At the police station, the man claimed, Ninness had pulled him towards a locker room and said: "We'll see how f---ing tough you are now.'' He had been punched, had struggled with and fallen over, and eventually had been pulled away from Ninness by others, in the course of which he was kicked by others.

The man, who had a criminal record, defended the charges vigorously over three days, but the magistrate preferred the police version that he and his mates had brought on the confrontation. The man appealed, but also lodged, through senior counsel, a private prosecution of Ninness for assault. In 1983, again after a lengthy trial, Ninness was convicted of assault and sentenced to three months' jail, suspended on his entering a good behaviour bond. The Federal Police, conspicuously, did nothing about Ninness' status within the AFP.

In due course the man's appeal came up before the ACT Chief Justice Sir Richard Blackburn. An even larger array of police witnesses were presented, but the judge found more than enough to justify a reasonable doubt about the police version of what occurred, and allowed the appeals.

Next – by now we are into 1986 – came the appeal lodged by Ninness, mounted on his behalf by Alec Shand, QC, from Sydney, paid for by the police association. The case, before Justice John Gallop, lasted 20 days and there were umpteen witnesses, not least glowing character references from some of Ninness' seniors – all old ACT Police colleagues such as Colin Winchester and John Johnson. The AFP successfully argued a claim of public interest privilege to withhold details of the internal investigation and treatment of a number of complaints of assault against Ninness over the years, all of which had been dismissed.

At the end of the day, Gallop could not decide who he believed. The two versions – from witnesses for the man and from the legion of police – were irreconcilable, and "someone is not telling the truth''. "Yet there was nothing in their respective demeanours in the witness box which enabled me to reach a definite conclusion one way or another about who was telling the truth ... I am left in some doubt concerning the commission of the offence by the defendant. I am not satisfied beyond reasonable doubt.''

This was exactly the same test that the Martin inquiry into the conviction of David Eastman for the murder of Colin Winchester applied to its recommendation that Eastman be freed. The judge made it perfectly clear that he was not convinced of Eastman's guilt beyond a reasonable doubt. He said Eastman should be pardoned.

The judge's findings do not depend only on his view that there was "a nagging doubt'' about Eastman's guilt. It also followed a positive finding that Eastman had been denied a fair trial, not least because Eastman, and his legal team, had not been told of a good deal of the evidence in the possession of police and prosecutors that could have helped him mount a better defence. One part of this, the judge suggested, involved material suggesting that the murder could have been done by someone else.

In the British common law system of justice, the onus is on the prosecution to prove guilt "beyond a reasonable doubt". If a judge, or a juror, thinks an accused person "probably did it'' but is not convinced beyond a reasonable doubt, the proper verdict is "not guilty".

I can think of any number of defendants, some of whom are or were in the Federal Parliament, who were probably guilty of charges made against them, but who had acquittals listed against their name, as Ric Ninness did, because the court was not satisfied beyond a reasonable doubt. It is rare indeed for a completely innocent person to be before the higher courts – whether for murder, or any other crime, if there was no evidence suggesting that they are probably guilty. Indeed, the guidelines of the Director of Public Prosecutions insist that no one should be charged on indictment before the courts unless the prosecution believes that a reasonably instructed jury would find the accused guilty.

At the end of the initial inquest into the death of Colin Winchester, the view of John Dee, from the DPP's office, who had been counsel assisting the inquest, was that the evidence available did not justify the prosecution of David Eastman, or anyone else. When, after the inquest was reconvened, the primary additional evidence available to the Coroner was evidence coming from a metallurgist working with the Victorian Police forensic laboratories, who said he had demonstrated that residue found in the boot of Eastman's car was the same as gunshot residue found at the murder scene. At the time, Dee, who had had some experience with overconfident assertion by this witness, formally warned the AFP, at the most senior levels, that there were a lot of doubts about the witness and his tendency to move from being the detached and independent expert to being the advocate for one side or another.

We now know that the warnings, which were effectively ignored, were justified. If the witness did the tests he claimed, he failed to properly document or record them, and, in some cases, it appears impossible that he could have found what he says he did. In any event, many of his experiments destroyed the samples being used, so were not reproducible by outsiders. Nor was there any process of proper peer or external review of what he was doing. There was also evidence of the mixing up of exhibits, including ones from different crime scenes.

One simply cannot say that the rejection of such evidence, or that doing so left a gaping hole in the prosecution case, was some sort of technicality. It was a fundamental flaw, one of which would have been clear to a more competent, and perhaps less zealous, investigation. Nor can one suggest that Eastman should be free today because of something only just found out. The risks and gaps were there before the prosecution, and the failure to see their significance, or to act on them, might be described as a matter of the forensic choice of the prosecution. If the consequence is that a possibly guilty man is walking free, that is not a consequence of his cleverness, or some quirk in the system, but of serious deficiency in the investigation, in the prosecution, in the conduct of the trial, in the way the appeals system so often fails, and in the failure of police to maintain an open mind, even after a verdict.

Over the past few days, the now-retired Ric Ninness has publicly affirmed Eastman's guilt and warned that he might well murder again. Another detective, John Lawler, whose work came under considerable scrutiny in the inquiry, seemed to be suggesting that the inquiry was a joke, or that one should substitute his complete conviction of Eastman's guilt. Lawler is, like nearly everyone else, now retired but a man who did so as head of the National Crime Commission, and was the man who promised much and delivered little about the "most shameful day in Australian sport''.

Eastman, whom the system failed to prove guilty, is entitled to the same presumption of innocence as Ric Ninness. It is better, as they say, that guilty men walk free than that innocent people rot in prison. It is a matter that many ought to ponder that our system will let a man whose guilt has not been proved rot in jail for 20 years before a proper re-examination of the case. For that, perhaps, we are all a little guilty. A worry too that there appears already to be the gathering of some sort of lynch party to override the verdict, or to make a reintegration of Eastman into society impossible. That's a risk some politicians are already pandering to by suggestions that a pardon will not be contemplated.