Hysteresis voting

Our founders held this as an axiom: government should reflect the will of the people. A government which does not is a tyranny, and should be torn down and replaced. Those who make peaceful revolution impossible will make violent revolution inevitable.

But our founders were also right to be concerned about the tyranny of the mob; in a direct democracy, if you can convince 50.0001% of the people to do break the system, it's broken forever. There have to be limits in place to slow things down, and ways to fix mistakes when they happen. That's why we have a representative democracy instead of a direct one. If the United States had been a direct democracy, then the 9/11 attacks might have resulted in misdirected nuclear retaliation and millions of innocent deaths. It's very possible that only our representative democracy kept that from happening.

But what if there's a terrorist attack two days before an election?

That might change the outcome, right? Perhaps Nathan Petrelli or Donald Trump wins instead of losing, and the course of the country is changed for at least the next several years. Maybe we invade Iraq again. Maybe we do something much worse, something we regret for centuries. And even as the people calm down, there's no chance for years to remove the horrible people we elected.

But if the same attack happened two days after the election? Now Nathan loses, and we go on an entirely different course for the next six years.

What's different between those scenarios? The attack happens identically. The people of the country react identically. They're still being asked the same questions on election day. The only difference is timing. Ask on the wrong day, and you set the course of the country for years to come, with no chance to undo it until the next time you ask the people what they want.

A similar thing happens in cases like Brexit. The way people vote is actually dependent on how they expect the outcome to go. Once people see the outcome of the referendum, they might change their vote (or lack of vote), but now it's too late. People are convinced their votes don't matter, so on the rare occasion they do matter, they're left wanting a second chance. The will of the people suddenly changed, but it only changed after the vote, so the government is left unable to respond to the change.

Signal processing and controls engineers, do you recognize this? It's a sampling problem! The people want the government to behave in a certain way, which changes over time. That's our setpoint we're trying to hit. The setpoint is able to change rapidly (even in response to its own sampling!) but it's only being sampled once every 2-6 years. If you tried to build an actual control system like that, you'd be fired!

Nyquist says that if you're sampling once every six years, the thing you're sampling can't change any faster than every twelve years! Does that describe the will of the people? Especially after something like 9/11? Absolutely not. To avoid getting a government stuck at the extremes, we need a higher sampling rate, to respond more quickly to the will of the people changing.

But that can't be the only change, of course. Having a government that's able to calm down with the people is good, but it also means having one able to get angry with them. Instant response to the momentary will of the people after a terrorist incident would be a disaster! Like in any engineered system, no matter how fast your sample rate is, you still have to have some filtering to slow down response to a reasonable level, or it will go unstable very quickly. The simplest way to achieve this filtering is called hysteresis.

So here's what we do: vote all the time, on everything. Polling stations are open 24/7, with the same set of proposals and candidates on them. Every month the vote totals reset, and everyone can go vote again. This has some immediate effects: any popular vote resolution can be changed at any time, and every elected official is facing a recall election, all the time. That's a recipe for chaos... but here's where the hysteresis comes in!

The catch is that you need a super-majority to affect change. Before an item is put on the ballot, you define some threshold for the change to be executed. Suppose the threshold for changing the mayor is 10%. Every month, you see how many people want a new mayor. Add up the percentage by which that position wins or loses, month by month. If that sum ever gets up to 10%, you get a new mayor.

So if there's one month that the vote is 55/45 for replacing the mayor, you do so immediately, because you got a ten percent difference. But say the outcome is 54/46; you only have eight of your ten required points. You have to wait another month, but that month you only need 51/49 to get that last two percent. This means that the more angry the people are at an elected official, the easier it is to remove them quickly.

But it also means that every month, you have some idea how much closer or further away that recall might be. If the embattled mayor's supporters happened to stay home the first month, maybe they'll come out the second month once they see he's in trouble. Perhaps that second month, the vote is 48/52, four points in favor of keeping him. Now the mayor is six points from being recalled instead of two, and his total will keep changing over succeeding months as more people show up to vote. You have continuously running polling of elected officials, giving a real-time approval rating, with actual consequences!

