When an army is repulsed with bloody losses, it has two choices. It can run away or try again. If it tries again, it had better pray that it understands the lesson of the previous disaster.

So should republicans. Their grim challenge today is to understand the real reasons why last year's referendum was lost, and plan accordingly.

The problem is that every republican has their own pet theory why the referendum failed and, surprise, those theories happen to vindicate the position of their author.

So what is needed now is not hype or spin. There must be a brutally dispassionate analysis of why one of the world's great independent democracies voted against the notion of having its own head of state.

In reality, this is not a particularly complex post-mortem, and the daggers in the corpse of the republic are fairly clearly labelled.

First and foremost, in Australia only bipartisan referendums succeed. The 1999 referendum was passionately opposed by the better part of Australia's conservative forces. Fade to black.

Second, referendums that are not supported by prime ministers have nasty accidents – such as ballot questions being framed in a way that puts the matter at hand in the worst possible light.

Third, whatever their party politics, the Australian people are constitutional conservatives. They know their system is in no danger of collapse, and are sceptical of the need for change. If there is any possible reason to vote no, they will.

Worse, the bigger the change proposed, the more nervy this dubious electorate will be. Even with the unadventurous model of 1999, every one of its 67 alterations was another reason to kill it.

Finally, republican politics is dirty politics. We now know that at any referendum, there is no argument too spurious, no tactic too disreputable that opponents of a republic will not use them. We also know that they are hideously effective.

What these republican realities clearly reveal is the identity of the only sort of republic that conceivably might succeed at some future referendum. The Realistic Republic will be bipartisan, have the support of the PM of the day, and be simple, safe and as little open to misrepresentation as possible.

Critically, despite the post-referendum crowing of Phil Cleary and Ted Mack, it is almost inconceivable that the Realistic Republic will be wearing the colours of direct election. This is because any direct election republic must run foul of almost all the hard-earned lessons of 1999, but most of all will breach the requirement for bipartisan support.

The vital element of republican bipartisanship is conservative backing. Given that Labor is intrinsically republican, the real challenge for any republican model always will be to garner the substantial conservative support required for referendum victory.

But the grim reality is that conservatives will never support direct election in any numbers. It simply is too radical. This is something that more venturesome republicans find hard to accept. They are convinced that if only they "explain it properly", the conservative republicans will come around. They will not.

What this means is that a direct election poll would be the referendum from hell. It would be opposed not only by Kerry Jones and her monarchist glee club, as well as outright constitutional conservatives such as John Howard and Nick Minchin, but by the full array of constitutionally conservative republicans.

This category obviously would include the Costellos, McGarvies (and Cravens) of the world. But numerous senior members of the Australian Republican Movement and the Australian Labor Party – such as Janet Holmes a Court and NSW Premier Bob Carr – also have expressed their profound disdain for direct election.

This formidable force could exact a terrible price for any direct-election republic to conform to other republican realities. Most notably, even the most limited form of direct election must involve a degree of change to the Constitution that would make 1999 look like the removal of a misplaced comma. Every change, every implication, every half-foreseen consequence would be a thorn in the heart of direct election.

So if direct election is a box canyon, in which direction should the republican movement go? This is a difficult question to answer, but the starting point must be adherence to the rules of republican reality. We need a proposal that is simple, safe and, above all, capable of attracting bipartisan support.

Paradoxically, that probably leads us back to the convention model that lost last year. Loser that it was, it is hard to see a model that comes closer to meeting the criteria for success. What we need to do is look at that model closely, with a view to improving its saleability. Can the appointment process include a more popular element? Can the mechanism for the dismissal of the head of state be improved?

Remember, the model managed about 45 per cent in the face of concerted prime ministerial opposition. But PMs, like seasons, pass. What would it or some variant achieve with the determined support of the leaders of both political parties, a scenario that is perfectly conceivable in the mid-term future? Unlike direct election, it might well win.

Greg Craven is dean of law at the University of Notre Dame and a member of Conservatives for an Australian Head of State. This article is based on his speech to a seminar organised by the Australian Association of Constitutional Law on Planning for a New Republic and Notre Dame University.