Cyrus Reed: Iris, hey — I want to start with a confession, actually. I came into this thinking gerrymandering was a boring civics problem. Lines on a map. Whatever.
Cyrus Reed: Now I think it's — wait, no, let me actually say the specific thing, because the specific thing is what broke my brain. In North Carolina, the mapmakers who drew the 2016 congressional map ran simulations — fifty-two separate election scenarios, different turnout, different national environments — and their map produced the same result every time. Republicans ten or eleven seats out of fourteen. In a state that was genuinely competitive.
Iris Holm: Fifty-two simulations is not a coincidence. That's a confidence level.
Cyrus Reed: And courts could see it! That's what I — the courts looked at this map and acknowledged, yes, this is engineered. And still couldn't act. How does that happen?
Iris Holm: Because proof and remedy are different legal questions. But before we get there — the underlying move here is older than anyone remembers. Elbridge Gerry, Massachusetts, 1812. He approved a district shaped like a salamander, newspapers literally drew it with a head and wings, and that word — gerrymandering — has been with us ever since.
Cyrus Reed: So the tactic is two hundred years old and we still can't stop it — but now we can prove it with fifty-two simulations, which somehow makes the powerlessness feel worse?
Iris Holm: That's the exact tension. Precision exposes the problem. Precision does not, by itself, provide the fix.
Cyrus Reed: Yeah — and that's what we're here to actually figure out.
Iris Holm: And the precision is the point — so let me actually show you the machine. Forget jargon. Imagine you're dividing a room of a hundred people into five voting groups of twenty. You don't change anyone's preference. You just choose who stands next to whom. If you're clever about it, you can arrange the room so your opponents win one group by a landslide and lose the other four narrowly. They got more total cheers — and one seat. You got four.
Cyrus Reed: Wait — same people, same preferences, totally different outcome just from the arrangement.
Iris Holm: That's the whole game. And it has two moves. Packing: cram your opponents into as few groups as possible. They win those groups by enormous margins. But those extra votes? Worthless — one group, one seat, surplus is gone. Cracking: everyone else gets spread thin across the remaining groups — never a majority anywhere.
Cyrus Reed: So packing and cracking aren't two separate tactics — they're one thing running at the same time.
Iris Holm: One machine. And the fuel it runs on is wasted votes — votes cast for a losing candidate in a cracked district, or votes piled on top of an already-won margin in a packed one. Neither kind elects anyone. Asymmetric accumulation of those wasted votes between the two parties is exactly what bends seat share away from the raw statewide numbers.
Cyrus Reed: Okay but how does that — wait, is this why total vote share and actual seats won look so different? Like you can win the state and still lose the chamber?
Iris Holm: Exactly because of winner-take-all. Each district sends one representative, chosen by simple plurality. That means spatial distribution of voters — where they live, which side of a line they fall on — determines seat totals more than statewide vote share does. Geometry becomes the variable that matters.
Cyrus Reed: That's — huh. Geography as a lever.
Iris Holm: North Carolina's congressional map is the clearest case. Democratic voters packed into two districts — they win those two huge. Republicans spread into the remaining twelve efficiently enough to guarantee ten or eleven seats in a state that was genuinely competitive statewide. The Princeton Gerrymandering Project flagged that signature: one party winning a small number of districts by overwhelming margins while the other takes the bulk narrowly.
Cyrus Reed: So the landslide wins are actually the evidence of the rigging — not the sign that party was popular.
Iris Holm: That's the tell. A genuinely competitive map produces close races across the board. What you see instead — one side stacking up eighty-percent blowouts in two districts — that's the waste being warehoused. Brennan Center frames it the same way: the lopsided victory is the footprint of the pack.
Cyrus Reed: And none of this requires changing a single vote. The preferences are identical — you're just moving the fences.
Iris Holm: Moving the fences is the old version. The new version — you don't even need a weird-looking fence.
Cyrus Reed: Wait — what do you mean it doesn't look weird anymore?
Iris Holm: Compactness, contiguity — those are the traditional legal tests. Is the district connected? Is it roughly square? A modern gerrymandered map can pass both with flying colors and still sort voters with surgical precision. Nothing salamander-shaped. Nothing that triggers the old red flags.
Cyrus Reed: So the disguise got better. Because — wait, okay, if you have granular precinct-level voter data and a good optimization algorithm, you don't need to draw a monster shape, you just... nudge a hundred small lines in ways that look totally normal.
Iris Holm: That's the North Carolina 2016 story. Fifty-two scenarios. Republican ten or eleven, every time. The map looked fine on paper.
Cyrus Reed: So how do you even — I mean, if there's no salamander, how do mathematicians catch it now? Because somebody did catch it.
Iris Holm: Two instruments. First: the efficiency gap. You count wasted votes for both parties across every district — losing votes in cracked districts, surplus votes in packed ones. Take the difference. A large gap means one party's votes are being systematically thrown away at a higher rate. That metric was central to the federal case against Wisconsin's 2011 Assembly map — a three-judge court ruled in November 2016 it was an unconstitutional partisan gerrymander.
Cyrus Reed: November 2016 — that's the same month as the presidential election.
Iris Holm: Right. And that case, Gill v. Whitford, went up to the Supreme Court — which is where the efficiency gap hit its ceiling. The Court didn't adopt it as a universal legal standard. Mathematical proof, no universal remedy.
Cyrus Reed: Okay but the second instrument — you said two.
