charlie's Substack

I Tried to Save the Electoral College. The Math Wouldn't Let Me.

charlie vargas — Fri, 29 May 2026 17:38:03 GMT

Your vote may not count. Not because it wasn’t cast, not because it wasn’t counted, but because the person you trusted to cast it for you has opinions of their own.

But a faithless elector is the rare case. It almost never happens, and ALL states can legally cancel or replace one. That version of the problem doesn’t really bother me. What bothers me is the quieter one, the system can screw you over while every single elector does their job perfectly and follows every rule. You vote. Your vote gets counted. Every official acts in good faith. And you still end up feeding a ballot into a system that was never built to let one vote out of 155 million be heard. The government works for the people. The people should be what decides who holds office, right?

This is not hypothetical. Four times, the candidate more Americans voted for has lost the presidency anyway. Two of those four happened since 2000.

The reason you’ll never hear for the Electoral College’s existence is slavery. The South wanted its enslaved population counted toward its political power without letting those people vote. A direct national popular vote would have counted only actual voters, which would have stripped the slave states of that advantage. The three-fifths compromise, and the Electoral College built alongside it, let the South inflate its weight in choosing the president. That is not a footnote. It is one of the load-bearing reasons the system took the shape it did.

The practical reason is obsolete, and the original political reason was shameful. Neither one is a reason to keep it.

The system has decided that some votes are worth more than others, and it decided it by geography. Live in California and your vote is essentially worthless next to a vote cast two thousand miles away. Move to Wyoming (I wouldn’t wish that on my worst enemy) and your voice suddenly counts for more, because your single vote is a bigger slice of a smaller pie.

Do the math. Wyoming has about 580,000 people and 3 electoral votes. Divide one by the other and each electoral vote there stands for roughly 193,000 people. California has about 39.5 million people and 54 electoral votes. Run the same division and each electoral vote stands for roughly 732,000 people. So a single vote in Wyoming carries about 3.8 times the weight of a single vote in California. Same country, same election, and the system has quietly decided that one of those people counts almost four times as much as the other.

Now think about how a presidential campaign actually spends its final months. The candidates circle the same small group of states, over and over. Roughly seven of them, depending on the year, decide the whole election. The other forty-some are scenery.

I couldn’t tell you the last time a presidential candidate tried to win my vote. Part of that is me. I turned 18 in New York City, voted in New Jersey by mail, and was not especially plugged into New Jersey politics. But part of it is that there was nothing to plug into. Neither New York nor New Jersey is a swing state. Neither one gets a visit, an ad, or a tailored promise. When you vote anywhere outside the seven states that matter, the system is telling you: it does not really matter whether you vote at all.

This should make you angry! The Electoral College is defended, constantly, as the thing that protects small states from being ignored. Wyoming gets ignored. It is a guaranteed lock for one party, so no campaign spends a dollar there. New Jersey gets ignored for the same reason in the other direction. The system is publicly defended for doing the exact opposite of what it actually does. A system that is defended out loud and then quietly works against the people it claims to protect is not a mistake, it is a choice in design.

I did not start out wanting to abolish the Electoral College. I started out trying to save it. I spent time on a fix that kept the College intact, allocating its electoral votes proportionally instead of winner-take-all. I wanted to base it all off of the least populated state. For example, if the least populated state had 100,000 voters, and your state has 150,000 voters, you’d have 1.5 votes towards the election. Now say a state voted 60/40 in favor of a blue candidate. That’s 90,000 votes for the blue candidate and 60,000 for the red. Dividing by 100,000, you’d get 0.9 votes for blue and 0.6 votes for red. At the end of collecting the votes, the national total for blue and red is just the national vote for each candidate divided by 100,000. That’s just the national vote. The divisor does nothing at all. Every honest attempt to keep the electoral college while also reallocating a proper amount of votes based on population lands back with the national popular vote.

The real answer was simpler than the thing I was building, and it annoyed me that I hadn’t dug deep enough to find it the first time.

Abolish the Electoral College. Count every vote in the country. Whoever gets the most wins.

Having a one person, one vote policy means you do not have to hand your ballot to an intermediary you have never met and trust them to carry it for you. It means your vote matters on its own, regardless of whether a family member, or the person next to you, or a stranger two thousand miles away, disagrees with you. Voting is democracy’s life. It is the mechanism that turns a population into a government.

The first objection to a national popular vote is the one I raised against my own idea before anyone else could: doesn’t this just let big states dominate? California has 39 million people. Doesn’t counting every vote hand them the country?

