Dan Balz had a great column over the weekend describing a key finding from political scientists: Presidential speeches don’t achieve the goal of changing public opinion.

Surely Barack Obama knows that, or at least someone in the White House does. Perhaps the White House doesn’t pay close attention to obscure academic journals, but this finding has been well publicized recently, whether it’s by Ezra Klein in the New Yorker and on Wonkblog, by other liberal reporters or by political science bloggers.

So why make a high-profile and yet almost certainly futile effort to change public opinion?

It’s a great question, because it points to one of the key facts about the presidency: Presidents often “must” do things just because everyone expects them to.

For example, presidents don’t have to give their State of the Union message before a joint session of Congress. But ever since Woodrow Wilson began the tradition, it’s become so ingrained that any president who refused would risk being denounced. Presidents don’t have to give formal press conferences, either, but again it’s a part of the expected presidency, and there’s a price for not doing them, even if the president substituted (say) much more informative and in-depth sit-down interviews.

At the same time, the meaning of presidential actions is in part determined by the traditions and norms of the office. So what Obama says in his speech tomorrow night and his network interviews tonight isn’t nearly as important as his doing it in the first place. What he’s saying, in effect, isn’t whatever rhetoric he adopts but rather (as George H.W. Bush might have said): “Message: I really care about this.”

Presumably, that could have an effect in a number of ways. It certainly can keep Syria atop everyone’s agenda, especially the news media’s. It may prod those who share the president’s party label to think twice before they abandon him. It may, on the other side of the aisle, be a price the president should be forced to pay — if an unpopular strike in Syria is to happen (or at least one that polls badly in advance), then the opposition party wants it identified as closely with the president as possible.

There’s more. It’s also a signal that the president is willing to risk his reputation for honesty on the claims he’s making, which (all else equal) actually provides stronger evidence that those claims are really true. You may think that’s silly; shouldn’t everyone look at the evidence itself? But some evidence is going to be classified, and all evidence is going to be subject to subjective evaluation from experts — and members of Congress are generally not experts. They ultimately have to trust someone. The best bet for them, often, is to raise the stakes for getting it wrong, thus pushing the president to only make claims on which he (and his own experts) are as confident as possible.

Obviously, this all did not work with George W. Bush and Iraq, but that’s precisely the point: When Bush was caught, everyone (including Republicans) stopped trusting him, which cost him again and again when he needed that trust. In other words, a strong incentive to get it right can’t guarantee that Barack Obama does get it right, but it’s a very real, and very strong, incentive. And it’s often the best that Congress can do.

So when Obama speaks tonight and tomorrow, the first point is simply to note that he’s doing it, and to understand what that means. And the second point is to pay especially close care to what claims he makes and doesn’t make. The odds are that everything has gone through one more scrub to eliminate as much chance of error as possible.

Just don’t focus too much on reactions from the mass public, because we already know that presidents can’t really do much to change that group’s mind.