Iris Holm: Ensemble analysis. You generate tens of thousands of alternative maps using only neutral criteria — compactness, equal population, that kind of thing — and then you compare the enacted map's partisan outcomes against that whole distribution. If the real map sits at the extreme tail, you've got your proof. And then at the precinct level, DeFord, Eubank, and Rodden built something called Partisan Dislocation — it measures how different a voter's assigned district is, in partisan terms, from her actual geographic neighbors. High score means the district was artificially constructed around her.
Cyrus Reed: Okay that's — wait, that's not just proving a map is rigged, that's showing which specific communities got split or stuffed. Like, imagine an epidemiologist in Ohio trying to map opioid addiction rates by congressional district and two adjacent neighborhoods — literally the same street, same demographics — show completely different numbers. Not because the addiction rates differ. Because a gerrymandered boundary cut the community in half and put each half in a different district with totally different surrounding populations.
Iris Holm: The map shapes what we measure about reality. Not just who wins seats. The Princeton Gerrymandering Project flags the electoral signature — but that Ohio epidemiologist is staring at the second-order damage.
Cyrus Reed: And frankly that asymmetry between what we can prove and what we can fix — that gets even stranger when you look at which party's actually been holding the pen, and what independent commissions do or don't solve. That's the part that keeps nagging at me.
Iris Holm: The pen question is the right place to pull. The Congressional Research Service frames packing and cracking as structurally symmetric — any party holding redistricting power can run the same playbook. And historically, that's true. Democrats drew maps that favored them for decades in the South.
Cyrus Reed: But the modern documented record is — wait, it's almost entirely one side. Wisconsin's 2011 Assembly map, North Carolina's congressional map. Those are Republican-controlled legislatures.
Iris Holm: Every named example the Brennan Center and MIT Election Lab keep pointing to — same story.
Cyrus Reed: Okay but is that because one party is more aggressive, or is it just — after 2010, Republicans won a huge number of state legislatures right before the census-year redistricting cycle? They held the pen because they won the timing lottery.
Iris Holm: That's the honest counter. Redistricting follows the decennial census — whoever controls the legislature that year controls the map for the next decade. 2010 was a wave election for Republicans at the state level. So yes, they held the pen more.
Cyrus Reed: So the asymmetry might just be — timing? Not ideology?
Iris Holm: Partially. But there's a second factor neither of us can dismiss. Democratic voters cluster — cities, dense precincts. That geographic concentration means cracking is actually easier to execute against them. You don't have to work as hard to disperse a party whose voters are already packed into urban cores.
Cyrus Reed: Wait — the geography does the first half of the job for you.
Iris Holm: The research doesn't resolve whether that's the dominant cause — scholars genuinely find mixed results on magnitude. Some find strong seat-distortion effects, others find limited or contested evidence. But the structural mechanism is real regardless.
Cyrus Reed: And the thing that makes me — okay, so the magnitude question almost doesn't matter when you look at the timeline. One census-year election locks the map in for a full decade. That's not one bad cycle, that's ten years of compounding.
Iris Holm: Which is why the public health finding hits differently than people expect. A 2024 study in the Journal of Public Health Management and Practice — peer-reviewed — found that gerrymandered district boundaries distort uninsurance rate statistics at the congressional district level relative to nonpartisan simulated maps. The boundaries designed to waste votes are the same boundaries Congress uses to allocate health funding.
Cyrus Reed: Hold on — so a map drawn to guarantee ten Republican seats in North Carolina is also generating a false picture of how many people in those districts lack health insurance? The data infrastructure breaks?
Iris Holm: Ohio's Congressional District 11 makes this concrete — racial and partisan gerrymandering overlapping in the same boundary. Packing minority communities serves both goals simultaneously. And the federal funding formulas downstream? They read those district lines as if they reflect real communities.
Cyrus Reed: So the question isn't just who wins the seat — it's what version of reality the seat produces. And if independent commissions just move the pressure to a different table without changing the incentive to distort that reality, then — I mean, what are we actually solving?
Iris Holm: That's the honest answer: nothing, yet. Independent commissions just relocate the pressure. Someone still has to decide what 'neutral criteria' the algorithm optimizes for — and that decision is political.
Cyrus Reed: Which means — wait, okay, so the North Carolina map encoded an algorithm into the cartography, and the proposed fix is... a different algorithm, chosen by a different room of people, with the same incentives just wearing a neutral-sounding name.
Iris Holm: The incentives migrate. They don't disappear. And next redistricting cycle — same computational tools, both parties. The problem stops being who holds the pen.
Cyrus Reed: It becomes — what do you do when everyone can build a map as precise as the fifty-two-scenario one? And courts still can't agree on whether the efficiency gap is even the right measuring stick?
Iris Holm: That's the thing I keep not being able to resolve cleanly. Gill v. Whitford proved the Wisconsin Assembly map was rigged well enough to overturn it federally. Didn't produce a universal standard. So mathematicians can prove it, courts can see it — and the legal infrastructure isn't there to catch the next one.
Cyrus Reed: The geometry was always the mechanism. That part's two hundred years old. It's just — the precision is new, and the precision is what we don't have an answer for.
Iris Holm: That's actually the small thing I buy. Not a grand frame — just: precision outran the referee. That's where we are.
Cyrus Reed: Yeah. Uncomfortable place to land, but it's the honest one. Thanks for walking through all of it.
