Recently we’ve seen two essays by people I highly respect in the field of AI and robotics. Their points are worthy of reading, but in spite of my respect, I have some differences of course.
The first essay comes from Andrew Ng, head of AI (and thus the self-driving car project) at Baidu. You will find few who can compete with Andrew when it comes to expertise on AI.
In Wired he writes that Self-Driving Cars Won’t Work Until We Change Our Roads—And Attitudes. And the media have read this essay as being much more strong about changing the roads than he actually writes. I have declared it to be the “first law of robocars” that you don’t change the infrastructure. You improve your car to match the world you are given, you don’t ask the world to change to help your cars. There are several reasons I promote this rule:
As soon as you depend on a change in the world in order to drive safely, you have vastly limited where you can deploy. You declare that your technology will be, for a very long time, a limited area technology.
You have to depend on, and wait for others to change the world or their attitudes. It’s beyond your control.
When it comes to cities and infrastructure, the pace of change is glacial. When it comes to human behaviour, it can be even worse.
While it may seem that the change to infrastructure is clearer and easier to plan, the reality is almost assuredly the opposite. That’s because the clever teams of developers, armed with the constantly improving technologies driven by Moore’s law, have the ability to solve problems in a way that is much faster than our linear intuitions suggest. Consider measuring traffic by installing tons of sensors, vs. just getting everybody to download Waze. Before Waze, the sensor approach seemed clear, if expensive. But it was wrong.
As noted, Andrew Ng does not actually suggest that much change to the infrastructure. He talks about:
Having road construction crews log changes to the road before they do them
Giving police and others who direct traffic a more reliable way to communicate their commands to cars
Better painting of lane markers
More reliable ways to learn the state of traffic lights
Tools to help humans understand the actions and plans of robocars
The first proposal is one I have also made, because it’s very doable, thanks to computer technology. All it requires at first blush is a smartphone app in the hands of construction crews. Before starting a project, they would know that just as important as laying out cones and signs is opening the app and declaring the start of a project. The phone has a GPS and can offer a selection of precise road locations and log it. Of course, the projects should be logged even before they begin, but because that’s imperfect, smartphone logging is good enough. You could improve this by sticking old smartphones in all the road construction machines (old phones are cheap and there are only so many machines) so that any time a machine stops on a road for very long, it sends a message to a control center. Even emergency construction gets detected this way.
Even with all that, cars still need to detect changes to the road (that’s easy with good maps) and cones and machines. Which they can do.
I think the redirection problem is more difficult. Many people redirect traffic, even civilians. However, I would be interested to see Ng’s prediction on how hard it is to get neural network based recognizers to understand all the common gestures. Considering that computers are now getting better at reading sign languages, which are much more complex, I am optimistic here. But in any event, there is another solution for the cases where the system can’t understand the advice, namely calling in an operator in a remote control center, which is what Nissan plans to do, and what we do at Starship. Unmanned cars, with no human to help, will just avoid data dead zones. If somehow they get to them, there can be other solutions, which are imperfect but fine when the problem is very rare, such as a way for the traffic manager to speak to the car (after all, spoken language understanding is now close to a solved problem for limited vocabulary problems.)
Here I disagree with Andrew. His statement may be a result of efforts to drive on roads without maps, even though Baidu has good map expertise. Google’s car has a map of the texture of the road. It knows where the cracks and jagged lane markers are. The car actually likes degrading lane markers. It’s perfectly painted straight and smooth roads which confuse it (though only slightly, and not enough to cause a problem.) So no, I think that better line painting is not on the must-do list.
He’s right, seeing lights can be challenging, though the better cars are getting good at it. The simple algorithm is “you don’t go if you don’t confirm green.” That means you don’t run a red but you could block traffic. If that’s very rare it’s OK. We can consider infrastructure to solve that, though I’m wary. Fortunately, if the city is controlling its lights with a central computer, you don’t have to alter the traffic light itself (which is hard,) you can just query the city, in those rare cases, for when the light will be changing. I think that problem will be solved, but I also think it may well be solved just by better cameras. Good robocars know exactly where all the lights are, and they know where they are, and thus they know exactly what pixels in a video image are from the light, even if the sun is behind it. (Good robocars also know where the sun is and will avoid stopping in a place where there is no light they can see without the sun right behind it.)
Working with people
How cars interact with people is one of Andrew Ng’s points and the central point of Rodney Brooks’ essay Unexpected Consequences of Self Driving Cars. Already many of the car companies have had fun experimenting with that, putting displays on the outside of cars of various sorts. While cars don’t have the body language and eye contact of human drivers, I don’t predict a problem we can’t solve with good effort.
