But then, bemusingly, other principles come into play. One, much fretted over after Iraq and the debacle of the WMDs that weren’t there, is the use of anonymous sources. What’s so “ordinary” about spinning cobwebs of tales from “high officials”, “senior intelligence analysts” and the rest of that wretched repertory company? Didn’t The New York Times – among other great American newspaper names – get the facts and the threat of Saddam Hussein’s arsenal lethally wrong because (like some of Bush’s boys) it relied on whispered testimony from Ahmad Chalabi and fellow exiles in his Iraqi National Congress? Maybe, if that source hadn’t been kept from the reader, but subjected to scrutiny and tough questions, the flimsy façade of Iraq’s mass destruction potential would have been exploded long before America’s first cruise missiles hit Baghdad. Maybe – no, absolutely – we should have been told where the NYT was getting its “facts” from.

That lesson, in large part, seems to be learned. The New York Times has cut back heavily on the use of anonymous sources, and it is not alone: USA Today has a most onerous new code restricting anonymity, and so does The Guardian here in Britain. “High officials”, at least in theory and editorial intention, have never had such a thin time. Judith Miller of the NYT, the main link to Chalabi, has surely learned a bitter lesson.

But wait a second, where is Ms Miller? In prison, as it happens, for defending the anonymity of a source of a story she never actually wrote, about the outing of a CIA agent who happened to be married to a former U.S. ambassador who got on the wrong side of the White House over WMDs (this time in Niger, after duff testimony from MI6 rather than Ahmad Chalabi). The printing schedules for the British Journalism Review make it hard to go too far while special prosecutors are still prosecuting away in search of administration leakers, but the overwhelming consensus as I write is that Judith Miller is defending the principle of a journalist’s right to protect her sources, not the reputations or jobs of some of the President’s top spinning and smearing men. You might as well expect the entire Westminster lobby to spend a night in Brixton rather than admit they ever received a phone call from Alastair Campbell.

The larger battle

Even The New York Times itself has some doubts: “To be frank, this is far from an ideal case. We would not have wanted our reporter to give up her liberty over a situation whose details are so complicated and muddy.” Why go to the stake to defend a spin doctor planting a bit of bile? Why commit an act of civil disobedience to defy an officer of the court charged, on behalf of ordinary, law-abiding people, with finding out who put a CIA officer’s life in danger by naming her and, in so doing, clearly broke federal law? We should begin, writes Ignatius, “by agreeing that the reporter-source privilege isn’t absolute – any more than attorney-client privilege or doctor-patient privilege”. The American Bar Association’s own code of ethics recognises, for example, that the confidentiality of conversations between an attorney and client is limited by what’s known as the “crime fraud exception”. The privilege can be breached if the attorney learns his client is planning to commit a crime, or if the attorney is himself participating in a crime or fraud.

See how marshy the firm ground of principle has suddenly become? Thirty-one out of 50 American states have “shield laws” which give journalists’ sources some measure of protection, while federal law raises no shield, but there’s nothing absolute here, nor has there ever been. The special prosecutor in the case pursued two particular reporters and institutions: Matthew Cooper of Time magazine as well as Judith Miller. Time fought the case through the courts to the last minute of the last hearing and then capitulated. There are no opt-outs from the rule of law if you lose rather than win, said its editor-in-chief, Norman Pearlstine. Ordinary people, ordinary Time readers, wanted to see judges and juries obeyed, not flouted. It was, whichever way you cut it, a principled decision from Time. The trouble comes when principles collide.

Such trouble always brings future fears, of course. American journalists, anxious about losing their First Amendment rights, fear a wholesale attack on source confidentiality from a hostile administration. The heroism of Deep Throat becomes an absolute, unquestioned good. The wisdom of NYT defiance is subtly questioned, dividing editor from editor. Isn’t there a little too much playing to the gallery here, unconscious restitution for Chalabi disgrace? Is this the cleverest way to fight a wide-ranging battle? And naturally, since Britain is not a media island, many of those arguments cross the pond.

Could it happen here? Indeed, has it already happened here? The last great British ruckus, after all, was when several UK editors declined to hand over forged (or at least heavily doctored) documents supposedly detailing a prospective takeover bid for South African Breweries by the big Belgian brewer, Interbrew. The forger’s idea was to make a killing. He came with criminal intent and raised keen attention from the Financial Services Agency, the regulator put in place to protect ordinary readers and shareholders. Could the regulator please have those forgeries so he could catch the crook? But no...without a First Amendment, but with considerable resolution, the editors declined to co-operate. The privacy of the source, even this source, was sacrosanct. Perhaps the American Bar Association, scenting a crimefraud exception, wouldn’t have agreed, but the line was held. And British legal experts – such as Martin Soames of DLA Piper, writing for Media Guardian – cite other trenches of defence.