Now, what about the end of a term? Well, under this system, there doesn't have to be any actual end of term! Instead, you just gradually lower the percentage required to remove someone from office. We design the system with a bias towards change over long periods. The first year, our mayor has to have 10% net disapproval to be replaced. The second year, he only needs 8% net to be replaced. After five years, 50/50 is enough. After ten years, he has to maintain 55% approval all the time to avoid replacement. So if you have someone who's actually consistently popular, they can stay in office for a very long time. But it gets harder and harder every time. You get all the advantages of term limits, without the problem of throwing out perfectly good elected officials arbitrarily.

So there you have hysteresis elections. There are a lot of possible details to be worked out, of course.


Advantages

• Poorly timed disasters and demagogues getting a temporary majority don't break everything
• A small majority of the electorate can't flip things back and forth rapidly
• Voters get warnings about changes before they happen, so more people can get out and vote for what they want
• Increase the value of votes and you increase turnout!
• Regret for voters that sit out is reduced
• It's much harder for the government to suppress voters if voting is happening all the time
• Term limits are handled much more organically

Questions

• Cost. Polling now costs at least 20x what it did previously.
• Or you do mail-in ballots or something, with all the security flaws that entails?
• Do we put less-important things on mail-in ballots, and only require in-person appearances past a certain limit?
• Who sets what's on the ballots and what the hysteresis limits are? If they mayor can set his own removal threshold, that's a problem.
• Are the proposals on the ballot and the thresholds also part of the same voting system?

Infrastructure Megaprojects: Communication

Think about what your life would be like without a phone, without television, without internet access, without books, without music. Really sit and consider that for a minute. I'm betting that if you're reading this, you can't even imagine what you'd do with most of your time. Now imagine your life to date without those things.

Will anyone dispute that information is a necessity in this world?

The US information infrastructure is pathetic compared to most of the developed world. But it's fixable! Estimates have Google Fiber costing about $1,500/home to install. Figure 100 million homes in the US, and to wire the country with high-speed fiber would cost something like $150 billion. Even if it's double that, it's trivial on the scale of projects we're talking about. And following the Google Fiber model, it should be possible to supply most households with free high-speed internet access, only charging for higher bandwidth connections.

But it shouldn't stop there. Wired communication is only part of our information consumption. Right now there are a large number of incompatible cellular networks in the country. How much could we save by standardizing those networks on a single interoperable technology? Think about that. With appropriate leasing agreements in place, you could use anyone's tower, and just let the providers haggle over who pays whom on the back end. And once there's a single universal standard, expanding coverage and service becomes much easier and more efficient.

How much would it cost to pay everyone to switch their towers over to a shared technology? Figure there are 200,000 towers in the US, and we want to change out 90% of them to match the rest. At $150,000 per tower, the entire network would cost $30 billion to build from scratch. Assuming the electronics involved are only a tenth the cost of the tower, we're talking about three billion dollars. Chump change. Once a standard was in place, the government would probably spend more than that building additional towers just to improve coverage.

Unfortunately, we're now beyond my technical knowledge. Are there actual technical advantages to Verizon's approach over, say, Sprint's? Is one objectively better? Is there some technical reason what I've proposed is unworkable? I can't say. But anyone who's ever considered switching cell providers knows what I mean when I say that anything to reduce vender lock-in is a good thing.

Oh, and while we're at it, let's get rid of bundling the cost of a phone into my monthly bill. If it's a $600 phone, don't tell me it's a $200 phone with an early termination penalty if I leave before 2029. Just tell me it's a $600 phone. Finance it, pay cash, whatever, but vender lock-in needs to die.

No, that's not a megaproject. But let's do it anyway.

Infrastructure Megaprojects: Arable Land

So now we have energy and fresh water. The next obvious human need is food. Food production comes down to three things: water, land, and fertilizer. We have a solution to get arbitrary amounts of fresh water, and we can develop fertilizer from the leftover potassium from the desalination plants. That leaves land.

