Squaring the hexagon, from Strange Maps. This map illustrates a proposal for dividing France into perfectly rectangular (well, as rectangular has something on a sphere can be -- but France is small enough that one needn't worry too much about this) regions, of which there were 80 or so. (As you may note, France is not a rectangle.)
This proposal was made around the time of the French Revolution, and it has the ring of proposals from that period; these are the same people that invented the French Republican calendar and the metric system. The metric system has turned out to be a good idea (although some people will argue that many of its units don't correspond to anything on a human scale). The Republican calendar, for those who aren't familiar with it, broke up the year into twelve thirty-day months, of three ten-day "decades" each (these are analogous to weeks); there seems to be no a priori reason why this doesn't work, except that we're used to seven-day weeks. It's rather inconvenient that seven is prime, in fact; it would be useful to have a unit of the "half-week".
If you look at a Major League Baseball schedule some time, you'll see that that's actually the fundamental unit of baseball scheduling; teams generally play two series a week, one of them running from Monday to Wednesday, Tuesday to Thursday, or Monday to Thursday, and the other from Thursday or Friday to Sunday. You might notice that Thursdays are kind of ambiguous in this scheme; sometimes Thursday is the end of a series, sometimes it's the beginning, and sometimes it's neither. Teams often have off on Mondays or Thursdays. The weirdness of Thursday in baseball scheduling is a direct consequence of the fact that seven is odd.) In fact, I'd argue that the fact that the week has seven days is a lot more inconvenient than a thirteen-month year would be; the fact that thirteen is prime, and that therefore no simple fraction of a hypothetical 13-month year is a whole number of months, is pretty much irrelevant. When was the last time you ever did something that took exactly three months (one quarter) or six months (half a year)?
Returning to geography, drawing straight lines as boundaries often seems to be a bad idea; lines that are drawn without regard for population centers inevitably, if you draw enough of them, pass through some population center and divide it up. This causes the people in charge on either side of the line to ignore the other side in government, making the region less able to function as a whole. Surprisingly, it's hard to think of population centers in the U.S. that lie along a state border that's just a line on the map; most of the big population centers near state borders are along rivers, such as Philadelphia, New York, or Washington (note that even if the District of Columbia didn't exist, Washington would be on the Maryland-Virginia border). This isn't surprising; rivers are natural dividing lines. But I suspect that the situation would be different in more densely populated France. And drawing lines on a map without regard for settlement patterns is a cause of the chaos in the Middle East.
The difference between the U. S. and Europe is that in Europe the lines have evolved along with the historical population patterns (which haven't changed that much even in centuries), whereas in the U. S. a lot of the lines between states were drawn before the states in question were settled. One can't expect the people making the maps to forecast in advance where people are going to choose to live, especially in modern times when the population centers grow up along roads (which people can build) and not rivers (which are where they are).
A lot more thoughts on how territory is often divided up -- countries into states, states into counties, etc. -- can be found in Ed Stephan's book The Division of Territory in Society, text available online. His hypothesis is that there's a relationship between the size of a state, county, etc. and its population density; smaller states within a country, counties within a state, and so on tend to be denser. This can be derived from the assumption that "social structures evolve in such a way as to minimize the time expended in their operation", although it only gives a relation between density and county area. It doesn't make forecasts about shape, and it seeems to assume that densities are locally uniform when this is very far from the case.