The Belfast Agreement was, and remains, an achievement of such a magnitude that it gives politics a good name. None of its opponents, either on the extreme fringes of republicanism or on the fundamentalist wing of unionism, has ever articulated even a vaguely credible alternative to the way it opens up possibilities for each side without closing them down for the other.
But the agreement gave certain hostages to fortune. What's happening at the moment is that fortune is demanding a heavy ransom before it releases them.
The first hostage was that, for the process to be possible at all, Sinn Fein had to be encouraged to place a rhetorical distance between itself and the IRA, even though no one really believes that the two organisations are other than inextricably connected.
There was a belief - ultimately justified by the course of events - that if Sinn Fein was allowed to pretend that it was a normal political party, it would begin to behave like one. Slowly and carefully, Gerry Adams, Martin McGuinness and their colleagues had to be moved out of the category of untouchable terrorists - who, by definition, could not be involved in democratic negotiations - and into the category of recognised political representatives.
That strategy was both necessary and successful. But what was good for negotiating a peace deal was bad for negotiating the decommissioning of IRA weapons. It allowed Sinn Fein to present itself as merely a political party like any other, representing its electorate and entitled to share in power purely on that basis. It could insist that weapons were a matter for the IRA, a faceless organisation out there somewhere beyond the reach of the political process.
Sinn Fein could take all the gains of the peace process - seats on the Executive, the release of IRA prisoners, the radical reconstruction of the RUC, a huge boost to its electoral prospects in the Republic - but insist that the bill, in terms of decommissioning, should be sent to someone else.
And that someone else - the IRA - could at the same time deny that it had any bill to pay. So, when the Mitchell Principles for decommissioning were laid down in 1997, the IRA could tell An Phoblacht they didn't really matter because "the IRA is not a participant in these talks".
Sinn Fein could make deals on the basis of democracy and the removal of the threat of violence, but they committed the IRA to nothing. And in spite of all the momentous events of the last two years, that basic position has not changed. It continues to echo through all the identical on-message interviews with Sinn Fein leaders this week.
Ultimately, though, unionists (and indeed everyone else with a stake in the process) were never going to buy the rhetoric. At some stage Sinn Fein would have to be forced to accept that, for the purposes of the peace process, it has to answer for the IRA. The ambiguity would have to stop.
The problem, though, is that the Good Friday deal was all about the creative use of ambiguity. It was achieved by stretching words as far as they could possibly go, constructing sentences that seem, at one and the same time, to underwrite the partition of Ireland and the union of Great Britain and Northern Ireland and to accept the legitimacy of all lawful attempts to destroy them.
Key political concepts like sovereignty and nationality have been stretched so that the competing claims of the British and Irish traditions can be accommodated in a fluid and open-ended arrangement. Instead of seeking precise definitions of the nature of the territory, the agreement enshrines the radical principle that whatever can be agreed by a sufficient consensus of both traditions will happen. Nothing has a fixed or permanent meaning. Everything is up for grabs, so long as nothing is grabbed by force.
But the one thing that doesn't yield to such rich ambiguity is weaponry.
The reason why decommissioning has always been so difficult to incorporate into the peace process is that it belongs to a completely different kind of language. If the creative politics of the agreement as a whole is poetry, decommissioning is cold, hard prose.
The questions that come into play when decommissioning is being discussed - when? how much? - are completely different to the questions about the future of Northern Ireland that the agreement tries to answer.
That is why decommissioning had to be left aside on Good Friday 1998. It is also why it has proved so difficult to resolve since then. The parties, especially Sinn Fein, were being asked, almost literally, to speak a different language to the one they had grown used to using in the rest of the process.
As for the IRA, one of the problems with a conflict that goes on as long as the one in Northern Ireland is that people start to believe their own propaganda. This has happened to the IRA and it is one of the deepest sources of the decommissioning problem.
Republicans have an image of their role in the conflict that is utterly at variance with the reality. At the heart of the present crisis is the IRA's determination to define the last 30 years as a war of liberation. To put their weapons "beyond use" would be to face up to the fact that the use of those weapons has been an unmitigated disaster.