Large sections of the United States are desert, and much of the rest is trending that way. Deserts may not all be dead and barren, but they're not particularly useful by human standards. The growth of deserts is a huge problem.

So let's fix them. It's possible to reverse desertification by planting trees. It's counter-intuitive, but think about it this way: plants don't just absorb water, they also release it through their leaves. That means that whatever rain that's fallen, the trees hold it temporarily, then release it back to the environment to rain out again. That means that whatever rain falls in the area stays in the area longer, cycling through the local ecosystem, rather than just evaporating and leaving.

There are about a quarter million square miles of desert in the continental US, and about as much semi-arid land. Figure fifty trees per acre, and that's sixteen billion trees. Sixteen billion trees to increase our useful arable land area by 20% sounds like a pretty good deal!

If you think that number sounds totally unreasonable, think again. During the great depression we planted three billion trees. More recently, seven billion trees have been planted in less than a decade. Moreover, this requires almost totally unskilled labor, so it's a great jobs project.

Of course, once the forests are established, we wouldn't just leave them untouched. Forests are great, and they have all sorts of positive effects on air quality, improving the health of those nearby. But forests aren't the only end goal. Over time we'd need to make some reclaimed areas into farms, taking advantage of the rebuilt soil. But we'd do that in a planned and controlled fashion. We need to make sure that we don't reclaim the deserts, only to recreate them later.

Ballot Order

The democratic party has selected their candidate for governor. He has no public policy statements, website, or twitter feed. He has an inactive Facebook group, a picture of himself with some fish, and his own name misspelled. From all appearances, this is the entirety of his campaign. Yet he won by a 2:1 margin.

How? He was first on the ballot. This isn't the first time.

There are two groups that should learn from this.

First, voters. If you don't know who you prefer in an election, don't just pick someone on the spot! All you do is water down the opinions of the informed voters, the ones that should be making the decision. If you don't have an opinion on one office, just don't vote for that office! Your votes for all the other offices will still count!

Casting a vote, any vote, is something you should take very seriously. If you're not prepared to do that, don't vote.

Second, legislators. Ballot order is fixed, by law, in alphabetical order by last name. Ballot order clearly has a significant effect on outcome, giving some candidates advantage over others. Laws should never, ever help particular candidates. That's undemocratic and unamerican.

Each voter should be presented with the candidates in a different, randomly chosen order. The uninformed voters who just pick the first candidate on the list would cancel each other out. If we're doomed to use these stupid electronic voting machines with no paper records, we should at least use them in a way that makes elections work.

And if anyone tells you that these machines can't be made to put candidates in random order, give the machine to me. I'll fix it for you. I won't even charge. This is not a technical problem. It's a legal problem, and it has a legal solution.

Infrastructure Megaprojects: Water

Water stress is the resource challenge of this decade, and probably a few more to come. Much of the US has been in drought for the last five years, driving up food prices. Some estimates are that this drought has cost the US economy $150 billion dollars each year! Water tables are being drained faster than they can refill, and polluted beyond use. Some bodies of water are being diverted so much that they've become poisonous, or ceased to exist. We need new sources of water, and we need them now.

Unfortunately you can't just create water unless you have a lot of hydrogen lying around. But we're not lacking for water; we're lacking for potable water. We have all the water we need, if we can just clean it up a bit.

We need to build desalination plants to make the seawater drinkable. There are already several in the United States, and quite a few more around the world. This is not a new thing, it's just a question of scale.

How much water are we talking about? Looking at a couple sources, we can estimate that the US uses around 400 billion gallons per day, most of which goes to run power plants and irrigate crops. An average desalination plant (based on Australian installations) could do 60 million gallons per day, consume about 24 MW, and cost $1.8 billion. So to replace every source of fresh water in the US, we're talking about 6,700 desal plants, consuming 160 GW and costing $12 trillion.

