Let's say a medium-sized 'Con has 100 full members who get to have a say in running things (scheduling panels, budgeting for Guests of Honor, where to put the screening rooms and wargaming rooms, etc). To simplify things a lot, let's say that 45 of them are grognards (hardcore wargaming fans) and 55 are otaku (hardcore anime/manga fans). It used to be less polarized, but the yearly fee for full membership has resulted in only people who Care A Lot About One Thing ponying up the fee. Because 100 people is a bit unwieldy, they elect a 5-person steering committee to make most of the decisions, with the full membership ceding most responsibility for convention planning to those five.

    But how do they choose the committee members?

    In a "school board" model, where everyone can throw their hat in the ring and the top five vote-getters win, the board will probably end up with 3 otakus and 2 grognards. The problem with this model is that it's very sensitive to how many people run...at the extremes you might have empty seats or everyone getting one vote from themselves and none from anyone else. And in the middle there's two other problems: what if no grognards run, and what if someone everyone hates runs but it's a field of five? You could end up with a pretty dysfunctional committee.

    In a "U.S. Senate" model, you'd have five separate races, usually with one otaku and one grognard running, and elect just one person each cycle (perhaps have two elections a year with 30-month terms). In a very polarized situation, where no one crosses fandom boundaries, this would result in a committee of all otakus, a situation the grognards aren't likely to accept. In a less polarized setting, where people are willing to vote for qualified people of the other fandom, you'd probably end up with 3-4 otakus. The problem of low hat-in-ring participation is solved because you only need two people willing to run in any given election, and if a total jerk runs it'd be easy enough to find someone willing to run against them. The downsides include the fact that if you have a high participation rate, you could have people winning with only 20 votes in their favor (or have to mess about with run-off elections), and if three otakus and one grognard run at the same time the grognard might win thanks to a split otaku vote.

    In a "U.S. House" model, the five seats on the committee are broken up geographically, so that roughly one fifth of the membership each votes on a separate representative every year. As long as even one person will run in each district, you have a full committee. Let's say that this convention is dominated by people from one state, so a simplistic model might split the state into quadrants with one seat for all out-of-state members.

    The first problem, obviously, is that the sizes of the districts might be pretty uneven. There might only be ten out-of-state voting members, which would give them a disproportionate voice on the steering committee (although it could be argued that since they have to make the largest financial committmen to attend, that's okay). You can also get a district where one of the two groups is so marginalized that people might drop their membership on the grounds that their vote never matters.

    Let's say that some grognards volunteer to use their map skills to solve that first problem, and create districts with more equal numbers. Aware that this could be abused, the membership as a whole agrees to this so long as two rules are followed: all districts must be geographically contiguous (i.e. no age-based districts, or putting all the college students in one district, etc), and each district can have no more than 10% more or fewer members than 1/5 of the total (in case the membership fluctuates). So they could pack 22 people into a district if it made geographical sense, or have as few as 18.

    And here's where the gerrymandering can commence. Armed with membership lists that make it possible to at least make a good guess at whether someone will vote otaku or grognard, the grognard district-making committee sets out to stack the steering committee with fellow grognards.

    Noting that the out-of-state district is already heavily grognard (they're more willing to travel long distances to get in a bunch of games), they grab a corner of the state that's slightly pro-otaku to create a district with 12 grognards and 8 otaku. It can be defended as "fair" because they combined grognard- and otaku-heavy groups, but it's a safe bet that grognards will get elected in the new "hinterlands" district.

    That leaves 80 people, broken down as 33 grognards and 47 otaku. One of the big cities in the state is a hotbed of otaku, so it's easy enough to create a nice contiguous district with 18 otaku and 2 grognards. Now there's three districts to go, with 31 grognards and 29 otaku. Center another district on another otaku-hotspot city with 15 otaku and 5 grognards, which leaves you two districts holding a total of 26 grognards and 14 otaku. It's easy enough to create two majority-grognard districts out of that, let's say a nice clean line down the middle of the state (curving around the cities already districted) generates a pair of 13-7 splits.

    So, now we have five districts with 20 each, and it looks vaguely like a yin-yang symbol, a bit of social engineering that's bound to make some of the otaku happy. Two cities, most of the rest of the state outside those cities split in half, and the hinterlands. It doesn't even look particularly abusive, and can be defended as letting the urban members get together with their representative more easily. But let's look at the voting patterns:

• District 1 (city A): 18 otaku, 2 grognard.
• District 2 (city B): 15 otaku, 5 grognard.
• District 3 (state A): 7 otaku, 13 grognard.
• District 4 (state B): 7 otaku, 13 grognard.
• District 5 (hinterlands): 8 otaku, 12 grognard.

    The numbers are a lot more complicated in regular elections, but the basic ideas are the same. If a group with minority support can get put in charge of drawing the lines for voting districts, it's pretty easy to stay within pre-computer-age rules and still make sure that the minority party can get and keep a majority of the seats. Once in a while you'll get blatant gerrymandering, such as a district that's two Democrat-heavy cities connected by a corridor ten feet wide and twenty miles long, but thanks to computerized map-making programs you can generally be a lot more subtle than that unless you want to give the other party a big ol' middle finger.