Could it happen here? “Not in the same way.” Journalists are given statutory protection against identifying sources under section 10 of the Contempt of Court Act of 1981. The Act says that a court cannot require someone to disclose a source of information, nor will they be held guilty of contempt of court for refusing to disclose it, unless “the court is satisfied disclosure is necessary in the interests of justice or national security or for the prevention of disorder or crime”. That’s a “useful shield”, says Soames, and so much better than America’s “rather fragile consensus, which is no substitute for legislation”. Like lawyers everywhere, he thinks new law or, at any rate, more law is the answer. But how easily he forgets...

I have a D.Phil in source protection theory and the continuing notoriety that goes with it. Sarah Tisdall, a young Foreign Office clerk, was sent to prison, on my Guardian editorial watch, for leaking details of 1983 cruise missile deployment plans and journalism legend blames me for that (which is not surprising, since I still blame myself ). Legend, at this distance, leaves a few things out, though.

As though by magic

What to do? I asked Geoffrey, and he gave, almost word for word, the answer that section 10 of the Contempt Act provides. The document in question concerned parliamentary tactics; it told how the missiles would go in early one morning to Greenham Common and the arguments that Michael Heseltine, then Defence Minister, would use in the House. It was about Westminster, not national security; it was politics, not disorder or crime.

We had a new shield against onslaughts like this, a law designed to give us succour. The lawyer suggested a legal defence which, to be activated, needed me, on behalf of the company, to pledge not to destroy the document. Sign here please... and I did. Some of those around me thought this was going too far. The unknown source hadn’t trusted us enough to tell us who he or she was. Why try to protect somebody who might already have taken every available means to disguise his or her identity? And some outside newspapers, such as The Sun, were similarly brusque. Too many Lefties scratching each other’s back. Hand it over in the name of GB and Mrs T. But, over long months, we went to law.

And we almost, almost won. In the High Court, Richard Scott (of the Scott Tribunal) ruled for us. Victory! Then, two hours later, the Court of Appeal struck us down. They found that the document itself was harmless and covered by section 10 – but that the sort of unreliable public servant who’d leaked it might leak something more dangerous next time, and so must be exposed forthwith. Now you see peril, now you don’t. Still without a blind idea whether or not capitulating would reveal anyone, I had my longest day. Did I tear the paperwork up willy-nilly and take what came? Gallantry or Quixotic gesture? The devil was that I’d signed on behalf of a company facing brutally escalating fines for non-compliance. I’d given The Guardian’s word, not just my own. Could companies that believed in the rule of law, owning papers that championed it, duck away when the going got tough?

You may remember what happened. The blanked-out hieroglyphics top right, which had seemed like a clue, were meaningless. But Sarah had used a FO copying machine and that helped track her down. Like Clive Ponting and others, she might have pleaded not guilty and been acquitted by 12 sceptical men and true, but she didn’t. I was left looking, and feeling, worse than Norman Pearlstine. And so the legend continues.

Living with it, though, helps square the confidentiality circle and sets all those principles in their place. No difference in law from ordinary people? Absolutely: throw your shields and section 10s and first amendments away. If a source needs protection, if you’ve given your word, then don’t call for a solicitor, because due process will drag you down. Anonymous sources? Use them utterly sparingly, only in extremis, for good cause. Then everyone will know they’re a fight waiting to happen if challenged. The bargain of protection? It’s better made explicitly between source and journalist, because that avoids potentially lethal confusion, and it shouldn’t apply to just any old confidence that’s pushed through the post, nor should it last only up to but not a second beyond when a story is published. Tales in print are the ones that count. Running that document, I thought after the Tisdall affair, was the bargain moment – and not destroying the document at once was the stupidest move. Why are journalists – as even over Interbrew – always so slow to shred evidence? Legal tuttings again.

Very little I’ve concluded here even features in the American debate; nor, as we see from Martin Soames and others, is it quite where Britain’s Sir Echoes find themselves. Everybody wants more protection, not less. Everyone wants better arguments for their friendly barrister. No good; no way. “Good” cases like Woodward and Bernstein come once in a blue moon; bad cases, full of calculation or crookery or ineptitude, are the norm. Don’t expect sanctity because it rarely features. Most of the time, perhaps, the certainty of a messy fight deters pursuers. (Whatever became of the FO chap who leaked those WMD memos to The Sunday Times during the general election?)

Sending journalists to prison, just like beating up on the BBC, brings politicians few boosts in the Gallup polls. Joe and Jane Public may not love our edgy trade, but they love politicians even less. But principle, schminciple, this is pure, desolate pragmatism. To do our job, we need to make simple choices about information and how it comes in – and we need to honour our bargains. Not the First Amendment, but brute common sense: pack away your wigs and take it from one who, miserably, knows.