Now, that's just an upper limit. There's no need to desalinate every last drop of water we use. Let's scale back a bit, and target 10%, which should be more than enough to relieve the water stress we're seeing. Now we're talking about 670 plants, 16 GW, and $1.2 trillion. That's eminently doable. And the system scales wonderfully. You can build it gradually over time. If you need more water later, you can build more plants.

Figure 3.5% salinity as a general average, so we're talking about having to find a home for 6 million tons of salt every year. That's enough to cause an environmental catastrophe if it's all in one place, so we need to plan for that. Luckily, the US presently produces 7-8 times that much salt in a year, so our economy could obviously absorb it.

Other resources are present in seawater. We'd be extracting 220,000 tons of potassium a year, about 1/5 our present production of potash. This has great possibility for fertilizers, though I can't speak as to the chemistry involved. We'd also get a comparable amount of magnesium, making us the world's fifth largest producer.

The total value of all those extracted solids comes several hundred million dollars a year. Trivial by comparison to the cost of the construction, but still, a nice offset to operating expenditures.

Of course, it's not just the coasts we're worried about; we also need a way to move the desalinated water from the sea to the midlands. We're talking about a huge aqueduct network. We already have quite a bit of experience building such things, but the scale would be unheard-of. Figure a trillion dollars to build the aqueduct network alone.

Now, we could do closed pipes, but I'm not sure that's what we want to do. Perhaps instead we should have open aqueducts, and let the water evaporate as it will. It will condense back out somewhere as rain, giving us the most efficient possible distribution method. Some combination of covered and uncovered aqueducts would probably be best.

Ultimately, the oceans will provide us with the only sustainable source of fresh water on the planet. We'll eventually have to start tapping it, and we're getting to that point.

Infrastructure Megaprojects: Energy

Without energy, nothing happens. That's not a hyperbole; nothing happens if there's not energy. No water is pumped, no food is moved to market, no computers or lights turn on, and come winter we all freeze to death. It seems fitting to start our megaprojects list here.

Electricity is our most efficient means of moving energy from place to place. There are many ways of generating electricity, but most have significant downsides. Fossil fuels pollute to varying degrees, and need continuous exploration to find new sources. (The negative effects of fracking for this purpose are tremendous. But that's another post.) Wind and photovoltaic solar cells are weather-dependent, and thus unreliable for continuous demand. Hydroelectric dams can only be put in a few places.

There are only two developed means of generation which are both emission-free and weather-independent. The first is nuclear. A well-designed and well-maintained nuclear plant is one of the safest means of power generation ever conceived. Adding up all the deaths due to nuclear accidents, those numbers don't come close to the damage of coal plants. Most nuclear accidents in the world have been due to old designs that were not properly fail-safe; there are vastly better designs now. Spent fuel rods can be reprocessed, eliminating most of the waste disposal concerns. And research into thorium reactors could further enhance both safety and pollution concerns.

But there's an even better way. Solar thermal power is completely pollution-free. It has all the upsides of a large-scale photovoltaic plant, and none of the down-sides. It can run at night, doesn't require complex chemical processes to build, and has no lifetime constraint. Right now there are about 1.5 gigawatts of installed solar thermal power in the US, with another 4 GW in planning.

Average electricity consumption in the US is on the order of 500 GW, about two-thirds of which is fossil fuel based. A large solar thermal plant can generate ~300 MW, so about a thousand solar thermal plants could eliminate fossil fuel plants entirely. Cost of construction for solar thermal plants is about $5500/kW, meaning it would cost ~$2 trillion to get the grid entirely off fossil fuels. That's a lot of money, but it's only about 4x the cost of the interstate system. Divide it up over 40 years, and we're talking about $50 billion a year. That's significant, but it's only about 1.5% of federal spending. What we would gain would be far greater than what we would lose.

