Every so often, I watch Law and Order, and see Sam Waterston twist the law somehow to catch a bad guy, and I get that happy sense of victory you do when the cinematic good guys win.

Somewhat later, reality sets in. The law isn't a game. (I know lawyers will disagree with me--so, okay, the law shouldn't be a game.) Obviously, the law is complicated, there will always be boundary cases where a hard law produces and unsatisfying result, and so forth. But we shouldn't celebrate this tendency.

I don't fault only the prosecutors, by any means--it's not clear to me why we have the exclusionary rule, rather than some other means of punishing government officials who poke their nose where it doesn't belong. The officials are, after all, not the ones who actually suffer when a guilty person goes free because they pushed the boundaries of a warrant. And I'm pretty unexcited about restricting juries from hearing various forms of evidence because it might taint their delicate little minds.

But there is a qualitative difference. A defender's job is to get his client off. A prosecutor's job is to serve the public, and the law, not to rack up convictions. Nor to find inventive ways to stack the deck against defendants. Prosecutors abuse their power, often from the best of motives, and worse, we hand them the motives by demanding jail time rather than actual justice.

Of course, it strikes a special chord in my heart when the abuser is the tax man.

From Joe Kristan:

Occasionally tax issues arise that affect many taxpayers. Tax shelters, for example, can be sold to hundreds of individuals. It doesn't make sense to issue hundreds of identical decisions. To avoid results like the 90 virtually identical Antarctica foreign earned income exclusion decisions that have been issued in the last couple of years, the IRS and taxpayers agreed to resolve a set of cases involving the "Kersting" tax shelters. After test cases were tried, the parties agreed to "stipulate" the remaining cases based on the result of the test cases.

Then the IRS attorneys decided to stack the deck. They worked out a favorable secret settlement with the taxpayers in the test case. Perhaps not coincidentally, the taxpayers then didn't defend the shelter successfully.

The Tax Court had resisted applying the secret settlement to all similar taxpayers, but following a reversal by the Ninth Circuit, they changed their mind. The Hartman decision issued yesterday ordered the IRS to apply this secret settlement to all of the taxpayers involved in the shelter to correct "a fraud on the Court." It is no small group; according to the tax court's decision yesterday: "As of Mar. 13, 2008, 1,173 Kersting project cases remained on the Court's inventory of docketed cases in which decisions have never been entered."

The court ordered the IRS to administratively adjust the accounts of all of the Kersting project taxpayers.

Paul Caron also has a long post on it, excerpting the Court's decision, which reads in part:

“Men must turn square corners when they deal with the Government.” ... “To say to these appellants, ‘The joke is on you. You shouldn’t have trusted us,’ is hardly worthy of our great Government.” To tell Kersting project petitioners they should not have trusted respondent to try the test cases honestly and fairly and the Tax Court to formulate an appropriate sanction when respondent failed to do so would be equally unworthy. ... “Wisdom too often never comes, and so one ought not to reject it merely because it comes late.” ...

Respondent’s attorneys committed a fraud on the Tax Court during the Kersting test case proceedings that was a fraud on the Court in every case bound by the results of the test cases. Extending to every petitioner whose case was bound by the results of the Kersting project test cases, by piggyback agreement or the Court’s order to show cause procedure, the benefit of the Thompson settlement strikes us as an appropriate accommodation of the competing considerations; it is a sanction for the misconduct that is consistent with Dixon V and is “no more than necessary” to maintain public trust in the judicial process that employs test case procedures. ... We are protective of the integrity of our judicial process and concerned about deterrence. We are “entitled to send a message, loud and clear.” .... We hold that sanctions should be imposed in the cases of all Kersting project petitioners in which stipulated decisions were entered on or after June 10, 1985, the date the Kersting project test case proceedings began.

In an email, Paul follows up:

As often happens in big cases (because there is no Tax Court procedure for class actions), the IRS and investors in this tax shelter agreed to try one case as the "test case" to decide the issues and then the other cases would follow it. But the IRS reached a "secret settlement" with the test case taxpayers, thereby screwing the taxpayers in the other 1,300 cases. The U.S. Court of Appeals for the Ninth Circuit exposed all of this and ordered the Tax Court to sanction the IRS by imposing the same settlement on the remaining taxpayers. But the IRS entered into settlements with some of the other taxpayers on less favorable terms (to the taxpayers), and the Tax Court in the earlier Lewis opinion refused to reopen those settlements. The Tax Court yesterday changed its mind, finding that the IRS's conduct was so sleazy that it should impose the same sanction in all 1,300 cases.

The opinion is extraordinary -- it is rare for a tax court judge to quote ethics and morality, via Aristotle, Hobbes, Rawls, in criticizing the IRS's conduct (page 74 of the opinion).

It's nice to see a court reminding the IRS that its mission is not to collect the most taxes from the most people, but to ensure that tax law is accurately and fairly applied.