Avoid lawyers, architects, project managers and NATM, says Manuel Melis, head of the Madrid metro.
After building a major underground system in record time, he is entitled to his opinion.
As the first passengers stepped onto the new trains on the recently expanded Madrid metro in 1999 they could hear further works going on behind screens at central stations.
South of the city there are now construction sites dotted throughout smaller towns.
After completing a stunning 56km length of new working metro in just four years, 38 km of it underground, the city is battling to repeat the achievement.
By 2003 it hopes to have another 54.7km of mainly underground line in full operation; again a four year programme.
As the small construction team likes to boast quietly, from engineering pride rather than braggadocio, the original line extensions came in at £21.3M/km, or £26M for the underground line. There were 37 stations on this first extension.
This compares to £109.7M for each of 18km of the Athens metro or £109M for the shorter Paris Meteor line. London's, admittedly difficult, 16km Jubilee Line extension cost £264M/km.
'And they took longer, ' points out Manuel Melis, the civil engineering professor who has been the driving force for the civil engineering in the expansion programme. He is the official president of Metro de Madrid and director general of the transport infrastructure division at Communidad de Madrid, the regional government for the area in and around the capital.
The hands-on Melis puts safety and caution right at the top of his list for achieving speed of construction and low cost.
What counts most, he says, is keeping the work moving and avoiding disasters. 'Delay and collapses are what cost money.
Clients will pay almost anything to get stalled work going again.
And contractors are good at exploiting that. Claims mount up.'
Perhaps he concedes, Madrid has some natural advantages.
Athens, for example, had special problems because tunnelling experience was limited and the archaeology a nightmare. Compared to London and Paris, Spanish contract bids are cheaper. Planning issues can be dealt with more quickly.
And in Madrid, routes pass through less congested areas.
Much of the second phase, for example, comprises a new circle line linking five developing dormitory towns south of the capital. There is undeveloped land between them and also along part of the extension to Line Ten which will connect the 40km loop into the main city system (see box).
The streets above this 'MetroSur' are wide and relatively traffic free, and the buildings are mainly apartment blocks of four or five stories. There are no major river crossings.
But that is not the whole story.
Madrid's work has been complex too, with the first phase, creating two new lines and extending others, located mainly in the city centre. The second stage also includes city centre work, threading an extension to the new airport line right into the business district, where an intown check-in terminal is being built.
The soft ground of the Madrid basin is no easier than that of other cities, being mainly alluvial sands and clays. It is similar in places to London Clay, says Melis, though with more sand lenses and water filled pockets.
And with the decision to keep the lines shallow at around 16m, to make the system easier and quicker to use, there are increased risks of settlement damage. Melis says he is thankful for the 'marvellous' tool of compensation grouting.
But despite such tools he moves with caution. He is vigorously opposed to large scale open face tunnelling in soft ground. 'I would absolutely forbid these methods whether it is NATM, pre-cutting or face stabilisation' he declares. 'I would not authorise it even in soft rock.'
He believes NATM is not only unproven but has such a history of collapses and failures that to use it for anything other than harder rock is 'criminal', particularly 'as you can put lives at risk'. Pre-cutting and face support systems like glassfibre rods are too time consuming and expensive, he believes.
'Show me figures on rates of progress and cost that prove otherwise and I will consider them.'
But even then he says the science of soil mechanics is still too undeveloped to guarantee safety. 'A huge face 10m high or more could have major variations within it and you have to test all over. It's time consuming and expensive and still you might miss something.'
He believes the only solution is to use the now well developed EPB tunnel boring machine, 'allowing you to keep a robust steel wall against the face at all times.'
Where the machine cannot be used or is not available you should keep open face to a minimum, he says. At Madrid a variant of the Belgian method is being used, with timber shoring and boards supporting initially all but a 3m 2excavation opening.
The top heading is widened cautiously, before concreting an arch and then working down.
The bench comes out some metres back from the face for stability and with an initial centre excavation before taking the two sides.
'The Phoenicians developed this 3,000 years ago and you can achieve 2.5m a day with it, ' he says. He agrees that where the use of pneumatic hand held tools is now facing restrictions, this method may be more problematic.