Ethan Brown's Snitch is rich in description and anecdote, but the author nearly chokes on his outrage and, as a result, fails to come to grips with the hard choices.

Here are the Brown's basic points, as I understand them:

1. In investigating and trying drug-dealing, organized-crime, and corruption cases police and prosecutors rely heavily on criminals who testify in return for reduced sentences.

2. Those witnesses are massively unreliable, especially since the draconian sentences now handed out under the drug laws create intolerable pressure. (Brown doesn't, but I would, compare giving someone a 20-year sentence and offering him an "out" in return for testimony to the practice of judicial torture.)

3. As a result, innocent people go to prison.

4. Some prosecutors and investigators are insufficiently skeptical about the stories informants tell.

5. Since people with greater involvement in a crime are more likely to have, or be able to concoct, information useful to prosecutors, the result can be an inversion of justice, in which the big fish get off lightly and the small fish bear the full brunt of punishment.

6. Some of the people who get sentence reductions in return for their testimony go on to commit terrible crimes.

7. Some investigators go too far in protecting informants.

So far, Brown and I entirely agree. These are important facts, and underappreciated by journalists and legislators.

But Brown goes further: he thinks there's something intrinsically wrong with testifying against co-conspirators, and thinks that a return to an imagined earlier period where there was honor among hustlers and no one snitched would represent moral progress. And he's strongly sympathetic with the "Stop snitchin' " movement, which he sometimes identifies with the "stand-up guy" ideal and sometimes claims stands in opposition only to dishonest informant testimony.

Brown acknowledges that witness intimidation is a major problem in urban law enforcement, especially with respect to murder. But he decries the identification of "Stop snitchin' " with witness intimdiation, as if the throat-cutting gestures he describes murder defendants as making to witnesses against them had no relation to the "Stop snitchin' " T-shirts. He seems to imply that the witness-intimidation problem could be solved by creating local equivalents to the Witness Protection Program operated by the Feds (which elsewhere in the book he denounces because some of the participants commit crimes).

I would offer an alternative account:

1. The investigation of transactional, conspiratorial crimes relies on intrusive law enforcement: electronic surveillance, undercover operations, and informant testimony. This is a reason for caution, and a disadvantage of the current policy of massive drug law enforcement.

2. Informant management has always been, and will always be, a nightmare. Most "cooperating informants" are very bad people, and their sworn testimony isn't worth the spit behind it. They're also likely to continue to commit crimes, since committing crimes is both their profession and their avocation.

3. Sowing distrust among potential conspirators is a socially useful accomplishment.

4. A "stand-up guy" may be admirable in contemplation, but in real life a "stand-up guy" helps keep the underlying conspiracy in business. Yes, in some weird sense Gordon Liddy and Scooter Libby displayed a kind of courage and integrity by taking the heat themselves, but John Dean did a public service by "snitchin' " on Richard Nixon and his cronies. The "Stop snitchin' " movement among police is called the "blue wall of silence." Does Brown think that a good thing? Neither do I.

5. Yes, of course some informants go on to commit terrible crimes. But some of the people they help put away are thereby prevented from committing equally terrible crimes. Brown offers no evidence that the "Stop snitchin' " movement in practice distinguishes honest from dishonest informant testimony, or the testimony of "civilians" from the testimony of "hustlers." And he offers no practical suggestions about how to deal with the witness-intimidation problem, which has driven homicide clearance rates in cities such as Baltimore, Washington, and New Orleans to well below 50%. Yes, you read that right: in those cities, you have a better than even chance of literally getting away with murder.

Gang homicides almost never prosecutable because none of the witnesses dares to testify and because the victim's friends are bound by the "Stop snitchin' " ethic. Any serious attempt to counteract the gang violence that terrorizes so many neighborhoods has to start with convincing people that testifying against bad guys is a civic duty.

I'd like to see shorter sentences for drug crimes not involving violence, a switch to drug-enforcement tactics that emphasize protecting neighborhoods rather than preventing transactions, better controls on the use of informants (even at the expense of making fewer drug-dealing cases), and more skepticism by prosecutors, investigators, and judges. But I'd also like to see the street (and prison) gangs put in the position eventually occupied by the Klan, the Mafia, and the American Communist Party: so informant-ridden as to be unable to function.

Yes, "Stop snitchin' " is a terrible indictment of the state of police-community relations in many of the neighborhoods where the need for better law enforcement is greatest. If I were running a police department, I'd take that as a primary problem to work on. But the idea that we could improve public safety or serve justice by drying up the flow of testimony by criminals against their co-conspirators doesn't appeal to me at all.