Maybe it's time to start from scratch in Congress.

From left, Devin Goins, Victoria Flores and Daniel Paterson, all from Chattanooga Occupy, protest on the west lawn of the Capitol. (Melina Mara/The Washington Post)

At least that's the message voters are sending in a new NBC News/Wall Street Journal poll released Wednesday. Nearly six in 10 (57 percent) said they would vote to defeat and replace every single member of Congress -- including their own representative -- if they could, a high dating back to 2010. Just 39 percent said they would not.

The response is striking, even as it's become crystal clear in recent years that Congress's overall image is in the toilet. (Americans give Congress an embarrassing 12 percent approval rating in the survey.) Why? Because while Congress as a whole is unpopular, incumbents continue to keep their jobs at an impressive clip. Roughly nine in 10 members of the House and Senate were reelected in 2012.

Not bad for a body less popular than cockroaches.

So do the new numbers mean we are going to see major turnover in Congress come 2014? Not necessarily. Consider that in January of 2012, 56 percent of voters said in the same survey that they would vote to defeat and replace every single member of Congress, nearly identical to the percentage saying the same thing now. But, that clearly didn't prompt voters to kick out their members en masse last fall.

A Gallup poll released in May showed that respondents approved of the job their representative in Congress was doing much more than they approved of Congress as a whole. Call it the institution-vs.-individual paradox.

What the latest NBC/WSJ numbers suggest is that voters have grown very weary of what they have seen from members of Congress, including their own. And it's not a partisan issue. Voters are split down the middle over whether they prefer a GOP- vs.-Democratic-controlled Congress.

Still, the devil-you-know attitude voters have displayed will probably keep many members of the 113th Congress in place for the 114th Congress.