Ecological benefits are obvious: our particulate and carbon emissions go way down. Economic benefits are high as well, as our fossil fuels now become something we can sell on the world market, rather than something we must burn here just to keep our civilization going. We'd become a huge supplier worldwide, greatly increasing our soft power. Along with this, we should improve our infrastructure links to Canada and Mexico, allowing us to become a net exporter of electricity and helping improve those countries as well.

Now, that's just the macro picture. There would clearly need to be a robust program in place to retrain whatever workers were displaced by the shift. And there would be second-order effects as electricity prices drop, possibly shutting down other plants. There's no changing one thing without changing fifty others, and we'd want to minimize the overall damage as much as possible. But once the shift was over, having a large, distributed, clean, free source of electricity would make the United States and our neighbors far better countries to live in.

Infrastructure Megaprojects: Introduction

The US economy tanked in 2007, and is still lagging significantly behind where we would all like it to be. Right now, I don't want to bicker and argue about who did what to whom. I don't want to argue about what can best be done about it. I'm happy to do those things, just at another time.

Right now, I want to argue from an assumption, as a thought experiment. It's often said that government spending on infrastructure construction is a good way to kickstart the economy. It puts people to work, and the net gains eventually outweigh the immediate costs. So let's start from that premise. If government spending on infrastructure is good stimulus for the economy, what form should that infrastructure spending take?

Oh, there's the obvious collapsing bridges and dams, things that need fixing. But what new could we do? What one-time projects would make the United States a better place to live for centuries to come, like rural electrification or the interstates did? What can we build that is, in a word, awesome?

I'll be talking about several possibilities in a series of posts. Each post will be one answer to the question, "What do people need?"

The Perfect Voting Machine

Ballots must be physical. The ballot should exist as a physical medium, allowing for recount by multiple independent parties.

Ballots must be machine-readable. The ballot must be marked in such a way that it can be counted by machine, for speed, efficiency, and repeatability.

Ballots must be human-readable. The ballot should be marked in such a way that an average human can read it, in case the counting machines are called into question or unavailable.

Ballots should be marked by machine. This prevents human error in ballot-marking in such a way that the ballot becomes invalid.

Ballots should be confirmed by the voter before final casting. Before the ballot is counted and stored, it must be read by the same machines that would count the votes in a recount. This ensures that the ballot is readable, and reduces the possibility of both mechanical and human error marking the ballot.

Each ballot should have the choices printed in a random order. It's been shown that earlier placement on the ballot conveys an advantage in winning the election. This means that whoever writes the laws defining ballot order can give themselves an electoral advantage, which is a clear conflict of interest. Each ballot should randomize the order of the candidates.

No machine, document, or person besides the voter should possess both the voter's identifying information and ballot contents. The ballot must not be marked with any potentially identifying information such as the user's name or a timestamp with precision finer than one hour. No person should see the marked ballot after it is marked besides the voter. Neither the printing nor reading machines should have any knowledge of the voter.

Ballots must be impossible to counterfeit. Exactly as many valid countable ballots should exist as votes are cast. Voters must be prevented from walking in with a pocket full of blank ballots; similarly, false ballots should be impossible to insert after the election ends. Practically, this means all ballots must be marked on-site with unique information that can be confirmed valid, but which is different on every ballot. As a first-guess suggestion, perform a one-way encryption on a timestamp. Then perform a two-way encryption on that, plus GPS coordinates, using an encryption key that is known only to a few high-level election officials. This ensures every valid ballot is unique, and allows each ballot to be tracked to the polling location it was marked, while maintaining timestamp (and thus voter) anonymity.

Ballots should only be issued to registered voters. It should be impossible to issue a ballot without also marking the name of a single registered voter off the roll. Similarly, it should be impossible to mark a name off a roll without issuing a ballot. Each registered voter should be issued a voting card before arriving at the polling location, containing cryptographically unique information to that voter. Only with the presence of that card will a ballot be issued.

Only one ballot should be issued to each voter. After a voter is issued a ballot, their name is marked on the roll. If that voter determines that their ballot was mismarked, they must turn it back in to receive a replacement. No replacement ballots can be issued without the original being returned.

