Leveson, uncontroversially, found the Press Complaints Commission (PCC) unfit for purpose. But instead of devising a different model, his recommendation was for yet another quango of the reasonably great and moderately good, without press representatives (so it would be completely independent of the industry) but with statutory backing, which would ensure it had teeth for biting rather than gnashing (up to £1 million in fines—this is not "light touch" regulation). This estimable body, with a resplendent chairperson, would adjudicate complaints over breach of the code of conduct, and provide a mediation service for potential defamation claims. But the press, with prime ministerial support, will go to almost any lengths to avoid any kind of statutory "underpinning". They want newspapers bound by contract rather than statute, submitting to fines of up to £1 million and publishing adjudications on their ethics imposed by a lay body chaired by the likes of an ex-chief justice.

The mistake that is made in these models is that they pivot upon decisions about journalistic ethics and accuracy made by worthies who may have no background in journalism, or ability at fact-finding. They will be part-time, taking decisions on arguments presented by modestly-paid staff, rather than after hearing evidence or making their own investigations. In so far as any form of regulation "works", it has been best attempted—in several countries—by a hands-on press ombudsman. This person is usually a distinguished lawyer or journalist (or both) who runs a full-time office giving most media victims what they want—a quick and efficient declaration of demonstrable inaccuracy, a right of reply to personal vilification, and a mediation service to which libel plaintiffs must resort before they can take their case to court. There are no "million pound fines"—these would probably be struck down by the European Court of Human Rights as disproportionate—but the ombudsman model offers a speedy remedy for untruths or mistakes and an effective alternative to strung-out and costly libel actions.

The simplest solution-and the one that would appear most advantageous for press and public alike-would be to slip into the Defamation Bill a provision for an ombudsman, who would have two functions. Every defamation action would be referred to the office to attempt mediation, and the ombudsman could, if the partries agreed, arbitrate the dispute, with compensation capped at, say, £20,000. A newspaper could always refuse and take its chance in the libel courts, but lengthy and bitter actions would become much less common. Secondly, the ombudsman would be empowered to receive complaints about inaccuracies (many, if not most, inaccuracies are not libellous), to fact-find and to issue decisions, which an erring newspaper would be bound to carry with directed prominence. The error would be exposed in a short and publishable decision. There would be no directed apologies—here, Leveson went wrong in principle, because sincere apologies must by definition be voluntary. (As Richard Ingrams replied, when asked if Private Eye had ever knowingly published falsehoods, "Yes, the apologies.") The ombudsman, as a statutory office, would be paid for by the government. I would also entertain complaints about serious falsehoods on credible websites like Wikileaks, which would not be bound in any way to participate, or to publish the ombudsman's finding. But they would be foolish to ignore it, as the blogosphere would not.

Newspapers could then get on with establishing their Voluntary Standards Regulator, which would confine itself to ethical lapses and empower itself through contracts with major newspapers. It would react to systemic failures in the industry, and could valuably provide ethics training courses for journalists. Its reports and adjudications on standards might be published by contractual direction, but newspapers should be fully entitled to dispute them. The press must be careful of what it pretends to wish for: the problem with self-regulation is that it produces self-censorship.