Brooks’ credentials are also superb, as founder of iRobot (Roomba) and Rethink Robotics (Baxter) as well as many accomplishments as an MIT professor. His essay delves into one of the key questions I have wondered about for some time — how to deal with a world where things do not follow the rules, and where there are lots of implicit and changing rules and interactions. Google discovered the first instant of this when their car got stuck at a 4 way stop by being polite. They had to program the car to assert its right to go in order to handle the stop. Likewise, you need to speed to be a good citizen on many of our roads today.
His key points are as follows:
There is a well worked out dance between pedestrians and cars, that varies greatly among different road types, with give and take, and it’s not suitable for machines yet.
People want to know a driver has seen them before stepping near or certainly in front of a vehicle.
People jaywalk, and even expect cars to stop for them when they do on some streets.
In snowy places, people walk on the street when the sidewalk is not shoveled.
Foot traffic can be so much that timid cars can’t ever get out of sidestreets or driveways. Nice pedestrians often let them out. They will hand signal their willingness to yield or use body language.
Sometimes people just stand at the corner or edge of the road, and you can’t tell if they are standing there or getting ready to cross.
People setting cars to circle rather than park
People might jump out of their car to do something, leaving it in the middle of the street blocking traffic, where today they would be unwilling to double park.
People might abuse parking spots by having a car “hold” them for quick service when they want to leave an event.
Cars will grab early spots to pick up children at schools.
Brooks starts with one common mistake — he has bought into the “levels” defined by SAE, even claiming them to be well accepted. In fact, many people don’t accept them, especially the most advanced developers, and I outlined recently why there is only one level, namely unmanned operation, and so the levels are useless as a taxonomy. Instead the real taxonomy in the early days will be the difference between mobility on demand services (robotaxi) and self-drive enabled high end luxury cars. Many of his problems involve privately owned cars and selfish behaviour by their owners. Many of those behaviours don’t make sense in a world with robotaxis. I think it’s very likely that the robotaxis come first, and come in large numbers first, while some imagine it’s the other way around.
Brooks is right that there will be unintended consequences, and the technology will be put to uses nobody thought of. People will be greedy, and antisocial, that can be assured. Fortunately, however, people will work out solutions, in advance, to anything you can think of or notice just by walking down the street or thinking about issues for a few days. The experienced developers have been thinking about these problems for decades now, and cars like Google’s have driven for 300 human lifetimes of driving, and that number keeps increasing. They note every unusual situation they encounter on every road they can try to drive, and the put it into the simulator if it’s important. They’ve already seen more situations than any one human will encounter on those roads, though they certainly haven’t driven all the types of road in the world. But they will, before they certify as safe for deployment on such roads.
As I noted, only the “level 4” situation is real. Level 5 is an aspirational science-fiction goal, and the others are unsafe. Key to the improved thinking on “levels” it is no longer the amount of human supervision needed that makes the difference, it is the types of roads and situations you can handle. All these vehicles will only handle a subset of roads, and that is what everybody plans. If there is a road that is too hard, they just won’t drive it. Fortunately, there are lots of road subsets out there that are very, very useful and make economic sense. For a while, many companies planned only to do highways, which are the simplest road subset of all, except for the speed. A small subset, but everybody agrees it’s valuable.
So the short answer is, solutions will be found to these problems if the roads they occur on are commercially necessary. If they are not necessary, the solutions will be delayed until they can be found, though that’s probably not too long.
As noted above, many people do expect systems to be developed to allow dialogue between robocars and pedestrians or other humans. One useful tool is gaze detection — just as a cheap flash camera causes “red eye” in photos, machines shining infrared light can easily tell if you are looking at them. Eye contact in that direction is detectable. There have been various experiments in sending information in the reverse direction. Some cars have lasers that can paint lines on the road. Others can display text. Some have an LED ribbon surrounding them that shows all the objects and people tracked by the car, so people can understand that they are being perceived. You can also flash a light back directly at people to return their eye contact — I see you and I see that you saw me.
Over time, we’ll develop styles of communication, and they will get standarized. It’s not essential to do that on day one; you just stay on the simpler roads until you know you can handle the others. Private cars will pause and pop out a steering wheel. Services like Uber will send you a human driver in the early days if the car is going somewhere the systems can’t drive, or they might even let you drive part of it. Such incrementalism is the only way it can ever work.