The design and firmware of all machines involved must be open and inspectable. Maintaining voter trust in the system is paramount. Open-source and open-hardware systems ensure that no back doors or remote access is possible, and allow review for flaws by many thousands of coders and engineers.

It should be impossible to lose ballots. The to-the-minute vote count shall be constantly shared via network with the central election office. This creates a check against large numbers of ballots suddenly "disappearing" before being counted.

So here's the process.

1) I receive my voter card in the mail. The card is marked with a crypto-hash of my personal information, making it effectively impossible to fake.

2) I arrive at the polling location and present my card. My card is scanned, marking me from the roll as having voted. (Optionally, some biometric identification may be performed here, to prevent people from voting with others' voting cards.) A ballot is printed with spaces for all races in my district, plus a unique code identifying the ballot as legitimate and from this polling location.

3) I take that ballot to the marking machine. I insert my ballot, manipulate a touchscreen, and the machine marks my ballot for each race as I indicate

4) I take my marked ballot to the reading machine. I insert my ballot, and it tells me who it thinks I voted for. This machine also confirms that my ballot is properly marked with a valid crypto-stamp indicating a legitimate ballot.
4a) I confirm that my ballot is printed correctly, both visually and by machine. The machine keeps my ballot and counts my vote. I get a sticker and leave.
4b) I find an error in my ballot marking. I return to the poll worker, who inserts my ballot into the ballot-printing machine. The machine confirms that my ballot was valid, marks it with information which renders it invalid (including a human-readable timestamp), and issues a new one with new markings. Return to step 3.

What attacks are possible against this architecture? Obviously we have a problem with running out of ink. Perhaps we mark everything with high-power lasers?

What's better than public transportation?

I'm a big fan of public transportation. The ability to survive without owning a car would lead to a tremendous reduction in cost-of-living. Reduction in traffic is in everyone's best interest. And not having to drive every day would let me get a lot more reading done. (Or more realistically, sleep.)

But here in Nashville, the bus system is more or less a joke for much of the city. I live in the city limits, and I could get to work by bus, but I'd have to spend ninety minutes instead of twenty, I'd get there late, and I'd spend more money. The many thousands in the exurbs are pretty much hosed except for the Music City Star, and even then your options once you reach Nashville are limited.

The more I think about it, the more I think the emphasis on trains and buses may be misplaced. Mass transit as a concept has one inherent limitation: each rider wants to stop at only two places, and no others. The more riders there are wanting different stops, the less convenient it gets for everyone involved. But the fewer stops the bus (or train or whatever) makes, the fewer people the bus is convenient for. You want a very high person-to-stop ratio. This only works for high-density end-to-end traffic paths, like an express from a park-and-ride, a commuter train, or a small local circuit in a high-density area.

But what about those of us (and I'd guess we're the majority) who neither live nor work in high-density areas? By definition, we collectively have more stops to make. To keep a high person-to-stop ratio, you have to reduce the number of people per vehicle. Perhaps to, say, five.

I argue that mass carpooling could have more effect getting cars off the road than any imaginable public transportation system. Say we're comparing three options: 60-person buses, 5-person carpools, and the default single-occupant car. If there are six thousand people commuting from Clarksville (to pick a random number and exurb), that's six thousand cars, twelve hundred carpools, or one hundred buses. Obviously both carpools and buses are vast improvements to traffic. And if everyone would ride the busses, they win over carpools. But not everyone will bus, because of the lack of flexibility.

So the next question is, how many people are willing to bus? How many are willing to carpool? And at what point does realistic carpooling get more vehicles off the road than busing? I won't bore you with my algebraic prowess (maybe later), but in our case the answer works out to be pretty interesting: regardless of the number of people involved, if just 25% more people are willing to carpool than are willing to bus, mass carpooling gets more vehicles off the road. This even though a bus holds twelve times more people! Since a carpool is far more convenient than a bus, I'd expect far more than 25% greater ridership.

