Imagine a scene in which a cop and a parole officer are grilling an ex-con about a fire set in the foothills of Southern California. What's wrong with the following block of dialog and dialog attribution?

“You underestimated the speed of the fire. You set it at the top of the hill and ran down. People always underestimate brushfires around here when the Santa Anas blow. You probably didn’t think the fire would get that big and move that fast. But it did, and if you’re lucky and smart enough to cut a deal, you’ll fall a second time for manslaughter. If you’re not smart enough to cooperate, you’ll go down for murder,” the cop said.

From the way the dialog is attributed, it's not clear who is speaking until the very last line, after the speaker has given a 30 second speech. It could be the cop, or it could be the parole officer speaking. If the concept behind dialog attribution is to let the reader know who is speaking, the above use of attribution comes too late to help the reader hear the dialog.

When attributing dialog that contains more than one complete sentence or a couple of phrases, consider inserting the speaker’s identity earlier in the block of dialog, rather than later. Why? Because the character who is speaking is important to the dialog’s interpretation, and ear. If you promptly tell the reader who is speaking, the reader can then hear the character’s voice in the dialog.

When revising the example quoted above, where would you place the attribution, letting the reader know who is speaking?