People taking advantage of timidity of robocars
I believe there are solutions to some of the problems laid out. One I have considered is pedestrians and others who take advantage of the naturally conservative and timid nature of a robocar. If people feel they can safely cut off or jaywalk in front of robocars, they will. And the unmanned cars will mostly just accept that, though only about 10% of all cars should be unmanned at any given time. The cars with passengers are another story. Those passengers will be bothered if they are cut off, or forced to brake quickly. They will spill their coffee. And they will fight back.
Citizen based strong traffic code enforcement
Every time you jump in front of such a car, it will of course have saved the video and other sensor data. It’s always doing that. But the passenger might tell the car, “Please save that recent encounter. E-mail it to the police.” The police will do little with it at first, but in time, especially since there are rich people in these cars, they will throw a face recognizer and licence plate recognizer on the system that gets the videos. They will notice that one person keeps jaywalking right in front of the cars and annoying the passengers. Or the guy who keeps cutting off the cars as though they are not there because they always brake. They will have video of him doing it 40 times, or 100. And at that point, they will do something. The worst offender will get identified and get an E-mail from police. We have 50 videos of you doing this. Here are 50 tickets. Then the next, and the next until nobody wants to get to the top of the list.
This might actually create pressure the other way — a street that belongs only to the cars and excludes the non-car user. A traffic code that is enforced to the letter because every person inconvenienced has an ability to file a complaint trivially. We don’t want that either, but we can control that balance.
I actually look forward to fixing one of the dynamics of jaywalking that doesn’t work. Often, people like to jaywalk and a car is approaching. They want to have the car pass at full speed and then walk behind it — everybody is more comfortable behind a car than in front of one. But the driver gets paranoid and stops, and eventually you uncomfortably cross in front, annoyed at that and that you stopped somebody you didn’t intend to stop. I suspect robocars will be able to handle this dynamic better, predicting when people might actually be on a path to enter their lane, but not slowing down for stopped pedestrians (adults at least) and trust them to manage their crossing. Children are a different matter.
People being selfish with robocars
Brooks wonders about people doing selfish things with their robocars. Here, he mostly talks about privately owned robocars, since most of what he describes would not or could not happen with a robotaxi. There will be some private cars so we want to think about this.
A very common supposition I see here and elsewhere is the idea of a car that circles rather than parking. Today, operating a car is about $20/hour so that’s already completely irrational, and even when robocar operation drops to $8/hour or less, parking is going to be ridiculously cheap and plentiful so that’s not too likely. There could be competition for spots in very busy areas (schools, arenas etc.) which don’t have much space for pick-up and drop-off, and that’s another area where a bit of traffic code could go a long way. Allow facilities to make a rule: “No car may enter unless its passenger is waiting at the pick-up spot” with authority to ticket and evict any car that does otherwise. Over time, such locations will adjust their pick-up spots to the robocar world and become more like Singapore’s airport, which provides amazing taxi throughput with no cab lines by making it all happen in parallel. Of course, cars would wait outside the zone but robocars can easily double and triple park without blocking the cars they sit in the path of. Robocars waiting for passengers at busy locations will be able to purchase waiting spaces for less than the cost of circling, and then serve their customers or owners. If necessary, market prices can be put on the prized close waiting spaces to solve any problems of scarcity.
So when can it happen?
Robocars will come to different places at different times. They will handle different classes of streets at different times. They will handle different types of interactions with pedestrians and other road users at different times. Where you live will dictate when you can use it and how you can use it. Vendors will push at the most lucrative routes to start, then work down. There will be many problems that are difficult at first, and the result will be the early cars just don’t go on those sorts of streets or into those sorts of situations. Human driving, either by the customer or something like an Uber driver, will fill in the gaps.
Long before then, teams will have encountered or thought of just about any situation you’ve seen, and any situation you’ve likely thought of in a short amount of time. They will have programmed every variation of that situation they can imagine into their simulators to see what their car does. They will use this to grow the network of roads the cars handle every day. Even if at the start, it is not a network of use to you, it won’t be too long before it becomes that, at first for some of your rides, and eventually for most or all.
CES has become the big event for major car makers to show off robocar technology. Most of the north hall, and a giant and valuable parking lot next to it, were devoted to car technology and self-driving demos.
Gallery of CES comments
Earlier I posted about many of the pre-CES announcements and it turns out there were not too many extra events during the show. I went to visit many of the booths and demos and prepared some photo galleries. The first is my gallery on cars. In this gallery, each picture has a caption so you need to page through them to see the actual commentary at the bottom under the photo. Just 3 of many of the photos are in this post.
