New year, same question for Londoners: when will the Jubilee Line Extension be finished? 'Our target is still the end of September this year to get trains running,' confirms project director Hugh Doherty confidently.
But just as confidently, it should be noted, as he last year assured that the extension would be finished on 31 March 1998.
All right, it is a cheap shot. The reality is that all track-laying in the tunnels and stations is now complete and - strikes permitting - contractors are rushing to complete installation of linear services: the signalling, optical fibre backbone, tunnel telephones and lighting. All this, says Doherty, should be complete early this year.
Test running between Stratford and North Greenwich should then start at the beginning of April, extending soon after into London Bridge and down to the crossover beyond Waterloo. Trial operations are scheduled from summer through to the end of September 1998.
So the JLE team will be just six months late delivering arguably one of the most demanding pieces of infrastructure that Europe has seen this decade.
But, (and there is, of course, always a but) as most Londoners are now aware, the JLE will not actually be complete in September 1998. And the service which does start this September will be poorer than on the existing Jubilee Line. There will be fewer trains, it will stop at 10pm, will not run at weekends and will not even link directly into the existing tube network. The only assurance from London Under- ground chairman Peter Ford is that the line will be fully operational in time for the millennium celebrations - at the end of 1999.
Focus is now on problems with the signalling system. JLE aims eventually to boast one of the most sophisticated software driven signalling systems in the world - a 'moving block' system - but it is not yet remotely close to being ready.
The original concept was to deliver 36 trains per hour compared to the usual maximum of around 25 on a conventional 'fixed block' system. The grand plan is then to introduce the new signalling system, funds permitting, right across the network.
So does it matter if it is a bit late?
'Of course it matters,' says Doherty. 'We said we would open on March 1998 and maybe we were over ambitious. Certainly the technical challenges have been greater than we had envisaged, but yes it matters. It's a matter of pride that we said we would open in March 1998.'
That said, Doherty is proud of what has been achieved in the last four years - huge problems overcome, no deaths and a quality of construction that most would agree is first class. 'It is regrettable that we haven't hit the March 1998 date,' he admits. 'But I won't walk away from here bitterly disappointed.'
And while frustrated that the software is not yet ready, Doherty defends this plan. 'We are not saying let's develop a super-duper signalling technology for the hell of it. The present underground needs this form of technology if we are going to increase capacity within the present infrastructure.'
So there is the dilemma. The underground wants and needs a signalling system that gives a high capacity. Old tunnels with small diameters mean you cannot pack more people into bigger trains, so running more trains is the only realistic answer until cash is available to build more or widen existing tunnels.
Doherty's problem is that Westinghouse, the company creating the software, is struggling to make the new moving block system work. It will definitely not be ready by September 1998, and it may not even be ready by September 1999.
Sadly this is all very familiar in the construction of modern railway infrastructure. The Channel Tunnel construction programme was brought to its knees by electrical and mechanical installations; Docklands Light Railway ridiculed through years of signal and control failures. Civil engineering wonders turn rapidly, in the public's eyes, to blunders when targets are not met.
But apportioning blame is not Doherty's style, at least not publicly. He is a problem solver. 'I could quite easily turn around and be critical of [the contractor] but why, when they are endeavouring to deal with quite difficult problems? My focus is to work with them.' It's a railway project not a civil engineering project, he booms. It is a team effort, though he accepts that commercially and contractually there will be issues that will have to be resolved. 'Hopefully in private,' he adds.
He is quick to point out exactly what Westinghouse is trying to create and why it is taking so long to get it right. 'It is a safety critical system,' he explains. 'It needs to work without fail, time and time again.'
But for Doherty's team, not having this new system has meant a total rethink on how to handle the line's opening, and forced time-consuming and expensive contingency plans so JLE does not miss the millennium.
It is also the reason why the service starting in September 1998 will not link into the existing Jubilee Line at Green Park, despite all the stations, bar Westminster, being ready.
In fact, things appear worse when you consider that when the opening was pushed back by six months last year, Westinghouse's problems had not then been uncovered. But rather than plumping for a conventional fixed block system, JLE has stuck to its task. It is now installing 90% of the hardware necessary to run the moving block signalling but will then modify it to run in a fixed block mode, a temporary but safe system that gets the trains moving.
But this thrown-together signalling will not be as efficient as a devoted fixed block system, and with a maximum capacity of 17 trains an hour will substantially underperform the 24 trains per hour on the existing Green Park to Stanmore Jubilee Line.
So running a through service, explains Doherty, could disrupt the services on the existing line. And with a high degree of reliability and user satisfaction on the existing line, LU was clearly keen to wait for Westinghouse.
But just how long, no-one will say.
Regardless of his confidence in Westinghouse, Doherty is no fool. Rather than risk the unmissable millennium deadline he now has engineers working on the modified system to squeeze out extra capacity and get it up to the magic 24 trains an hour to match the existing line.
'If we do that before September 1999 - which we believe we shall - we will then operate through running. So it could be early 1999,' he says.
And having spent four years struggling to construct Westminster station largely in the four hour shut-down period in the middle of the night, he is keen to take every opportunity to maximise time for modification work to be completed after September 1998. With LU estimating that 80% of the JLE's demand would be satisfied by a weekday business hours service, Doherty describes squashing the remaining work around a full service as 'crazy'.
'Of course we are heaped with abuse for doing that,' he accepts. But there is a precedent in his defence - the Victoria Line. LU's last major new line opened in 1969, nearly two years ahead of the link to Brixton. And Pimlico Station didn't open until 1972. But the public was none too impressed that time either.
Doherty has no desire to follow this example. 'More time means more money, so we must drive to limit the overrun in time,' he says. 'But you spend only what is reasonable. You don't make September 1998 something that you die for and spend inordinate amounts of money to achieve it.'
All this leaves one outstanding question. Just how much will JLE cost? Present out-turn figures put the bill at £2.76bn, and if you juggle with indexation long enough this is around £650M more than first thought.
While a large proportion of this can be put down to civil construction difficulties - and 'time will tell' says Doherty on what claims may yet follow - some must also be attributed to LU's need for a solution to its current overcapacity problems; namely a state of the art signalling system.
JLE simply had the misfortune or fortune, depending on how you look at it, of pioneering this effort.