Hank, thanks for that post. It will take a while to process, and I'm afraid that my time is pretty limited these days. So I'll try to bite off a few bits at time. The basic question here is about decision-making and the costs and benefits of various approaches and how to evaluate particular issues to determine which approach is appropriate for that issue. The big question of trying to apply some generalized system and draw conclusions about whether we have enough or too much of one approach or another is an important one, but it's more than I'm willing to bite off without working through some of the ancillary issues.
First, I want to agree with you on the stupidity of this unidimensional visualization of politics that is popular in the U.S. There is a left and a right and various points in the middle and on the extremities and that's that. Not only is that inaccurate, but I really believe that this metaphor tends to promote partisanship and stifle substantive policy discussion. A real policy dialogue is founded on common ground, and it is much easier to find common ground with other parties when you visualize policy as existing on many dimensions rather than as a point on a line. And it trivializes everything. Somehow Joe Lieberman becomes a moderate Democrat, when he's not moderate about anything. He is a zealous social conservative and an unrepentant fan of big and aggressive (and irresponsible) government (is it any surprise that he has had a love affair with this administration). But somehow, under our unidimensional view of politics, his two opposing extremisms cancel out to create a "moderate". Hmm.. This has been a pet peeve of mine for a long time..
Anyway, on to my main point.. I'd just like to hash out some of the territory here. Some of what you describe about presidential politics is, I think, referred to as the "equilibrium displacement" theory of voter power. Or as it has been referred to in two of my classes it has come up in, the hot dog vendors on the beach theory. Say you have a beach full of people, and two hot dog vendors trying to sell to them. Where do the hot dog vendors inevitably set up? Side by side at the median point of the locations of the beach-goers. That is, I believe, a Nash equilibrium (neither side can unilaterally improve his position). The only power of any individual on the beach is to switch to the other side of median point. The median will shift one person over.
That is really the traditional view of American presidential politics. I think there are some real problems with that model, but to the degree that you subscribe to it, perhaps there is some sense in which the political spectrum is linear. No matter how many dimensions there are to issues, the positions will boil down to the two hot dog vendors and you can define a line that runs through the two points. Of course, I don't think this model properly accounts for a) voter turnout (and not so much strategic voter turnout (i.e. your Shotgun News story) as much as general voter turnout issues) and b) that there is a fairly broad range of "acceptable" policy positions beyond which non-policy issues (the charismatic frontman, the TV commercials, gamesmanship, etc.) are more significant.
The Shotgun News thing raises another issue: strength of preference/logrolling. If owning an AK-47 is worth 25 points of utility to you, having low taxes is worth 10 points of utility, and having high steel tariffs is worth 5 points (and those are the only three policy questions in the election), clearly you're going to be willing to compromise on taxes and tariffs in order to have the gun policy you want. And I don't think that is inherently problematic. People have varying strengths of interests and should be able to express those in voting preferences. I personally will overlook many policy differences in order to find candidates who support aggressive action towards clean and accountable government, campaign finance reform, and balanced budgets. Add in sane foreign policy and that's nearly my entire voter preference list. The single issue voter is only an extreme iteration of this.
Logrolling is problematic where people are logrolling for immediate financial gains. In general I'm referring to situations where commercial interests are engaged in the logrolling, and are using lobbyists and campaign contributions to do it. They may not care about any other policy except getting the earmark that delivers them a $5m contract, or an agency regulation that puts a competitor out of business. I think you have to distinguish that sort of logrolling (the car seat example) from the logrolling involved in individual voters deciding who to vote for (the shotgun example). The former, I agree, is a serious problem, but I have no objection to the latter.
I think that's about all I can handle for now. For next time: policy bundling in presidential elections. And I'll look into updating the general appearance of the blog, although it might not happen for a couple weeks...
edit: I should add, having now read Ryan's comment on Hank's post that my statements above assumed that the car seat law was a result of industry lobbying.