's preference, it has determined who is entitled to participate. That sounds like a philosophical question, and in the companion episode we will get into the stranger versions involving children, artificial minds, uploaded people, and whether a copy of you gets them the vote. But it is already an important issue as folks debate who counts as a qualified elector, whether someone who works in a town or owns property there should get a vote there, or whether someone temporarily living there for a couple of years

29:40should be voting there, back home, or even in both places. We talked earlier about losing information during voting, and this is a good example of where what you want the system to capture can vary. You could make a case that a highway into town ought to be principally funded by the folks from outside town who use it, perhaps through a sales or income tax. The wisdom and fairness of that are certainly debatable, but if it is being voted on only by the folks who live there, it is arguably rather unfair of the folks actually paying it. Resort or bed taxes offer a similar example

30:11and are often popular with local governments precisely because their own voters do not pay much of them themselves and thus don't get as angry when it gets raised. Others would argue that property taxes should only be voted on by folks who own property in a district, or that everyone who lives there gets a vote, but people who own property there should be able to vote on those taxes too. I have seen well-reasoned arguments for and against all those approaches over the years, though in the US at this time, the basic rule is one person, one vote, attached to the location

30:42designated as your primary residence. A lot of the reason for doing it that way is fundamentally administrative necessity. Trying to run a property tax-only ballot that everyone who owns property in town can vote on makes me shiver, but technology could change that. It gets hard enough as it is. I have had to conduct hearings where we decide whether someone, usually a candidate for office, rather than a regular voter, though voters too, actually lives at the place they say they do, and whether it really matters for the question in front of us. A lot of people designate a place

31:13as their primary residence for tax or mortgage rate purposes, but mostly don't live there or even rent it out. Others have a place they truly consider their home, vote from it, but spend more time away from it than at it. That can get tricky legally and ethically. Voter registration systems have to deal with people moving, dying, changing names, becoming citizens, crossing jurisdictional boundaries, duplicate records, and data that is simply wrong. As those systems become more automated and interconnected, registration can become much less visible

31:44to the voter, and in this context, that can be a good thing. You register once, and your registration follows you more or less automatically as you move around. We can already print ballots on demand, and I could see that reaching a point where your ballot gets printed at whatever authorized polling location you visit, complete with any unusual add-ons your jurisdiction permits, perhaps a property tax question where you own a cabin, or an income tax question where you work. That might be awesome, or a nightmare, your mileage probably varies,

32:15but it's another door technology can open. Does Election Day

Does election day need a day

32:20Need to Be a Day? For much of history, elections happened at a particular place and time because that was how people, ballots, observers, and officials could be gathered together. Modern technology has been dissolving that constraint for years, and voting during COVID seems to accelerate that trend, as it did with remote meetings. In the 2024 US General Election, only about 37% of ballots were cast in person on Election Day itself. Roughly another third were cast early in person, and about 30% by mail.

32:52Election Day remains the legal deadline and cultural focal point, but for a large share of voters, the election is already an election period. Convenience is the obvious reason that may continue. Overseas voters on business or in the military cannot simply come home on Election Day, and even locally, a citizen who works a 12-hour shift, has limited transportation, cares for children, is traveling, or lives far from a polling place, pays a higher practical cost for a system that insists on one location during one narrow window of time.

33:24Resilience matters too. A storm, power outage, cyber incident, epidemic, or local emergency is much less capable of disrupting an election if voting is spread across time and multiple locations. Technology makes that easier through electronic poll books, real-time registration data, ballot-on-demand printing, improved mail tracking, and voting centers where a citizen can vote at whichever authorized location is convenient rather than one assigned precinct. Those technologies can also make same-day voting more flexible.

33:55A student from Texas attending college in Florida could conceivably have the correct Texas ballot printed at a Florida vote center and scanned there that same day. On the flip side, there is a lot to be said for an extended voting period, especially with absentee or mail voting, because it gives you a nice window to review the ballot at home and research down ticket contests where most voters do not recognize every candidate. You can sit around googling things over coffee in the comfort and privacy of your own home. People feel rushed at polling locations,

34:27even early voting. But extending the window creates costs too. Ballot collection and chain of custody questions become more important, especially when third parties gather absentee ballots. Election officials have to secure ballots, equipment, and records for longer. And mail voting adds more reliance on a postal service whose normal job is delivering all the other mail, too. Election officials and post offices butt heads a lot, incidentally. More voting methods also means more procedures to explain and more failure modes

34:57to train for. An election held over several weeks means some people vote before debates, scandals, emergencies, or major news that later voters know about. So future societies may not simply maximize convenience. They will probably gravitate toward whatever balance of convenience, common information, observability, and administrative control their technology and political culture can support. But people generally like having more options, so I suspect we end up with more ways to vote and a longer period to do it in, such that the phrase

35:28Election Day survives long after most people stop voting on one specific day at one specific location.

