Here is a paradox. It is pretty easy to predict the voting patterns of new Supreme Court justices. But it can be exceedingly difficult to predict the votes of justices in specific cases — which means that it can be difficult as well to predict how those cases are going to be decided.

When President Bill Clinton appointed Stephen Breyer to the Supreme Court in 1994, those who knew Breyer's work knew that his voting patterns would be moderately liberal — more centrist than those of progressive heroes Justices Thurgood Marshall and William Brennan, but to the left, for sure, of Justice Antonin Scalia.

And when President George W. Bush appointed John Roberts as chief justice in 2005, it was clear, from Roberts' record, that his voting patterns would be moderately conservative — distinctly more conservative than those of Breyer, but distinctly less so than those of Scalia.

Something similar can be said about every Supreme Court nominee over the last half-century.

At the same time, the court is capable of big surprises. For example, many experts did not predict that in 2015, the court would rule in favor of same-sex marriage; that in 2020, it would strike down President Donald Trump's decision to repeal President Barack Obama's Deferred Action for Childhood Arrivals program, designed to protect some unauthorized aliens from removal; or that, in the same year, it would interpret the Civil Rights Act of 1964 to forbid discrimination on the basis of sexual orientation.

Indeed, a careful effort to test whether legal experts are good at forecasting the decisions of the Supreme Court found that such experts were right just 59.1% of the time.

That's better than flipping a coin — but not a lot better.

Judge Amy Coney Barrett's background shows unmistakably conservative inclinations, and it is a good bet that she would count, broadly speaking, as a conservative justice. But would she vote to strike down the Affordable Care Act, the 2010 health care law widely known as Obamacare? Would she vote to overrule Roe v. Wade, the 1973 ruling that women have a constitutional right to have an abortion? To forbid any and all affirmative action programs? Would she take a strong stand against gun control laws? Would she give more protection to commercial advertising?

On these questions, any answer would be speculative. Even though Barrett is highly likely to show conservative voting patterns, and even if we know the same about the five other Republican appointees, it is hard to be certain how they will vote in individual cases. There are three main reasons.

The first is that even if justices can be characterized as conservative, they might have very different views about their appropriate judicial role and in particular about their obligation to respect the court's own precedents.

For example, Chief Justice Roberts insists on that obligation — far more so than (say) Justice Clarence Thomas, who is not reluctant to overrule precedents if he thinks that they are wrong. (The record of a lower court judge is not very informative, because such judges are bound by Supreme Court rulings. The Supreme Court is empowered to overrule itself.)

Preliminary signs suggest that with respect to precedent, Justice Brett Kavanaugh, one of President Donald Trump's two previous nominees, may be aligned with Roberts, while Justice Neil Gorsuch may take Thomas's approach. On this count, Barrett's inclination remains unclear.

The second reason predictions are hard is that even if a justice's voting patterns are predictably conservative or liberal, his or her views on particular issues, and on particular cases, might turn out to be surprising. For example, conservative icon Scalia liked to invoke the "rule of lenity," which tells courts to rule in favor of criminal defendants when statutes are ambiguous.

It looks as if Gorsuch may have libertarian leanings, helping to explain his decisive vote, in 2020, to protect a Native American against prosecution in state court. "Today we are asked whether the land these treaties promised remains an Indian reservation for purposes of federal criminal law," Gorsuch wrote. "Because Congress has not said otherwise, we hold the government to its word."

Does Barrett have libertarian leanings as well? That, too, is unclear.

Finally, specifics of cases and issues matter, whether they involve law or fact. A justice might show broadly conservative voting patterns, but in a particular case might encounter a convincing argument that the stereotypically conservative position is wrong — whether the issue involves sex equality, presidential power or environmental protection. The logic of the argument, and the particular dispute, might move any justice in what might seem to be an uncharacteristic direction.

In an unusual public statement in 2019, Roberts said, "We do not have Obama judges or Trump judges, Bush judges or Clinton judges." He wasn't entirely right, but he wasn't entirely wrong, either.

Elections have consequences, and presidents almost always get the kind of justices that they want. But it's also important, and perhaps a consolation, to say that on particular occasions, they are going to be surprised — and disappointed.