This suit by an Indian tribe (and a member of the tribe, but we can ignore that detail) claims that a county board of supervisors redistricted the county in a manner that violated both the equal protection clause and the Voting Rights Act, 42 U.S.C. § 1973. The district court granted summary judgment for the defendants (collectively, the County), Frank v. Forest County, 194 F. Supp. 2d 867 (E.D.Wis.2002), and the tribe appeals.

The oddness of the tribe's claims is made dramatically clear by the facts of the case. Forest County is a large (1014 square miles) but sparsely populated (barely 10,000 people) county in the extreme northeastern corner of Wisconsin, fronting on Lake Superior. The population is unevenly distributed across the county; 57 percent of the 1620 census blocks into which the county is divided have no human inhabitants at all. The county is governed by a board of supervisors each of whose 21 members is elected from a single-member district in nonpartisan elections held every two years. The board redistricted the county after learning the results of the 2000 census, which revealed a marked increase in the Indian population since the 1990 census, from 8 percent to almost 12 percent (45 percent of the Indians live on reservations). The rest of the county's population is white, except for a very small number of blacks (only 118 on census day) almost all of whom (106) are residents of the Blackwell Job Corps Civilian Conservation Center, which offers a comprehensive educational and job-training program for "at risk" youth ages 16 through 24. The residents of the center, a majority of whom are black, are transients; the average length of their stay is only 188 days. Only 57 percent are Wisconsinites, and very few either come from Forest County or plan to remain there when their stint at the center is up.

If the 21 districts that elect supervisors each had 477 residents, the districts would be of equal population. The district map adopted by the board of supervisors and challenged by the tribe does not hit this nail on the head. One district, the largest, has 514 residents, and another, the smallest, has 428 residents, the others being of essentially uniform size. The difference between the largest and the smallest (86) is 18 percent of 477. The County admits, prematurely as we are about to see, that this deviation from perfect equality, because it exceeds 10 percent, constitutes a prima facie denial of equal protection. But it argues (and the district court agreed) that the deviation is defensible because to redraw the district lines in a way that would reduce the deviation to 10 percent would produce districts that were not compact and that crossed many local-government boundaries, so that school districts, fire districts, and so forth would straddle board of supervisor districts. The districts created by the board's plan are compact; there is no evidence that they are gerrymandered, whether along political or racial lines. The tribe points out, however, that the board's argument for why the 18 percent deviation is tolerable in the circumstances appears nowhere in the minutes of the meeting at which the board adopted the new district map over the tribe's objection.

The 10 percent norm on which the tribe's equal protection claim pivots represents the latest in a series of steps toward the ever greater automation of the redistricting process. Because the U.S. population is so mobile, there are large population shifts across thousands of federal, state, and local government districts between the decennial censuses. Redistricting is an intensely political process and there is no theoretical guidance to how to balance the various considerations that political science might deem relevant to conforming districted governments to the principles of democracy (themselves contested). So the judicial tendency has been to insist on a very close approach to mathematical equality, and one of the devices used is the 10 percent norm. "Our decisions have established, as a general matter," the Supreme Court has said, "that an apportionment plan with a maximum population deviation under 10% falls within this category of minor deviations. A plan with larger disparities in population, however, creates a prima facie case of discrimination and therefore must be justified by the State." Brown v. Thomson, 462 U.S. 835, 842-43, 103 S. Ct. 2690, 77 L. Ed. 2d 214 (1983) (citations omitted); see also Regensburger v. City of Bowling Green, 278 F.3d 588, 595 (6th Cir.2002).

Rules are attractive devices for economizing on litigation costs and minimizing judicial discretion; and safe harbors are particularly welcome to the bar. But a rule applied to circumstances remote from those contemplated when it was adopted can produce perverse results. The 10 percent rule, viewed not as a safe harbor (which it is in part, and unexceptionably) but as a rule of prima facie liability (which it also is — both aspects are clear from the passage we quoted from Brown v. Thomson), was devised for elections in large electoral units. See, e.g., Voinovich v. Quilter, 507 U.S. 146, 149, 161-62, 113 S. Ct. 1149, 122 L. Ed. 2d 500 (1993); Brown v. Thomson, supra, 462 U.S. at 838-39, 103 S. Ct. 2690; Connor v. Finch, 431 U.S. 407, 416-17, 97 S. Ct. 1828, 52 L. Ed. 2d 465 (1977); Daly v. Hunt, 93 F.3d 1212, 1215, 1221 (4th Cir.1996); Garza v. County of Los Angeles, 918 F.2d 763, 773 n. 4 (9th Cir.1990). Even the city council districts in the modest-sized city of Bowling Green, Ohio, were almost 15 times more populous than the districts in our case. Regensburger v. City of Bowling Green, supra, 278 F.3d at 592 n. 1. The smaller and more scattered the population of the area to be redistricted and the more numerous the districts, making it harder to create districts of equal population without creating weird shapes that straddle the boundaries of the smaller government units, as recognized early on by the Supreme Court in Abate v. Mundt, 403 U.S. 182, 185, 91 S. Ct. 1904, 29 L. Ed. 2d 399 (1971), the more arbitrary the rule of 10 percent prima facie liability becomes, until finally it becomes absurd. It is true that in Chapman v. Meier, 420 U.S. 1, 24, 95 S. Ct. 751, 42 L. Ed. 2d 766 (1975), the Court said that "sparse population is not a legitimate basis for a departure from the goal of equality." But the districts in question were roughly 25 times as populous as the districts in our case and anyway it would be improper to give sparsely populated areas greater representation than densely populated ones. See id. at 25 n. 16, 95 S. Ct. 751. At all events, even if intended to apply to the tiniest electoral districts, the 10 percent rule is merely one of prima facie liability, and is therefore rebuttable, Voinovich v. Quilter, supra, 507 U.S. at 161-62, 113 S. Ct. 1149 — and more easily so the smaller the population of the area to be redistricted, the more unevenly the population is distributed across the area, and the more numerous the districts. Let us elaborate on these points, beginning with population size.