To the left you see BMW’s concept car, which starts to express the idea of an ultimate non-driving machine. Inside you see that the back seat has a bookshelf in it. Chances are you will just use your eReader, but this expresses and important message — that the car of the future will be more like a living, playing or working space than a transportation space.
The main announcement during the show was from Nissan, which outlined their plans and revealed some concept cars you will see in the gallery. The primary demo they showed involved integration of some technology worked on by Nissan’s silicon valley lab leader, Maarten Sierhuis in his prior role at NASA. Nissan is located close to NASA Ames (I myself work at Singularity University on the NASA grounds) and did testing there.
Their demo showed an ability to ask a remote control center to assist a car with a situation it doesn’t understand. When the car sees something it can’t handle, it stops or pulls over, and people in the remote call center can draw a path on their console to tell the car where to go instead. For example, it can be drawn how to get around an obstacle, or take a detour, or obey somebody directing traffic. If the same problem happens again, and it is approved, the next car can use the same path if it remains clear.
I have seen this technology a number of places before, including of course the Mars rovers, and we use something like it at Starship Technologies for our delivery robots. This is the first deployment by a major automaker.
Nissan also committed to deployment in early 2020 as they have before — but now it’s closer.
You can also see Nissan’s more unusual concepts, with tiny sensor pods instead of side-view mirrors, and steering wheels that fold up.
Several startups were present. One is AIMotive, from Hungary. They gave me a demo ride in their test car. They are building a complete software suite, primarily using cameras and radar but also able to use LIDAR. They are working to sell it to automotive OEMs and already work with Volvo on DriveMe. The system uses neural networks for perception, but more traditional coding for path planning and other functions. It wasn’t too fond of Las Vegas roads, because the lane markers are not painted there — lanes are divided only with Bott’s Dots. But it was still able to drive by finding the edge of the road. They claim they now have 120 engineers working on self-driving systems in Hungary. read more »
You may have seen a lot of press around a dashcam video of a car accident in the Netherlands. It shows a Tesla in AutoPilot hitting the brakes around 1.4 seconds before a red car crashes hard into a black SUV that isn’t visible from the viewpoint of the dashcam. Many press have reported that the Tesla predicted that the two cars would hit, and because of the imminent accident, it hit the brakes to protect its occupants. (The articles most assuredly were not saying the Tesla predicted the accident that never happened had the Tesla failed to brake, they are talking about predicting the dramatic crash shown in the video.)
The accident is brutal but apparently nobody was hurt.
The press speculation is incorrect. It got some fuel because Elon Musk himself retweeted the report linked to, but Telsa has in fact confirmed the alternate and more probable story which does not involve any prediction of the future accident. In fact, the red car plays little to no role in what took place.
Tesla’s autopilot uses radar as a key sensor. One great thing about radar is that it tells you how fast every radar target is going, as well as how far away it is. Radar for cars doesn’t tell you very accurately where the target is (roughly it can tell you what lane a target is in.) Radar beams bounce off many things, including the road. That means a radar beam can bounce off the road under a car that is in front of you, and then hit a car in front of it, even if you can’t see the car. Because the radar tells you “I see something in your lane 40m ahead going 20mph and something else 30m ahead going 60mph” you know it’s two different things. read more »
Thursday night I am heading off to CES, and it’s become the main show it seems for announcing robocar news. There’s already a bunch.
BMW says it will deploy a fleet of 40 cars in late 2017
Bumping up the timetables, BMW has declared it will have a fleet of 40 self-driving series 7 cars, using BMW’s technology combined with MobilEye and Intel. Intel has recently been making a push to catch up to Nvidia as a chipmaker supplier to automakers for self-driving. It’s not quite said what the cars will do, but they will be trying lots of different roads. So far BMW has mostly been developing its own tech. More interesting has been their announcement of plans to sell rides via their DriveNow service. This was spoken of a year ago but not much more has been said.
Intel also bought 15% of “HERE” the company formerly known as Navteq and Nokia. Last year, the German automakers banded together to buy HERE from Nokia and the focus has been on “HD” self-driving maps.
Hyundai, Delphi show off cars
There are demo cars out there from Delphi and a Hyundai Ioniq. Delphi’s car has been working for a while (it’s an Audi SUV) but recently they have also added a bunch of MobilEye sensors to it. Reports about the car are good, and they hope to have it ready by 2019, showing up in 2020 or 2021 cars on dealer lots.
Toyota sticks to concepts
Toyota’s main announcement is the Concept-i meant to show off some UI design ideas. It’s cute but still very much a car, though with all the silly hallmarks of a concept — hidden wheels, strangely opening doors and more.