It doesn’t. A “big state” is not a thing that votes. 39 million Californians vote, and they do not vote as one bloc. In 2020, California split roughly 63/34. About 6 million Californians voted for Trump. Under the system we have right now, every one of those 6 million produces exactly zero electoral votes. The state went blue, the bucket tipped entirely blue, and 6 million people were erased. Meanwhile 580,000 people in Wyoming deliver three electoral votes, guaranteed. The current system is not protecting small voices. It is deleting six million of them in a single state.

A national popular vote does not make California dominate. It makes Californians count, all of them, including the 6 million the current system silences. “California dominates” would require all 39 million to vote the same way. They never do. Nowhere does.

The unit stops being the state and becomes the person. A voter in Wyoming and a voter in California each get one vote, the same weight. California has more total influence only because more people live there, and more people having more influence is not domination. It is the definition of the thing we keep saying we want.

The system that causes the most neglect is the one we already have. Right now about seven swing states decide the presidency and the other forty-some are scenery. Under a popular vote, a vote in Wyoming is worth a vote in Pennsylvania, which gives a candidate more reason to show up in Wyoming, not less. The fear that small places get ignored describes the present, not the proposal.

Two other defenses deserve a quick answer. The Electoral College is said to force candidates to build broad geographic coalitions. It does the opposite. It forces them to chase seven states, a narrower map than a popular vote would produce. It is also said to contain recounts, keeping a disputed result inside one state instead of triggering a national one. That one is partly true. It is a real cost, and I will come back to it honestly.

I have to be honest about my own proposal, because doing anything else is the move I just spent an essay attacking.

I will not tell you the national popular vote is flawless. But “isn’t flawless” is an empty thing to say. It is true of every voting system ever designed, and it carries no information. Flawless was never on the menu. So the real question is not flawed versus flawless. It is which specific flaws you would rather live with.

The first is the serious one. A national plurality vote has no majority requirement. In a race with three or more real candidates, someone can win the presidency with well under half the country behind them. In 1992, Bill Clinton won with 43% of the popular vote, because Ross Perot pulled 19% of it. Under a pure national popular vote, you could seat a president on 38%. That is a genuine legitimacy problem. The Electoral College does not cleanly solve it either, but I am not grading my proposal on a curve. It is a flaw. The fix, a national runoff or ranked-choice voting, is its own redesign with its own tradeoffs.

The second is practical. A genuinely close national race means a national recount, every precinct in the country rather than one state. Slower, messier, more open to litigation. Florida in 2000 was a disaster, but it was a contained one. A national popular vote does not contain it.

These flaws are real, and I am not going to wave them away. But I’m not conceding either. Every system on offer is flawed. The only honest question left is which set you would choose.

So here is the choice. One system can seat a president the public was less inclined to back in the first place, and hold him up with structure instead of support. The other gives you a closer election, decided by where the general population actually landed. I will take the closer election every time. A president should be held up by the people who voted for him, not by the machinery that happened to count them.

And wanting this should be easy, because it costs almost nothing. We are not building a new machine. We all already vote for the candidate we prefer. The popular vote does not add a single step to that. It removes one. It takes out the intermediary standing between your ballot and the outcome, and lets democracy do the plain thing it was always supposed to do.

We aren’t far from this reality at all. It does not need a constitutional amendment that will never pass. The National Popular Vote Interstate Compact is already law in 19 jurisdictions holding 222 electoral votes, and it takes effect at 270. The fix is 48 electoral votes away from being real.

That gives me hope. It should also come with urgency, because 48 votes do not close themselves. A government that actually works for the people is close enough to reach, but close is not the same as done. It still takes effort. It still takes us deciding, out loud, that we want it.

Your vote may not be heard. But it can be.

References:

The Paper Schedule

charlie vargas — Sun, 17 May 2026 21:31:26 GMT

There’s a paper on the wall of every train station in Tokyo. It tells you when the next train will come. Not “in 4 minutes” or “delayed” or “cancelled” on a digital display that updates whenever it feels like it. An actual, printed schedule, with actual times, down to the minute.

Best part. The trains come when the paper says they will.

I moved to New York for school about 2 years ago. I take the subway every day. The first time I came home from Tokyo and stood on a platform waiting for the Q, I started thinking about something I’d had wrong for a long time.

For most of my life I assumed the difference between New York’s subway and Tokyo’s was money. That if the MTA was just willing to spend more, the trains would work the same way. I really hate being wrong about this, but I was.

The MTA has a $21.3 billion operating budget^[1]. Per rider, it is one of the most-funded transit agencies on the planet. New York pays for its subways many times over, whether in fares, dedicated taxes, federal grants, or the congestion pricing fight that took most of a decade to settle. Only 39%^[1] of that budget comes from fares and tolls. The rest comes from taxes and government contributions, meaning the entire region has been quietly subsidizing the system for years. The money is there. The problem is what happens after we hand it over.

