WHEN YOU talk about a soft drink without naming a brand, do you call it 'pop', 'soda', or 'coke'? Or something else?
What you call it depends on where you grew up. It's one of those words that changes depending on where you are in the nation (or the English-speaking world), like 'cellar' vs. 'basement'. The map on the right is the result of a project by Alan McConchie, who appears to be a student at Cal Tech. His Pop vs. Soda web page collects user responses as to which word they use, and plots it by zip code. The scripts are written in Perl, and there is a real-time plotting function in Java, which is pretty cool, but a little too low-rez, IMO. The map on the right represents the data a little better, and the large version does it better yet, but I suspect neither is up-to-date with the real-time Java version.
I find this fascinating, with the caveat that it isn't terribly scientific, and the data-gathering mechanism isn't very robust. The people who submit data are self-selected, not a random sample, and there are apparently no mechanisms to prevent someone from skewing the data by making hundreds of false entries. As, for example, in the infamous People Magazine 1998 web poll for the Most Beautiful Person of the Year, which People expected to be won by the likes of Leonardo DiCaprio and Kate Winslet, and which was won instead by a write-in campaign for Howard Stern creation Hank the Angry Drunken Dwarf.
However, if we accept the data as roughly accurate... There's an obvious pattern.
And it's 'Red vs. Blue'. It's the pattern of the 2000 presidential elections, in which the coasts and the big cities went for Gore, and the middle of the country for Bush. People who say 'soda' are likely to be Gore voters, people who say 'pop' or 'coke' are likely to be Bush voters. Since the election, a lot has been made of this geographic pattern, and what it means. Of whether there is a profound cultural split in the nation, that has a geographic aspect that mirrors its social/aesthetic/religious aspects.
BTW, the idea of doing this kind of map of the geographic dimension of word choice is not new. Linguists have been doing this for decades. However, I haven't seen it done in real-time before, on the Web, and most of the previous examples I know are rather small scale, e.g., mapping the use of the terms 'farmer's cheese' vs 'pot cheese' vs 'cottage cheese' in Massachusetts.