Quanergy announces manufacturing plans for $250 solid state LIDAR
Quanergy (Note: I am on their advisory board) announced it will begin manufacturing this year of automotive grade $250 solid state LIDARs. Perhaps this will stop all the constant articles about how LIDAR is super-expensive and means that robocars must be super-expensive too. The first model is only a taste of what’s to come in the next couple of years as well.
New Ford Model has sleeker design
Ford has become the US carmaker to watch (in addition to Tesla) with their announcement last year that they don’t plan to sell their robocars, only use them to offer ride service in fleets. They are the first and only carmaker to say this is their exclusive plan. Just prior to CES, Ford showed off a new test model featuring smaller Velodyne pucks and a more deliberate design.
I have personally never understood the desire to design robocars to “look like regular cars.” I strongly believe that, just like the Prius, riders in the early robocars will want them to look distinctive, so they can show off how they are in a car of the future. Ford’s carm based on the Fusion hybrid, is a nice compromise — clearly a robocar with its sensors, but also one of sleek and deliberate design.
Nvidia keeps its push
Nvidia has a new test car they have called BB8. (Do they have to licence that name?) It looks fairly basic, and they show a demo of it taking somebody for a ride with voice control, handling a lot of environments. It’s notable that at the end, the driver has to take over to get to the destination, so it doesn’t have everything, nor would we expect it. NVIDIA is pushing their multi-GPU board as the answer to how to get a lot of computing power to run neural networks in the car.
Announcements are due tomorrow from Nissan and probably others. I’ll report Friday from the show floor. See you there.
These matters are studied both by statisticians, who focus on the science of measurement, particularly of things about groups, and election theorists, who also are interested in that but add the study of votes/polls which do not deliberately sample a subset of a population, but attempt to consider the will of the entire group. Both of them are highly concerned about how to deal with the fact a substantial fraction of the population may not participate.
One way to look at the difference is to consider this: An election is not supposed to be just a measurement. It is that, but more than that it is an action. It is the actual enactment of the will of the voters. While there are government officials who count the votes and report on them, a person is not put into office by those officials. Rather, it is the voters who put the candidate into office through their votes. (In Canada, it’s different. The Queen and her Governor-General technically have the legal power, and they observe how the people voted and invite the winner to form a government in the Queen’s name.)
Because voting is an act, rather than just an expression of opinion, we have come to deal with the non-participators as still acting. By not registering to vote or not showing up, they have still taken an action; they have deferred to the others to select the winner.
We tolerate this, though we don’t like it. Low turnouts reduce confidence in the results, and they also mean that election results can be more easily manipulated through “get out the vote” efforts. On the other hand, we get quite upset when people don’t vote for other reasons outside their own will, particularly if somebody else impeded their ability to vote, or manipulated them into not voting. Both voting and not voting must be acts of the free person.
Election theorists join with statisticians in some ways. All are interested in making sure that the aggregate will that comes from counting the votes most accurately reflects the aggregate will of the voters. We debate the merits of different counting systems. Many feel that multi-candidate ballots/preferential ballots do a much better job than first-past-the-post plurality systems. But in all case the counting system is simply the means of calculating the voters’ will so it can be enacted.
In the US Presidential elections, in spite of what is written on the ballot, the voters are appointing a slate of members of the electoral college. This is done independently in each state. In the swing states, all is as you would expect. Candidates campaign. Major efforts are made to woo voters and to get voters to come out. Voters go to the polls knowing and expecting that their will shall be done. They expect they might be part of the group which gets to designate the slate of electors.
In the safe states, it’s very different. In these states, who the electors will be is already well established from polls and the historical patterns of the state. The voters will picks the electors, but it’s a foregone conclusion. Nobody campaigns. There are no major efforts to get out the vote. There will be other races on the ballots which will bring out voters, who will vote within the known constraints. A decent chunk of voters will also show up because “this is how we do things” and together the knowledge that this will happen seals the fate of the state. On top of that, in the safe states, one knows that if things got so far outside the predicted norms as to make the vote actually close, then long ago the election will already have gone to the unexpected party, which in that situation will win all the swing states and victory. This is particularly true on the west coast, where the result is almost always decided before the polls close, and will certainly be decided long before that in a strange situation. If today’s California came close to going Republican, the rest of the USA would also be going so Republican that California’s shift can’t matter.