Pretty well everyone can admit that the Troubles in general have been an appalling affliction. With 3,600 gruesome deaths in a place with a million and a half people, this is undeniable. Consensus on this basic fact is reflected in the Belfast Agreement and all that led to it.
But disagreement about what explains the horror lies behind the current crisis. For what you think about decommissioning is really a function of what you think about the nature of the conflict. It depends on how you answer an apparently simple question: on whom did the IRA make war?
The IRA's answer seems simple enough: we made war on the British army, the UDR, the RUC and the loyalists in order to defend the nationalist people against aggression. Our weapons were turned on other people with weapons. Since those other people say they will not be using arms, it is safe to assume that we will not be doing so either. We suffered just as much as anyone else in the Troubles, but you don't hear our political representatives threatening to bring the house down unless the Brits or the loyalists get rid of arms.
This answer reflects the IRA's self-image and its preferred version of what has gone on since 1969. There are probably many members of Sinn Fein and the IRA who believe it. To sustain a long and sordid conflict, you have to censor, not just opposing views, but your own knowledge of the truth. But if the war is really over, the IRA ought to be able to look back with some degree of honesty. If it does so it will be forced to face certain crucial facts.
First, the IRA has been much more than merely one source of violence among many. It has dominated the killing game. Over the course of the Troubles the IRA was directly responsible for 1,684 deaths - well over half of all the killings. And it was not a defender of the Catholic population. Not including members of the British army or those killed in Britain or on the Continent - whose religious affiliations are generally not recorded - it killed 745 Protestants and 381 Catholics. For both communities, in other words, the IRA has been a very significant aggressor.
Second, it is not true that the IRA fought a war only against other official and unofficial armies. Republican paramilitaries killed 713 innocent civilians - the vast majority of whom were victims of the Provos. The IRA killed 73 children under the age of 18. It killed building workers on their way home, shoppers having a cup of tea, women collecting census forms, young couples having a jar in a pub in Birmingham, people honouring the dead of the two world wars, mothers looking for a tasty bit of cod in the local fish shop.
Third, the IRA itself did not carry the brunt of the suffering. The IRA's war was unusual in that, for all its talk of "blood sacrifice", it took far fewer casualties than it inflicted. Normally in guerrilla campaigns (like the IRA's campaign against the British in the War of Independence, in which it suffered about 700 deaths compared to army and police losses of about 500), the guerillas suffer most.
The Provos, by contrast, killed at least five people for every death they suffered, causing over 1,600 deaths and suffering 355. For every Bobby Sands and Mairead Farrell there were 10 forgotten, anonymous people who don't get their names on the gable walls. They didn't volunteer for death, they were volunteered. They didn't sacrifice their lives for the cause. Their lives were sacrificed to the cause.
And, of course, many of the republican dead themselves were the victims of internal feuds or of "own goals". The largest number of republican paramilitaries killed in the conflict were murdered, not by the RUC or the British army, or the loyalist terror gangs, but by their own comrades.
The INLA and the IRA have been responsible for the deaths of 164 of their own members. The British army, RUC, UDR and loyalist paramilitaries killed 161. Even in a largely Catholic area like west Belfast, more republican paramilitaries were killed by their own side (42) than by the British army, UDR, RUC and loyalists (41) put together.
Together, these facts have a huge bearing on what happens to IRA arms. For they remind us that the IRA's weapons have not been a conventional military threat.
They have been used against a very large range of people: the armed forces of the British state and republican militants; loyalist mass murderers and grannies out doing their shopping; Protestants and Catholics; politicians and babies; alleged drug dealers and members of the Garda Siochana. And it was the IRA, with its elastic and infinitely expandable definition of "legitimate targets", that chose to present itself as a threat to anyone and everyone.
Arms used in that way can't be regarded as the mere tools of a conflict, which can be allowed to rust away when the conflict is over. They are a source of fear for everyone and anyone that some faceless militarist chooses to define as an enemy.
In the face of that fear, ambiguity loses its usefulness and clarity becomes a matter of life and death.