Now let's consider cost. If you've got 100 busses, that's at least $30,000,000 in capital expenditure. Probably more. Each bus costs around $100/hr to run. Even if you assume they only run four hours a day (two round trips), that's $40,000/day, or $10,000,000 a year in operating costs. Assuming each bus lasts ten years, that's $13,000,000/year to get 5,900 vehicles off the road. This seems like a lot, but consider that adding a lane of interstate between Nashville and Clarksville would cost something like $150,000,000 and take several years.

How about gasoline? A hundred mile round trip at 25 mpg costs $15 a day. That's $22 million a year in gasoline saved by getting those 5900 cars off the road! That's a number so big I almost want to cry.

Further, consider that traffic can add half an hour to your commute each way. One hour a day saved, times six thousand drivers, is 1.5 million man-hours per year. Figure an average wage of $12/hr, and $13,000,000/year starts to sound cheap.  If we come up with a solution that makes a significant reduction in traffic that only costs, say, a million dollars a year, we collectively are coming out way ahead.

So here's the idea: we should pay people to carpool. But not at a flat rate. Put a million dollars in a pot, and declare that that pot will be distributed evenly among everyone who carpools that year, weighted by how many days they do it. Imagine how people would respond to an incentive like that! If only five people carpool all year, boom, easy $200k each. Pay it out more often than once a year, too. Say every two weeks. Within a few months you should reach an equilibrium point where exactly the right number of people are carpooling for the money being offered. After that you can see just how good the system is and how much it's worth.

As an added bonus, you could let people without cars sign up for the system, and basically turn every driver in the city into a government-provided taxi for the carless. The driver gets paid by the number of passengers, so its a win for them. And the car-free individual gets better service than busses.

The problem is making the matches. If you could get our hypothetical six thousand people to put their schedules in a system, a relatively simple computer program could make matches between them and make a huge dent in traffic. There are already such systems, but there's relatively little data in them. Paying people to participate will fix that. And with smart phones becoming ubiquitous, hitching a ride without advance planning becomes easy.

Now, how does this apply to the Amp? To be clear, I’m not presently taking sides in the debate over the Amp. Traffic on West End is abysmal, and something needs to be done. The Amp is, indeed, something. But what alternatives are there? I think this alternative is better from almost every perspective. Let's run the numbers.

Amp is projected to cost $4 million a year to operate, plus the $174 million startup costs. Add some for inevitable overruns, divide that over 20 years, and you get about $14 million annually. A comparable BRT system in Cleveland has ridership of around 14,000 daily. Each rider represents at most one car off the road, but maybe not even that. Depends on whether they count the same person going both directions as one rider or two. So being generous, we’re spending at least a thousand dollars per year to get each individual car off the road.

How about instead, we just pay people to carpool? If someone paid me a thousand dollars a year for my trouble, you can bet I’d be carpooling! Set up a good smartphone-based system to make ride matches, and I guarantee you you’ll get more cars off the road for less money. The result helps all of Nashville, not just one dense strip. And there’s zero construction disruption.

Obviously there’s a lot of variation possible. Who gets paid? The driver? The rider? Both? How do you keep track and minimize gaming the system? I don’t have all the answers. I can tell you that the system has to be set up well from the start; I've seen far too many systems like this half-complete with clearly zero ease-of-use consideration. The problem isn't trivial. But it is solvable, and I think this is what we should be looking at as an alternative to ripping up West End for a few years.

Footnote:

Suppose you have two forms of transit available to people. Cars, which hold 5 people, and busses, which hold 60. Suppose your goal is to get as many vehicles off the road as possible. For cars to get more people off the road than busses, more people have to be willing to use cars than are willing to use busses. How many more?

Define DB to be the number of people held by a bus, and DC to be the number of people held by a car. The ratio of people willing to carpool vs. number willing to bus must be at least (1-1/DB)/(1-1/DC)

If this criterion is met, more vehicles will be off the road by carpooling, even though each vehicle holds fewer people.

