Seamus wonders whether politicians can signal sincerity by advocating policies that they know will be unpopular: the doctor who prescribes bad-tasting medicine may be more credible than the one whose remedies taste sweeter.
As with everything else, there's a literature on this. Long story short: politicians do better by pandering.
Let's run a Wittman line as a starting point. Specify two parties and two potential policies. Party A says policy 1 is best for the country; policy 1 is generally disliked by voters. Party B says policy 2 is best for the country; voters' priors tell them policy 2 is better. A bunch of economics wonks jump up and down about policy 1 being the better one; a bunch of talk show demagogues say economists can't predict anything. If voters are rational truth seekers who care about achieving the best outcomes, they'll discount the demagogues and vote for Party A. In that world, pandering can't be maintained, just as a used car dealer can't pass off a Trabant as a Porche.
Add the Caplan / Brennan twist that because voters really don't affect outcomes, they've little incentive to vote instrumentally; they vote to ally with groups with which they affiliate, to express their preferences about what the world could be like, to tell themselves and others what good people they are. Pandering can work in this world.
Now, look out into the real world. If Seamus were right that advocating sensible policies were best, or even were occasionally best, then we'd expect a whole lot better policy to emerge from the political process. Political markets are competitive; if voters wanted the effective but unpalatable medicine, they'd vote for it. Pointing to the NZ reforms of the 80s just says that voters will recant from Christian Science when the deathbed otherwise looms.
As for the lit, here's a couple worth checking: