The losing guess

TWO 20-year-olds stood accused of robbing a man who had been dozing on a park bench one night. According to the prosecutor's ``Statement of the Government's Case,'' after the men had run off with the complainant's wallet, he flagged a police cruiser.

While riding in the back seat, he spotted the culprits standing on the curb. The police detained them, and he confirmed his identification from a face-to-face distance. Arrest followed immediately, and indictment later.

If, as was true of both men, a defendant cannot afford a lawyer, the Constitution requires the judge to appoint one. Most states (and the federal government) have created special organizations, generically dubbed ``public defenders,'' to provide the necessary representation.

The same lawyer, however, must not represent two defendants accused of the same crime. So careful are the courts to prevent this risk that they usually extend the prohibition against dual representation to entire law firms. Thus the public defender's office could represent only one of the accused. The judge appointed a private lawyer for the other.

The odds, of course, strongly suggested that the case would not reach trial: About 85 percent of defendants plead guilty. The process that produces this result is accurately called ``plea bargaining.''

Although other variables affect plea probabilities, the most significant is the strength of the government's case. With weak evidence, a prosecutor will often agree to ask for a light sentence, rather than risk acquittal. Conversely, when the state's witnesses are solid, defense counsel will advise the client to accept whatever concession the district attorney may offer.

In the case of the park bench robbery, conviction seemed so certain that both defense counsel quickly reached accord with the prosecutor. The private lawyer's client accepted the deal and changed his plea forthwith. His co-defendant, however, flatly refused, despite the urgings and (presumably) the dire predictions of his experienced public defender.

Once the trial began, the victim, whose evidence had appeared so compelling on paper, dissolved like ice in August. His eyesight turned out to be bad, his memory worse. The jury promptly returned an acquittal.

Different juries, hearing essentially the same evidence, sometimes reach contradictory results: During the Teapot Dome scandal, one jury convicted former Interior Secretary Albert B. Fall of accepting a bribe; another acquitted Edwin L. Doheny of tendering it.

It was, therefore, neither illogical nor improper that one man should confess, while his confederate escaped. Like many other decisions in life, a guilty plea is a prediction, based on then-available information; like any prediction, it may prove less than completely accurate.

Even as he concluded that justice cannot demand 20/20 foresight, the judge felt uneasy. Should the defendant who pleaded receive credit, as it were, for the other's acquittal? Should the judge try to forestall a possible later claim that in letting his client plead, the lawyer had demonstrated incompetence?

Knowing the evidence, the judge seriously doubted that any jury would have convicted; yet what he learned during the change-of-plea procedure convinced him that the man had committed the crime.

In the end, it was that thought which solved his dilemma. No criminal-justice system can work if defendants who are in fact guilty routinely demand trials. Nor can it operate if any defendant who comes to believe, for whatever reason, that he has made a bad bargain can change the deal directly or indirectly.

The system can guarantee defendants only an informed choice. And it must measure the fairness in terms that are absolute, not relative. A late winning guess does not justify giving the early losers a second chance.

Hiller B. Zobel sits on the Massachusetts Superior Court.