No one supposes that the census is totally accurate, see Wisconsin v. City of New York, 517 U.S. 1, 6-7, 116 S. Ct. 1091, 134 L. Ed. 2d 167 (1996); and to bring the 18 percent deviation of which the tribe complains down to 10.9 percent would require subtracting only 16 people from the census count in the largest district and adding only 20 to the count in the smallest, which are changes that may well be within the census-takers' margin of error. Over a larger population, mistakes of overcounting and mistakes of undercounting will tend to cancel out. For example, the 1990 census is believed to have overcounted the national population by about 1.6 percent, but the small (relative) size of the error was the consequence of the fact that the 15-million gross overcount and 10-million gross undercount largely canceled each other out. See U.S. Census Monitoring Board, "Final Report to Congress" 16-17 and n. 11 (Sept 1, 2001), available at http://govinfo.library.unt.edu/cmb/cmbp/reports/final_report/FinalReport.pdf. But this cannot be assumed in the case of a minute local population. Moreover, more than three years have elapsed since the 2000 census was conducted, and during that time the movement of a few families across district lines in Forest County may have decisively altered the deviation on which the tribe relies to show a denial of equal protection. The significance of population shifts as a function of the time from the census-taking to the court challenge was recognized in Abrams v. Johnson, 521 U.S. 74, 100-01, 117 S. Ct. 1925, 138 L. Ed. 2d 285 (1997); here we add that the significance is also greater the smaller the population-and also the smaller the redistricted area is geographically, since people are more likely to move short than long distances.

Number of districts is important too, as we recognized in Sutton v. Dunne, 681 F.2d 484, 487 (7th Cir.1982). The more districts there are, the less meaningful a comparison between just two of the districts is. If there are four districts in the electoral unit, and the largest and the smallest have very different populations, the unit itself is seriously malapportioned. If there are a hundred, the malapportionment created by two outliers is apt to be at once trivial and unavoidable. Twenty-one, the number of districts in our case, is a large number from this perspective.

All this said, the board didn't provide much in the way of an explanation for the particular districting choices it made out of the almost infinite array theoretically available. (There are an unbelievably vast number of district configurations that would result in districts in Forest County of equal population.) But as with most other cases in which governmental action is challenged as a violation of the federal Constitution, it should be enough that good reasons can be given for the action, whether or not they were articulated in advance of the action. E.g., FCC v. Beach Communications, Inc., 508 U.S. 307, 315, 113 S. Ct. 2096, 124 L. Ed. 2d 211 (1993); McDonald v. Board of Election Commissioners, 394 U.S. 802, 809, 89 S. Ct. 1404, 22 L. Ed. 2d 739 (1969); Estate of Kunze v. Commissioner, 233 F.3d 948, 954 (7th Cir.2000). This precept is especially appropriate when one is dealing with so tiny a polity as Forest County, Wisconsin. We intend no disrespect in saying that the exceedingly modest functions that the State of Wisconsin has assigned to the board of supervisors of this remote rural county preclude any reasonable expectation of a deliberative process that we might expect of the Congress of the United States on one of its better days. Indeed we are mystified as to how this suit, with its impressive array of expert witnesses, was financed by either side; we are reluctant to impose upon the board a requirement of process that will burden the county government with expenses disproportionate to either its resources or its responsibilities. The supervisors are elected in nonpartisan elections, and their principal responsibility appears to be limited to keeping the county roads in good shape.

