As I've discussed at length, but not recently, there's more to representation than just policy promises. That's on of the main findings of Richard Fenno's work: promises also include all sorts of things, including style of representation. Among other things, that to me, is the "solution" to the question of whether representatives "should" be delegates or trustees; the answer is that they should do whatever they said they would do when they were campaigning. But that's just one of the many things that can go into representative style and the representative relationship that politicians develop with their constituents.
At any rate, one of the things I find most interesting is why any of this happens. Ragusa is good on this:
Finally, if citizens are (a) unaware of who their elected lawmaker is and (b) as a result woefully ignorant of their representatives’ position on key votes, the question is: what keeps lawmakers honest? The answer is that while private citizens may not know how their representative or senator voted, general election opponents and interest groups sure do. Thus, while the electorate is generally inattentive to lawmakers voting record, the reelection incentive—and the threat of attack ads from one’s rival—keep lawmakers honest.
Indeed, in the underlying paper, Tracy Sulkin argues that risk-averse politicians will likely attempt to keep their promises as if they would get in trouble from abandoning them. That sounds right to me, as is her argument that politicians are probably likely to make promises in the first place that they have an interest in keeping.
What fascinates me about this, however, is that I really don't buy the idea that politicians, no matter how risk-averse, are really keeping promises because of electoral incentives. It just doesn't wash that it's a significant enough constraint. Especially when it comes to promises about representative style. What we know about the ignorance of voters is just very hard to square with the idea that they will punish their representatives' misbehavior.
Sulkin's suggestion that politicians may choose to emphasize particular issues (and presumably style) for personal reasons, rather than for electoral reasons, is promising.
But overall, I think it's one of the more puzzling problems in representation, and therefore in democracy. I have an old paper in which I try to get to it (it involves political parties, Arendt, and the "touch of Harry in the night" scene, and yes, I suppose I should try to do something with that paper). I do think that representation is real, and really "works," and that the fact it works is central to how democracy actually does what we want it to do...but the whole topic is filled with difficulty.