‘The best interests of the child’: a retired judge on a standard that decides less than it seems

For twenty-six years, in the family and children's courts of Porto, I decided where children would live. Which parent would have them on school days. Whether a father who had been absent for two years could start again with alternate weekends. Whether a grandmother could take a boy whose mother was not, that year, able to care for him. Every one of those decisions was made under the same phrase: the best interests of the child.

The phrase is everywhere. Article 3 of the United Nations Convention on the Rights of the Child, which Portugal ratified in 1990 and almost every country in the world has ratified since, says that in all actions concerning children, whether taken by courts, welfare institutions, administrative authorities or legislatures, the best interests of the child shall be a primary consideration. National codes repeat it. I quoted it in hundreds of judgments.

Now that I have retired, I want to ask a question I avoided while I was using it. What does it actually tell a judge to do?

The case against the phrase

The sharpest critique I know is a 1975 article by the American legal scholar Robert Mnookin on custody decisions and the problem of indeterminacy. His argument has two parts, and I recognised both from my own courtroom.

First, the standard asks for predictions nobody can make. To know which home is better for a seven-year-old, a judge must forecast how two adults, a child and their circumstances will develop over ten years. Psychologists cannot do this reliably. Judges certainly cannot.

Second, even if we could predict, we would still need to know what counts as better. Is it the child's happiness at twelve, or her stability, her education, her relationship with both parents, her religious upbringing? Reasonable people, and reasonable judges, rank these differently. The phrase hides the ranking inside the judge.

Mnookin's conclusion was that the standard often gives a discretion so wide that the outcome depends on who is sitting. The philosopher Jon Elster later pressed the point further: if two parents are both fit and the judge cannot honestly say which home is better, might tossing a coin be more honest than a long judgment that pretends to know?

I never tossed a coin. But I sat through days when I understood why he asked.

What the critics proposed instead

The alternatives are rules that narrow the judge's discretion. Some American courts adopted a presumption in favour of the primary caretaker, the parent who had done most of the daily care before the separation. Others proposed dividing a child's time after separation roughly in proportion to how care was divided before it. These rules are more predictable, and they reward the work of care that the best-interests standard can overlook in favour of income or a larger house.

They also make mistakes a judge would not. A rule cannot see that a primary caretaker has started drinking, or that a twelve-year-old is now old enough to say clearly where she wants to live. Article 12 of the same Convention gives a child who is capable of forming her own views the right to express them, and requires that they be given due weight according to her age and maturity. No presumption can do that weighing.

What the phrase is really for

So I find myself where Aristotle found himself on equity. In Book V of the Nicomachean Ethics he says that law must speak in general terms, and that where a general rule goes wrong in a particular case, it is right to correct it by saying what the lawgiver would have said had he been present. That correction is equity, and the equitable person is better than the strictly legal one, not because she ignores the law but because she completes it.

The best-interests standard, I now think, is not really a rule. It is an instruction to be equitable about one particular person. It tells the judge where to look, at the child, not at the parents' grievances or entitlements, and it forbids certain reasons. A judge may not give a child to a father because he is the father, or to a mother because she is the mother, or to the parent who wins the argument about who betrayed whom. That is real content. Most of the work the phrase does is excluding bad reasons, not selecting the right answer.

John Locke made a related point about parents three centuries ago in the Two Treatises of Government. He preferred the word parental to paternal, since the mother shares the power, and he argued that it is not a power over the child for the parents' sake, but a temporary trust held for the child's good until the child can govern herself. On that reading a family court is not dividing property between adults. It is deciding who will hold a trust.

What I would tell a young judge

Hart would say the standard is a clear case of open texture: a general term with a large penumbra, in which the judge must use discretion. Dworkin would say there are principles in the law that guide her even there. I spent my career somewhere between them. I was choosing; I was also bound, by reasons I had to state and that others could check.

If I could speak to my younger self on her first day, I would not tell her the standard is empty. I would tell her it is a discipline, not an answer. Write down every reason you are giving. Strike out every one that is really about the parents. Ask the child, if she is old enough, and say in the judgment what she said. And when you have done all that and the two homes are still equal, do not pretend your choice was dictated by her interests. Say that both were good, that you chose, and why.

Children read their files when they grow up. Several have written to me. None of them complained that I had chosen. Some complained that I had pretended not to.