People know this, and this makes a big difference. A vote in California is technically an action, but only technically. It’s technically a vote but that’s an illusion. In reality, it can never change the result. It’s only for show. The candidates know it too. Because of that a lot of people don’t even register, and a lot stay home. The vote in California is not an election, but only a measurement. A survey. All it ever does is change the number printed in the paper.
Statisticians know all about surveys. They can be pretty good at measuring aggregate opinion if done well, but it is hard to do them well. The problem is what we call sampling bias. In an election, not voting is an implicit action. In a survey, not participating is just not participating. When there is nothing to gain or lose from participating or not participating, the motivations are different.
In 2016, the average swing state Presidential turnout was 64.6% of eligible voters. California’s turnout was 56.1%, just under the 56.6% average of the safe states. In Hawai`i, which knows the election is always decided before it votes (pretty much always for Democrats) the turnout was 41.7% A lot of people don’t show up.
This turns the safe-state votes into something closer to a self-selected survey. Millions are not voting, and those who are voting do so for other reasons than to enact their will. The self-selected survey is the most common class of what is also called the “non-scientific survey.” The name is intended to be derisive. It is easy to jump to false conclusions from a self-selected survey.
It isn’t that simple of course. The vote in safe states is a mix of actual polling and self-selection. As noted, there are people coming to vote on other races. We know how many of those there are. Turnout in off-year elections is around 40%, sometimes worse. And, as we can see, a lot of people show up because there is a Presidential race, in spite of the lack of power in their votes. Some do it from duty. Some from the excitement of a Presidential race. Many do not understand the impotence of their vote, and certainly many do not look at it the way it is described in this article, with a statistician’s eye. So many are voting as though their vote counted. Many have studied the race in detail, as though their vote counted. I can’t even vote and I study it as deeply as any.
But some vote very differently because they know their vote lacks power. Around 9 million don’t vote at all, who would have voted if they were in swing states. Almost surely many millions of those who do vote will do it differently than they might if their vote counted. But there is also no denying that a considerable majority of the voters are treating their vote as just as real, voting just as they would if it could change things. But a considerable majority is not enough. As long as a large group — even if it’s a small minority, even just 5% — are altering or withdrawing their votes, the total loses scientific validity, and has much larger error bars on it.
It is worth noting that by the normal definitions of a popular vote election, it is invalid to add the results of two distinct elections. There is no question that the Presidential elector selections of each state are distinct elections, run by the states. Even on that grounds you can’t add them and treat it as a popular vote. Because ballots replace the actual candidates (slates of electors pledged to the candidates) with the names of candidates, it makes people forget that they are two distinct elections. Thus it becomes necessary to understand how they are not just distinct because they are in different states, but because they operate on different principles as well.
This is why I wrote that, in spite of the fact that it is possible to sum up the votes cast in the 51 different electoral college contests and call it the popular vote, it is nonsensical to do so. You can’t add the totals from people who were voting with the full power of voters in a popular vote election to the totals from people who were participating in a voluntary survey. Aside from the real accuracy problems of the latter class, they are just different things. They can be added on a calculator, but to do so is to announce a misleading number, a meaningless one. You can call it “the popular vote” but it is not like a real popular vote, the kind used in all the other elections of the USA and most of the rest around the world. Calling it the popular vote makes many people — we’ve seen this — think it has a winner and a loser. They think it has meaning. They think it supports or questions the legitimacy of the winner of the electoral college. Since real popular votes are, in our modern democratic world, seen as superior to systems like the electoral college, calling it “the popular vote” implies to many people that it is superior, when in fact it’s meaningless. It would only be superior if it were an actual popular vote election like the others.
The common statistic reported after the US election was that Clinton “won the popular vote” by around 3 million votes over Trump. This has caused great rancour over the role of the electoral college and has provided a sort of safety valve against the shock Democrats (and others) faced over the Trump victory.
I’m here with concerning analysis, which I offer because it is a mistake on the part of the US left to underestimate the magnitude of Trump’s victory, or to imagine it was only because of a flaw in the system which he gamed better than Clinton.
The problem is that the US does not officially have a thing called “the popular vote.” That exists nowhere in its rules. There is no popular election of the President. Rather, there 54 elections with popular votes in 51 jurisdictions, which newspaper reporters then sum up into a number they incorrectly describe as “the national popular vote.” Of course, Clinton did win that invalid sum by around 3M votes. But bad statistical practice by the press, though it has created a common convention — for many decades — of calling that number “the popular vote,” does not make it valid. True popular votes involve all voters being free and equal, and we criticise any foreign election that pretends to call itself a popular vote when the voters are not free and equal. A popular vote, by its proper definition, is the vote total in a single election. Not 54 of them. As such, the sum is no more a popular vote total than adding the results of the 2008 and 2012 votes would get you a popular vote for or against Obama.
