So this time the White House really lost a battle, right?

In the wake of overwhelming public pushback against his deeply-unpopular policy of separating children from their migrant parents, President Trump on Wednesday signed an executive order to allow detained families to be housed together. Given how reluctant this administration is to ever admit error, backing down in this manner would seem to suggest that the administration grossly miscalculated.

Such a comforting conclusion is unfortunately premature.

First of all, it's too soon to call the president's executive order a retreat. The order almost seems designed to run afoul of immigration law — specifically, the Flores consent decree that forbids detaining minor children indefinitely. If the president's order is challenged in court and his zero-tolerance policy is kept in place, families would continue to be torn apart, but Trump could turn around and blame high-handed judges for ordering children pulled from their parents' arms, just as he had previously blamed the Democrats for causing the crisis by not funding his beloved wall.

But even if the administration ultimately does retreat, and at least temporarily abandon its zero-tolerance policy, that doesn't mean the White House was denied what it really wanted: outrage.

The Atlantic's McCay Coppins calls the policy's principal architect, Stephen Miller, the administration's troll-in-chief, and it's an apt description to apply to the policy itself. That policy wasn't accidentally implemented in a way that provoked anger at its excessiveness and cruelty. On the contrary: No attempt was made to implement that policy with a view to efficiency, effectiveness, or minimizing either the practical or political fallout. It was actively trying to be what we would normally call "unwise."

Why do that? Precisely because it sharpens political divisions and thereby breaks down the process of deliberation. That's trolling.

We're familiar with trolls in the online world, where they can easily destroy fora for deliberation by making them inhospitable to people who want to deliberate, so that those people eventually leave. Moderated fora face a constant challenge to maintain enough openness to allow for a free and unbiased exchange of ideas without turning those spaces into a breeding ground for trolls.

But a troll takeover is almost universally deplored by those who once valued a forum destroyed by them. Why, then, should trolling ever work politically? Why shouldn't their behavior always be counter-productive, a problem for their co-partisans rather than their opponents? The answer has to do with the deformation of our politics, and the breakdown of our political community.

Members of a political community generally share a set of values, habits, and principles that define the ethos of the community. But, inevitably, these exist in tension with one another. We want freedom and security and equality and a variety of other goods that cannot all be maximized simultaneously. So simply following a rule — whether derived from reasoned principle or from habit — is insufficient to resolve all potential conflicts. That's why political communities require leadership that manifests good judgment — what Aristotle called phronesis or practical wisdom. In a well-ordered democratic republic, prospective leaders use argument to convince the public that a given course of action strikes that proper balance — or, more correctly, that he or she is the right person to discern that balance.

But when the political community breaks down — when there is no longer a shared conviction on what we value — that process of phronesis breaks down as well. Public argument is no longer about how best to achieve our collective ends, nor even about which ends are more important, but first and foremost a demonstration of fidelity. Which is to say, not an argument at all.

Good judgment should not be confused with compromise or moderation. "Extremism in defense of liberty is no vice" is an argument for being more vigilant about protecting liberty than about securing other goods (such as alleviating poverty or social injustice). It presumes a shared appreciation of liberty. Dick Cheney's "1-percent doctrine" towards the possibility of nuclear terrorism is, similarly, an argument for extremism that presumes a shared appreciation of security. These arguments may be persuasive or not — and they may have been wise or not — but they are intended to persuade. If the targets conclude that these arguments show bad judgment, they will be rejected.

Trolling has different goals. Its governing assumption is that opponents do not have a shared appreciation of the value in question, and it aims to expose that divide. And that exposure is far more important than any other aim, including even realizing the objective of the policy.

Which brings me back to Stephen Miller.

He characterizes his opponents as believing in open borders and rejecting the idea of an American nation. His primary policy aim is not to convince them otherwise, nor even to secure the border from illegal entry. Rather, the objective is to demonstrate the administration's fidelity to its own values, and to force others by their very outrage to demonstrate their opposition to those values. As Coppins writes, "for Miller, the public outrage and anger elicited by policies like forced family separation are a feature, not a bug."

If that's Miller's goal, then wise policymaking is not only unnecessary but actually counter-productive. The worse the policy, the more sure it is to be furiously opposed — and thereby to reinforce the division of the country into mutually hostile and unpersuadable camps. But a moderate response to such a policy is arguably worse, as it effectively rewards the strategy of trolling and encourages its repetition.

This is the dilemma on whose horns the opposition to Trump has been impaled since he first emerged as a political force, and even a leader well-endowed with phronesis might hesitate to chart an escape. In the land of trolls, wisdom itself is a sign of weakness.