Why can't I vote from my phone

35:34Why can't I vote from my phone? This is probably the future election question I hear most often. I can transfer thousands of dollars from a bank account on my phone. I can file taxes online, buy stock, sign contracts, access medical records, send encrypted messages around the planet, and officials can issue extremely consequential orders by phone. So why am I still filling in an oval with a pen? Surely,

36:04phone voting is inevitable. Maybe. The demand for it is easy to predict. Younger voters grew up expecting government services to be available online. Military and overseas voters already faced distance problems. Indeed, I got into elections in part because I was furious that for various reasons, I never managed to vote one single time in four years deployed overseas, to both war zones and friendly garrisons like Germany. Voters with disabilities or seniors in nursing homes may benefit from

36:34electronic ballot delivery or accessible remote tools. Emergencies can make physical voting difficult, and a settlement on Mars is eventually going to find the idea of merely an absentee ballot to Earth mildly impractical. But elections, combined security requirements, ordinary online transactions usually do not. Your bank wants to know that you authorized to transfer and happily keeps a permanent record connecting you to it. If $10,000 vanishes, the bank wants an audit trail showing exactly where it went.

37:06An election must authenticate you too. We have to know you are eligible, that you vote only where and as permitted and that your ballot was included correctly. At the same time, we generally do not want a permanent record connecting your identity to your choices. We may specifically not want to give you a receipt that proves conclusively how you voted, which makes it harder to isolate a suspicious ballot and simply ask the voter, is this who you picked? A secret ballot is desired for more than hiding your vote from the government. It also makes coercion

37:37and vote buying harder because you cannot reliably prove how you voted. If someone offers $100 for a vote, they cannot demand a reliable proof of the choice. And the same applies to an employer telling workers whom they are expected to support or a pushy spouse or parent standing over somebody at the kitchen table. A polling place provides something a phone cannot easily reproduce. A controlled environment where election officials can make a reasonable effort to ensure the voter is alone and free to make a private choice. Your kitchen table does not.

38:07Then there is the device. Election officials can harden, test, seal, inventory, and observe equipment they own. We cannot secure every personal phone and laptop in the country. A device may be outdated, infected, stolen, rooted, remotely monitored, or running a free game whose developers seem unusually fascinated by your contacts and banking information. So too, even a perfectly secure election server cannot save a vote that malware changes before it reaches the server. Nor can we expect everyone to report

38:39a child, partner, or parent who knew their passcode and voted for them. Felonious behavior or not, they do not want them in prison. Blockchain does not magically fix that, and while it does have some potential for voting, it is not a magic wand. If malicious software changes Alex to Bob and invites Bob on an immutable ledger, then we have created an extraordinarily secure permanent record of the wrong vote. Artificial intelligence could make both defense and attack more capable. AI systems can help monitor networks,

39:10inspect logs, detect anomalies, and support election workers. They can also automate phishing, vulnerability discovery, social engineering, and attacks on human beings around the technical system. None of this means remote voting is impossible forever. Estonia already permits internet voting in national elections, and one response to coercion is allowing a voter to cast another electronic ballot later, with a later electronic vote replacing the earlier one, while an in-person paper vote can supersede

39:41the internet vote. Someone watching you vote at home, therefore cannot be certain you not change it afterward. This also helps if two days before the election, a candidate is found to have done something bad, but someone has already cast their vote, they can go in and change it. I should also note that electronic voting of this type probably makes recall elections easier to perform as well, for when people want to change their vote after the election is already over. Future systems may use strong or digital identity, biometrics, dedicated secure hardware,

40:11and end-to-end verifiable cryptography that lets voters and observers obtain evidence that ballots were recorded and tallied correctly without simply publishing everybody's votes. A big issue is voter confidence in such a system, though, which sadly does not have to match how secure the system actually is. People are not only worried about whether their own vote can be changed, they are worried about fabricated votes, or about people less careful than themselves getting tricked, or hacked, and given the sheer power and budgets available to governments,

40:42they will always be a target for that, and is a lot cheaper than a military invasion. So the most likely path may not be everyone suddenly voting through the same phone app. Remote electronic voting may expand first where distance or accessibility creates the greatest pressure, and gets a good stress test, while ordinary domestic elections remain more conservative and slowly adopt tested and upgraded systems. Secure voting kiosks, supervised remote locations, electronic delivery of blank ballots, or special systems for overseas voters

41:12may grow a lot before unrestricted home voting does. And even if cryptography eventually makes the system extraordinarily secure, we return to the problem I worry about constantly. Can ordinary voters understand why it is secure?

41:28Civilizations can survive natural disasters, technological upheavals, and all sorts of cosmic hazards, but sometimes the thing that does the most damage is simply the person running the place. That makes Mad Kings from real-life lore such a good companion to some of the questions we explore on SFIA. The six-part Nebula original looks at some of the modern era's strangest and most dangerous rulers, dictators, and would-be monarchs. People who gained enormous power and then used it for vanity, brutality, personal crusades, and decisions that would seem excessive

42:00even from a poorly written fictional villain. You can watch the complete Mad Kings series on Nebula, along with every episode of SFIA, early and ad-free. My monthly Nebula exclusives and exclusive collections like Space Industry. Use my link or the QR code on screen to get 50% off an annual plan. Just $30 for the whole year.

Why paper refuses to die

42:21Why Paper Refuses to Die When our office was looking at newer voting equipment some years back, vendors naturally emphasized what their systems could do and what they cost. They really wanted to emphasize the bells and whistles of one new set of voting machines. At one point, my response was to flat-out tell one vendor that even if they gave us those machines for zero dollars, I still would not want them. Our voters like paper ballots. They like taking a pen, physically marking a choice,

42:52being able to clearly read what they picked, feeding the ballot through a scanner, and knowing the ballot will be locked up securely in the event of a recount. People were even particular about the paper itself, make it too thin or flimsy, and some voters feel it does not seem sufficiently official and reliable. Thick, official, and rather expensive paper. From a purely technological perspective, that can sound silly. From an election administration perspective, it's not silly at all. Public confidence is one of the outputs an election system has to produce.

43:22Most of the time, some election board got hacked. It was their barely used Facebook page or website, not something related to actual voting or counting, but the effect of the public is still pretty dire. So too, if I design a mathematically brilliant system that nearly every computer scientist regards is secure, but half the electorate sees it as a mysterious black box, as suspects it can manufacture votes, I have not necessarily improved the election. Paper has remarkable virtues, precisely because it is an unimpressive technology.

43:53A scanner can count thousands of ballots quickly, but if somebody disputes the results, human beings can physically inspect those same ballots and count them again, with no special training or technology. You do not need to understand a cryptographic proof, you just need eyes, and perhaps reading glasses. This is not merely nostalgia. More than 98% of the US election jurisdictions now use systems in which voters mark paper, or which otherwise create an auditable paper record, and the Federal Election Assistance Commission has formally endorsed

44:24paper-based, software-independent voting systems. That points toward, what I suspect is one of the most likely futures, increasing the sophisticated election technology wrapped around a very stubbornly physical core. Registration databases become better, electronic portbooks become more capable, ballot-on-demand printers let voters use flexible vote centers, scanners become faster and more accurate, software automatically reconciles totals and flags anomalies, cryptography may help authenticate records and publish evidence,

44:54artificial intelligence may help officials inspect logs or notice patterns that deserve human attention, like 20 absentee ballots to actual voters somehow getting mailed to one address and coming back with identical handwriting on the envelopes. And after all of that, the authoritative evidence may still be a stack of marked paper. Post-election audits are a good example of technology and statistics strengthening rather than replacing paper. We do one after each election in Ohio, risk limiting audits take the idea further, instead of automatically

45:25recounting every ballot, officials inspect a statistically selected sample of the fiscal record and expand it if necessary until there is strong evidence that the reported outcome matches the ballots. New technology does not always replace the old system, sometimes its best job is helping the old system prove that it worked.

45:47Fundamentally, the deepest challenge is legitimacy. The winner matters, but the people who lose have to believe they genuinely lost fair and square, which is much easier to convince them of if it is also actually true. They do not have to like the result, approve of the winners, or stop arguing with them. The existence of another election later rather assumes they will not. They do need enough confidence in the process to accept that this is the result until they have another opportunity to change it,

46:17and shouting at them that everything works fine is typically not too effective, especially if they can point to errors, which will usually be common enough. That may be the hardest requirement of all to engineer because it does not live entirely inside the voting machine, the paper ballot, the cryptography, the voter database, or the constitution. It lives inside the minds of the voters. The future will give us more ways than ever to collect, count, verify, and translate people's preferences. The real question is which of those new abilities will make elections better

46:48rather than merely more complicated? Because democracy has never really suffered from a shortage of ways to count people's opinions. The harder problem is building a system in which people believe those opinions counted.

47:15everyone talks about how challenging it is to raise kids, but what about protecting them? Outlet covers, kitchen cabinet locks, baby gates, where do you even start? Well, protecting your home in auto is a great place to start. Progressive offers customized coverage options so you can be sure you have the right coverage. They're basically the baby gates of insurance.

47:46Get a quote at Progressive.com. Progressive Casualty Insurance Company, affiliates and other insurers. Not available in all states.