The tribe's voting-rights claim, to which we now move, is in great tension with its equal protection claim. The Voting Rights Act, so far as bears on this case, entitles a minority to seek district configurations that will sufficiently concentrate the minority population to enable it to elect some officials of its choice. 42 U.S.C. § 1973(b); Thornburg v. Gingles, 478 U.S. 30, 47, 106 S. Ct. 2752, 92 L. Ed. 2d 25 (1986); Barnett v. City of Chicago, 141 F.3d 699, 702 (7th Cir.1998); Milwaukee Branch of the N.A.A.C.P. v. Thompson, 116 F.3d 1194, 1196 (7th Cir.1997). The County's map creates one district in which Indians have 78 percent of the total population and 75 percent of the voting-age population, and another in which Indians have 53 percent of the total population but only 49 percent of the voting-age population. The final plan presented by the tribe in the district court would leave these districts intact but would alter the boundaries of another district to bring the Job Corps center that we mentioned earlier into the same district with a number of Indians, creating a district that would have a total population that was 49 percent Indian and 20 percent black and a voting-page population that was 39 percent Indian and 24 percent black.

What is remarkable about this plan is that it does not alter the largest and smallest districts in the County's plan and thus does not remedy the alleged denial of equal protection. As an afterthought on appeal the tribe argues that the deviation in population between those districts can be rectified consistently with creating the Indian-black district that is obviously the particular object of the lawsuit. But there is no map to indicate district boundaries and so we cannot evaluate the new plan. Barnett v. City of Chicago, supra, 141 F.3d at 702.

As it happens, both of the "deviant" districts are predominantly white; the larger is only 1.5 percent Indian and the smaller only 2.5 percent Indian, and the larger has a single black resident and the smaller none. The voters harmed by the deviation are the voters in the large(st) district, each of whom has less voting power than the voters in the other districts (and particularly the voters in the smallest district), but almost all of them are white — and there are no white plaintiffs. What is more, although the voting power of Indian residents is diluted in the large district, the dilution is more than offset in the small one because it has a larger Indian population, although the numbers are minute (8 Indians in the large district, 11 in the small). There thus does not seem to be any Indian interest in eliminating the alleged violation of equal protection, which is probably why the tribe did not bother to submit a plan to the district court that would eliminate the violation without compromising the tribe's goal of obtaining a combined Indian and black district.

In arguing for moving boundary lines to create a district that will give Indians and blacks together a majority of the population, the tribe points to studies of the voting patterns of blacks and Indians that indicate that the two groups have similar electoral preferences, so that by being concentrated in the same district each can hope for better representation of its interests than if Indians and blacks were scattered across districts dominated by whites. There are cases that support the argument, Campos v. Bay Town, 840 F.2d 1240, 1244, 1244 (5th Cir.1988); League of United Latin American Citizens, Council No. 4434 v. Clements, 986 F.2d 728, 785-86 (5th Cir.1993); Bridgeport Coalition for Fair Representation v. City of Bridgeport, 26 F.3d 271, 275 (2d Cir.), rev'd on other grounds, 512 U.S. 1283, 115 S. Ct. 35, 129 L. Ed. 2d 931 (1994); Badillo v. City of Stockton, 956 F.2d 884, 891 (9th Cir.1992); Concerned Citizens of Hardee County v. Hardee County Board of Commissioners, 906 F.2d 524, 526 (11th Cir.1990); contra, Nixon v. Kent County, 76 F.3d 1381, 1393 (6th Cir.1996) (en banc), but the Supreme Court has reserved the issue, Growe v. Emison, 507 U.S. 25, 41, 113 S. Ct. 1075, 122 L. Ed. 2d 388 (1993), and its problematic character is vividly shown by the present case. The studies of black and Indian voting on which the tribe relies are limited to voting in Presidential elections — a far cry from voting in county board elections — and particularly inapplicable to the only black population to which the tribe's plan pertains, namely the black residents of the Job Corps center. So far as the record shows, the only thing they have in common with the Potawatomi Indians who live in the proposed district is that they are not Caucasian. The Indians are among the most rooted inhabitants of Forest County, having lived there since the 1880s. The black residents of the Job Corps Center are the least rooted and most transient. So far as appears, they do not mix with the surrounding population. And while the Indians and these blacks may conceivably have similar preferences in state or federal elections, the suggestion that they have similar preferences, different from those of whites, concerning road maintenance in Forest County strikes us as ludicrous. County roads are not a racial issue.

It is no surprise that the residents of the Job Corps center apparently do not vote at all in local elections. On election day, assuming continuous turnover, the average remaining stay of such a resident is only 94 days (188/2). Even if a resident had strong views on county roads, the election of a supervisor sharing those views could not be expected to have any actual effect on road building or road maintenance until long after the resident had departed. A supervisor's term does not begin on election day, and contracts for road building and maintenance are not let and performed instantaneously.

The tribe admits that it has no evidence that the Job Corps residents have any interests in county government that are in common with those of the Indians, but argues that since no evidence is available on the question the studies of state and national elections should carry the day for it. But when the party with the burden of proof cannot obtain evidence to sustain the burden, he loses.

AFFIRMED.