Voter Fraud vs. Voter Suppression

There's a lot of talk lately about voter ID laws. Like many things, this issue has devolved into two sides. One side says that these laws are necessary to prevent voter fraud. The other side says these laws disenfranchise legitimate voters.

I'm not going to get into each side accusing the other of having negative motives. That's just not productive.

Let's instead start with basic premises of democracy, and identify our common ground. In a perfect system, no legitimate voter would be prevented from voting, and no illegitimate voter would be allowed to vote. This applies regardless of political persuasion, race, religion, location, age, anything. The goal is to have a good election, even if it results in your side losing.

(If you don't agree with that, I suggest you go live in Iran, which you might find more to your liking.)

Now that we've identified the ideal, we have to acknowledge that we don't live in a perfect world. Some number of fraudulent votes will be cast, and and some number of legitimate voters will be turned away. We wish to minimize both these numbers. But what if we have to decide between them?

That's what these voter ID laws ask us to do. So instead of arguing about hypothetical models, let's talk about what can be quantified.

These laws will prevent some number of fraudulent votes (and before someone accuses me of taking sides, zero is a number). These laws will also suppress some number of legitimate votes. We can measure both those things and see what effect the laws have.

But before we do that, let's define a standard. Having a standard, then comparing evidence against our standard, gives us a path to admit we were wrong, and thus become right. Again, let's try to find common ground.

Say we wanted to avoid even a single case of voter suppression, but as a consequence we had to accept ten billion fraudulent votes. This is a bad trade, because the ten billion fraudulent votes now dictate the course of the election. The one vote we "saved" didn't matter at all.

Now the other side. Say we wanted to prevent one fraudulent vote, and to do so we had to suppress every other vote in the country. I think we'd all agree that was a bad trade. Sure, there are no fraudulent votes, but there's also no election.

These absurd extremes demonstrate that we all live somewhere in the middle, and that we can (at least in principle) put a number on this problem. The only real difference any of us have is exactly where between these extremes we draw our line. So ask yourself: how many fraudulent votes have to be prevented to be worth suppressing one legitimate vote? The one suppressed voter loses his voice entirely; the one or five or fifty fraudulent voters reduces the value of everyone's vote. Where's the balance point?

I'm pretty confident that if you have to suppress more than one vote to prevent one fraudulent vote, we're definitely in the wrong territory. One fraudulent vote does less damage to the election than one suppressed vote does.

I'm also pretty confident that if you allow more fraudulent votes than the margin of error in the election, you're ruining the integrity of the election for all involved. So that puts a hard upper bound on the number of fraudulent votes that should be allowed in the entire election.

In between those numbers, things get a little fuzzy. I could see easy arguments for any ratio between 1:1 and 1:10, and I'd like to hear arguments for numbers outside that. So we have our standard. Now we ask, how do the numbers work out? Because if more votes are being suppressed than frauds are being prevented, we've created more problems than we've solved.

Let the data gathering commence!

Once more unto the blog

I find myself wanting to say random things with no particular direction. The things I want to say can't typically be compressed into 140 characters. Thus, blog.

I've done this a few times before, but not with any serious regularity. We'll see how I do now.

Work's been interesting of late. I've been working on what's basically a grid-tied inverter for regenerative braking. I wrote a Python script to autogenerate my sine wave lookup table. The output and switching frequencies can be controlled by two constants in the script, and you can immediately recompile. It's quite excellent. The unit seems to work under all tested circumstances. Next, abuse.

I've also been training a new engineer. He's learning well how we do things, having had to drop two projects already for higher-priority ones. And now he's trying to modify and update my code from three years ago. (Poor sod.) See, I built this tester unit. It met spec, but I was never happy with the overall architecture. Now, three years later, someone wants several of them, somewhat modified, and they want them now. Since I never went back and redid it in what I now (in my since-acquired infallible wisdom) know to be the One True Way, and since we're in a rush, we have to carry over the bad old architecture to Yet Another Product.

Be sure your sins will find you out.