It’s especially invalid because it’s really summing two fairly different types of results.
True Popular vote totals from “swing” states where both candidates actively campaigned, turnout was higher, and voters expected their votes to count
Low-accuracy popular vote totals from “safe states” which candidates did not contest, and where voters knew their vote would not change the result
Statisticians will tell you these are two very different animals. We probably wish we knew who would have won the popular vote, if there had been a real national popular vote. Because there was no such vote, the hard answer is we don’t know what its result would be. In particular, with a statistically invalid sum like the published national popular vote, it is incorrect to say one party “won” or “lost.” There is no actual contest to win or lose, and while you can pretend that a higher total is winning, it is not a mathematically valid conclusion.
We do know that in the 16 contested regions, Trump surpassed Clinton in a simple sum by about 500,000 votes. (As you would expect, since he needed to win the swing states to win the college.) In the uncontested states, where the Presidential choice was closer to a self-selected survey than a vote, a sum of those popular votes has her about 3.4M more than Trump. While you can’t add popular votes, each popular vote is a statistic, and you can combine statistics if you follow correct statistical procedures.
There are many factors which will introduce error into the results from non-contested states, making it harder to figure out what the actual popular vote might have been.
Voters knew their votes didn’t matter. Many stayed home; these states had generally lower voter turnout. The states with the lowest turnout (HI, WV, TN, TX, OK, AR, AZ, NM, MS, NY, CA, IN, UT) were generally safe states with large margins. Average turnout in 16 contested states was 65%, in non-contested states 57%.
To get specific, a rough calculation suggests 8 to 9 million more votes would be cast in the non-contested states if they had a 65% turnout. This is a giant disenfranchisement.
The two candidates had the lowest approval ratings ever. Many Clinton voters were not supporting her, but were out to stop Trump. Trump’s ratings were even lower, so many of his voters were only out to stop Clinton. I suggest that in states where you know your vote will not elect or stop anybody, there is less motivation for nose-holding votes.
As noted, campaigns were not active in these states. In some states, like California, Clinton did campaign, though presumably to raise money rather than votes. Having only one candidate campaign skews things more.
More safe state voters felt comfortable voting for 3rd party choices, which they would have been less likely to do in a swing state. Many of the 4.6M votes for 3rd party candidates in safe states may have gone to major party candidates, though in what direction is unknown.
In some safe states, even the downballot races are predetermined, discouraging voters. In California, the election of Democrats in most down-ballot races was assured; the primary was the real contest. (However, contentious ballot propositions can counter this in some states.)
In the end, though, results from a race that everybody agreed didn’t matter are just a different animal from results in a contested race. You can’t add apples and oranges, or perhaps more correctly, oranges and lemons. Different, though not entirely. You can add them and get a total number of citrus (votes of any kind,) but you can’t call it the count of oranges (real votes.)
In spite of the frequent description of the US vote-total as a popular vote, this is at odds with common usage. The thousands of other elections in the USA are actual popular votes, as are the vast majority of elections in free countries. The US national vote sum, and similar sums published in some parliamentary elections, are the rare exception where an official and incorrect tally gets called a popular vote.
A century ago in 1916, women could not vote for President in most of the USA — except for Illinois, which recognized women’s right to vote in Presidential elections in 1913. President Wilson did not support suffrage in 1916 but his opponent, Hughes, did, and suffragettes campaigned for Hughes as a result.
Wilson won, but Hughes won Illinois handily, in fact his margin there of 202,000 votes was his highest in any state (and 2nd highest in the land) — in part because the addition of women to the rolls meant Illinois had more voters than any other state. I have to speculate that this margin had to do with women voting for the candidate ready to defend their basic human rights.
Wilson won the college 277 to 254. And he won the so-called popular vote by 600,000 votes. But that “popular vote” in this case consisted of adding the popular vote from states like Illinois where women were human, and other states where they were less than human. Who can defend adding those totals together, cast under such different rules, and calling it “the popular vote” and declaring that Wilson “won” the popular vote in 1916.
Today, the difference between California and other states is not so dramatic as disenfranchising an entire sex. But because Californians are told their vote for President doesn’t matter, the turnout there was 56% and an average of 65% in the swing states. If California had that average, that’s 2.3 million more voters. Millions disenfranchised not because of their sex, but because the system says their vote doesn’t matter. California’s “popular vote” is a sham, and not too different a sham from that of men-only New York in 1916 or “Dear Leader of course” North Korea today. Oh sure, they have something they call the popular vote in North Korea, but the result is known in advance and nobody thinks their vote counts. (And yes, they know they could be punished if they put their ballot in the wrong box.)
