How we count, and use, our prisoners.

Census block 2054 lies in Murray, N.Y. The block is 645 acres of flat farmland that, in 1990, held only 63 people (and nearly as many cows). But the 2000 census mysteriously reported a sudden influx of 1,395 new residents, 97 percent of them female, even though nothing had been built in Murray during those 10 years. Did the numbers lie?

They did, as it turns out. There is a cluster of 1,400 women in this rural county, but the cluster is actually one town to the west, in Albion Correctional Facility. County-planning officials caught the mistake; the population was transferred to the town of Albion, and that was that. But counting those 1,400 women as Albion residents makes almost as little sense as counting them in Murray. They do not vote or pay taxes; they have no ties to the community. As one Albion corrections officer puts it, "I don't think of them as residents. I mean, when they leave, they're not going to stay here longer than it takes to get a bus ticket out."

Almost all of Albion's inmates are from elsewhere (mostly New York City), and most will return home. Most of them would never even have set foot in rural Orleans County if they were not forced to. So, why does the federal census treat them as upstaters?

The way a society counts its people depends on what it plans to do with the data. As David Plotz recently wrote in Slate about the census taken in the Book of Numbers, the ancient Israelites only paid attention to those citizens they considered potential soldiers: adult males. Perhaps they should have included women or excluded the elderly, but at least the Biblical census did what it set out to do. The United States census, on the other hand, is not a military tool but a democratic one. And by this standard, it fails.

Our legislators use census data to apportion electoral districts. In doing so, they are supposed to follow two basic legal principles. The first is "one person, one vote." For every vote to count equally, every district should contain the same number of people. The second principle urges legislators to protect "communities of interest" by drawing electoral lines around unified voting blocs. Our current census policy—counting inmates as residents of the town in which they are locked up—violates both of those principles.

In 1964, in Reynolds v. Sims, Chief Justice Earl Warren established his "one person, one vote" principle, writing, "legislators are elected by voters, not farms or cities or economic interests." But legislators are not elected by prisoners: In 48 states, including New York, prisoners cannot vote. Of Albion's 8,000 "residents," about 2,500 are behind bars, so when a free Albion resident goes to the polls, his or her vote counts for more than it should.

As for unified "communities of interest," the interests of many inmates are actually directly opposite to those of the rural folk on the other side of the fence. Abraham Gilliam, a business owner in Albion and himself a former inmate, says, "In prison it's a lot of blacks, a lot of poor people, a lot of liberals. But they're all represented at the state level by white Republicans. … I can tell you from talking to people inside that they feel used."

The courts have traditionally overlooked the misrepresentation of inmates because it didn't make much practical difference. We didn't have all that many inmates, and until the 1970s, the ones we had usually served their time near their homes. Last May, however, the 2^nd U.S. Circuit Court for Appeals questioned the long-standing practice of counting prisoners where they sleep. As that court noted, the New York state Constitution mandates that "no person shall be deemed to have gained or lost a residence" by being "confined in any public prison."

Today, after the prison boom of the '80s and '90s, inmate misrepresentation is no longer statistically negligible. Prisoners are still a minority, obviously, but a growing one. America has—by way of comparison—more prisoners than Muslims. The United States penal system has blossomed into the largest in human history with two million citizens behind bars—more prisoners per capita than any other country.

Most of New York's convicts come from downstate, but almost all of them are locked up in remote, majority-white communities. New York has built 43 prisons since 1976, all of them upstate. As the prison population grows, poor neighborhoods in the Bronx, Brooklyn, and Buffalo lose population while prison towns like Albion gain phantom constituents.