I’ve spent some time looking at the MTA up close. I built a fare analysis^[2] last year showing that New York’s flat $2.90 fare (now $3) could be replaced with a distance-based model that would generate around $913 million in additional annual revenue while lowering costs for about 15% of trips. The deeper I went into the data, the clearer it became that the system’s problems aren’t hiding. The numbers are fine. They have always been fine.

What’s actually broken doesn’t show up on a balance sheet. It shows up in who answers for the trains when they don’t run. It shows up in how much it costs us to build a single new mile of track.

Who actually answers when the trains don’t run?

The answer? It depends, and that’s the problem.

A 17-member board governs the MTA, and every single one of them is technically appointed by the Governor of New York. The state Senate confirms them. Of those 17 members, 6 are nominated directly by the Governor, including the Chair, who runs the entire agency. Four more are nominated by the Governor on the recommendation of the Mayor of New York City. Three are recommended by the county executives of Nassau, Suffolk, and Westchester. The remaining four are recommended by executives from Dutchess, Orange, Putnam, and Rockland counties, and together they cast a single collective vote^[9].

NYC Transit handles roughly 2.4 billion rides a year. Metro-North and the Long Island Rail Road combined handle about 175 million. Add those up. 2.575 billion total rides. NYC Transit’s share is 93%. So if you ride the MTA, there’s a 93% chance you’re riding inside New York City. And of the 17 board seats, only 4 are filled by people the Mayor of New York City had any say in. That’s 24% of the board. 93% of the riders, 24% of the board.

The actual people getting on the subway in the morning have no one who directly answers to them.

This is what governance looks like in practice.

In late 2018, Governor Andrew Cuomo personally took control of an MTA decision^[10]. The agency had spent years preparing for a 15-month shutdown of the L train so it could repair the Canarsie Tunnel, which had been flooded with salt water during Hurricane Sandy in 2012. The plan had gone through extensive public hearings. Businesses had relocated. Commuters had built workarounds. The MTA’s paid consultants (WSP and Jacobs Engineering, on contracts worth hundreds of millions) had explicitly endorsed the full shutdown as the “preferred approach”^[11].

Then, three months before the shutdown was supposed to begin, Cuomo called the engineering deans of Cornell and Columbia and asked them to take a look at the plan. They worked unpaid over winter break. Three weeks later, they came back with a different proposal: wrap the corroding bench wall in fiberglass instead of rebuilding it, and run new cables along the tunnel walls instead of encasing them in concrete. On January 3, 2019, Cuomo held a press conference announcing the shutdown was canceled. The MTA chairman stood next to him and endorsed the new plan. No MTA officials publicly raised any objections. The shutdown never happened. The new plan worked.

The detail that haunts me about this story is what was on the MTA’s payroll while it was happening. 173 civil engineers. 107 project managers. 73 architects^[11]. None of them had proposed the academic plan. The hundreds of millions in consultants couldn’t see what two unpaid deans saw in three weeks. The Riders Alliance called it “a last-minute Hail Mary idea cooked up over Christmas,” and they meant it as an insult, but it stuck precisely because it was true.

You can read the L train story two ways at once, and both are correct. One reading is that the MTA is so dysfunctional that years of expert planning could be improved by free academic labor over a holiday break. The other reading is that one elected politician unilaterally overruled years of public process with a plan invented in three weeks by people accountable to no one. Whether or not it worked this time, the next time it might not. There is no functional system between the people running the trains and the people riding them.

This is the structural answer to who is in charge of the subway. The Governor is in charge. The Governor has always been in charge. The Governor is elected by people who, for the most part, don’t ride. The people who do ride get to elect a Mayor who recommends four members to a board controlled by the Governor. The accountability gap is not an accident. It was built into the agency at its founding.

Now the cost.

The accountability gap doesn’t stay in the boardroom. It shows up in what the agency builds, and what it pays to build it.

A single mile of new subway track in New York costs about $2.5 billion^[3]. Phase 2 of the Second Avenue Subway, still waiting for the rest of its funding, is projected at around $4.3 billion per mile^[4]. The global average across 900 projects in 60 countries is closer to $350 million per mile. Similar projects in Italy, Sweden, Spain, Turkey, France, and Germany typically come in between $250 and $450 million. Seoul builds at roughly one-twentieth^[5] of our per-mile cost.

$2.5 billion divided by $350 million is 7.1x. That’s Phase 1 of the Second Avenue Subway versus the global average. Phase 2 at $4.3 billion per mile is 12.3x. And Seoul builds at one-twentieth our cost, meaning Seoul builds a subway for roughly 5% of what we pay. We are not 20 times worse at moving dirt or pouring concrete. We have just become 20 times worse at running the operation that does it.