You could not add the votes of Illinois and New York in 1916 and call it a true popular vote. You can’t add the results of California’s sham popular vote to Florida’s real popular vote and call it a true popular vote. I mean, people do that, but they should not.
Can we figure it out?
All this said, you could attempt to measure what the vote would have been. We may not have enough data, but we could make some estimates. We know that Clinton led Trump by 3.5% in national polls before the election, but we also know that Trump outperformed those polls by 1.5-6% in many contested states. To really do this would require much more careful analysis than you see in this paragraph, which is written only to show one extreme of what’s possible, and the difference is almost surely less than this from these two states. Full analysis would require looking at detailed voting and polling patterns and an understanding of what motivates people to stay home or vote differently in safe states. vs. swing states, and an understanding of how Trump outperformed his polls so broadly in the contested states. In the other direction, since the 8-9 million missing voters in the safe states are in states that swing Democratic, there are arguments Clinton’s total could have been even higher. However, even with that analysis we still would not really know.
My intuition is that such a result would show Clinton scoring higher than Trump, but not by 3M votes. And the margin of error would include results where Trump wins that popular vote, but this would be the outside condition. Certainly the only hard data on states that were actually contested has him win if extrapolated, but the Democratic party dominance in the big uncontested states is very strong. Also not factored in this is the effect of voter suppression techniques.
I should note to non-regular readers that I am anti-Trump. At the same time, having been shocked several times by underestimating his support, I write this because this underestimation must stop, and both sides need to come to much better understanding of how people voted for or against them, and why.
A slightly better approach would be to publish vote totals divided between swing and safe states. Because situations differ so much in the safe states, this is still not super accurate, but it’s a lot better. (I built this from an earlier download so numbers may not match final totals exactly.)
Clinton Trump Johnson Stein McMillin Others
Swing Total 25,946,624 26,423,193 1,783,571 434,433 203,500 351,415
Safe Total 40,582,344 37,227,033 2,770,706 1,031,304 435,055 468,484
It is interesting to note how much better Stein did in Safe states, 130% better. Johnson did 50% better, Clinton 55% more and Trump 38% more
So what should the popular vote be?
One might argue that in an ideal democracy, the popular vote would represent the aggregate view of all voters. Some nations make voting mandatory in order to get this. Australia gets 95% turnout using this technique, but Malta, New Zealand and several other countries get turnout around 90% without legal compulsion.
It might even be argued that a truly ideal democracy would not only have everybody vote, but have everybody study the choices to make an informed vote. We don’t get any of these ideals, and so in the USA it has come to be accepted that the popular vote is the vote totals from those who took the time to show up. The low turnout enables both voter suppression efforts and gives extreme value to successful “get out the vote” efforts, since it is far cheaper to convince a weak supporter to show up than to convince an undecided voter to swing your way.
Some election theorists have actually proposed that the best way to do elections would be to use a random sample, sometimes combined with strong incentives for members of this sample to vote, and possibly to also learn before voting. This seems strange to non-mathematicians but actually has strong validity. (In one variant, the selected electors are known weeks in advance and the campaigns and public interest groups focus their attention on “educating” them, in which case the number must be large so that truly personal targeting is not effective.) In a nation with 90% turnout these techniques make elections much cheaper but don’t affect results much. In a country with 60% turnout which switches to 99% turnout from the randomly selected electors, the result becomes a much more accurate measure of voter will than the current system.
It is also worth noting that the entire popular vote system for President is not in the US constitution, and so alternate systems, including sampling, actually are legally possible if states willed it, though politically unlikely. There are many advantages to sampling: Close to 100% turnout, more informed voters, the possible reduction of massive campaign spending and fundraising and the elimination of voter suppression. Its main disadvantage is that it doesn’t match non-mathematician’s instincts about how an election should work, and the added risk of corruption of the random selection.
In order to get a real popular vote, even one where we total the will of the 60% who show up, it is necessary to get rid of the college. The college could be nullified by a pact between California, Texas and two other large size republican safe states. If just those 4 states agreed to cast all their electors according to a popular vote result, it would be sufficient to make the college match that popular vote. Once it was known that this was the case, all voters would now know their vote counted, and all candidates would campaign in all states instead of just swing states, and we would have a true popular vote result.