There are limitations to this comparison, and I want to name them before someone else does. Tunneling in Manhattan is genuinely harder than in some cities. Older utility infrastructure complicates everything. American labor and material costs run higher in general. But the Transit Costs Project, which has spent the better part of a decade on exactly this question, controls for most of those. The 20x gap doesn’t disappear when you do.

The Transit Costs Project^[6] at NYU’s Marron Institute points to 3 clear pain points. Physical structures. Labor. Procurement.

I’ll start with the stations.

The Second Avenue Subway’s mined stations are roughly twice as long as the trains that run through them. The 96th Street Q station is closer to three times longer. By comparison, Italian and French subway stations are 5–10% longer than their trains. Stockholm’s Odenplan is 17%. Stand on the 72nd Street platform at some point. The station feels luxurious and grand. High ceilings, mosaics, a vast tunnel-like feel, and the steepest escalators known to man, all of which beat most of the rest of the system. But the train pulls in and there’s more empty platform than there is platform for trains to operate on. By no means am I saying the station isn’t beautiful. It’s just way larger than it needs to be. Phase 1 of the Second Avenue Subway used two different escalator vendors across three stations^[7], with each station essentially custom-built, even when standardization would have been cheaper and faster.

Next, labor. In Turkey, Italy, and Sweden, labor costs are roughly 20–30% of total construction^[6]. In the Northeast US, labor is about 40–60%. That difference doesn’t come from American workers being paid more. At most a quarter of the gap is wages. The rest is headcount.

During Phase 1 of the Second Avenue Subway, the tunnel boring machine was operated by 46 workers per 8-hour shift, with another 29 supervisors overseeing them across the day. The Transit Costs Project’s interviewees said the same machine could have been run with 30 workers per shift, and that the number of supervisors exceeded any international norm.

Across the day, 46 plus 29 is 75. 75 divided by 30 is 2.5. We are sending 2.5 times the labor required to do the work.

Finally, procurement. Different agencies that share a worksite (utilities, transit operators, the city, the state) each insist on sending their own supervisor into the tunnel to watch the contractor. Each set of supervisors gets paid out of the project budget. The Transit Costs Project describes the broader pattern as “a pervasive culture of secrecy and adversarialism between agencies and contractors,” combined with a lack of capacity at the MTA to manage the consultants it hires. The result is a system in which paying for oversight has become a major cost driver, and the oversight itself doesn’t seem to make the work any better.

It is important to me that I tell you what this argument is and what it isn’t. It isn’t an argument that MTA workers are overpaid. The research is explicit that pay differences account for at most a quarter of the gap^[7]. It isn’t an argument that unions are the problem. White-collar overstaffing is just as severe. And it’s not an argument that New York has caught some uniquely American disease that can’t be cured. Alon Levy, one of the Transit Costs Project’s lead researchers, has given it to us straight^[8]. “The causes of high American (especially New York) costs are institutional, and fixable, without a revolutionary upending of any legal or social system.”

The important piece is that we aren’t fixing them. In late 2024, the MTA approved roughly $250 million in new consulting contracts for Phase 2 of the Second Avenue Subway. The same project the Transit Costs Project used as its case study for procurement dysfunction. The pattern the researchers identified as a primary cost driver is being repeated, on the same line, in plain sight.

None of what I’ve laid out is mysterious. The Transit Costs Project has spent years documenting it. Reporters have spent decades on it. The MTA itself publishes most of the relevant numbers in its annual reports. The diagnosis isn’t hidden. The governance structure that lets a governor override years of expert planning is written into the public authorities law. The construction costs are public record. The procurement dysfunction is named in a published academic report whose authors will happily give you a tour.

What we don’t have is a political incentive to fix any of it. The governor isn’t punished by NYC riders for bad decisions, because most of his voters don’t ride. The MTA isn’t punished by the public, because the public can’t tell whether their delay is the agency’s fault or someone else’s. And the consultants keep getting hired because the people doing the hiring don’t have the internal capacity to do the work themselves. Which is the same problem the consultants were hired to solve.

I think about the paper schedule in Tokyo a lot. Not because the United States is going to become Japan, but because the paper schedule is evidence that a transit system can be built with the goal of working, and succeed at it. We do not need to copy Tokyo. We need to want this for ourselves.

charlie's Substack

I Tried to Save the Electoral College. The Math Wouldn't Let Me.

The Paper Schedule

Who actually answers when the trains don’t run?

Now the cost.

Footnotes