Japan has never had a fatal high-speed train crash. So what's gone wrong here?

'We're making the railways safer. But we've higher expectations of safety than we had a generation ago'

It is too early to make judgements on the cause of the Hatfield tragedy, or indeed to try and attribute the blame. For the people killed, injured or bereaved by this, or any of the other rail accidents of recent years, there can only be anger and grief. And for those of us who use the railways frequently there is that awful "what if?" feeling when the train thumps over some bit of particularly uneven track. But judgement and blame - these are surely for later.

It is too early to make judgements on the cause of the Hatfield tragedy, or indeed to try and attribute the blame. For the people killed, injured or bereaved by this, or any of the other rail accidents of recent years, there can only be anger and grief. And for those of us who use the railways frequently there is that awful "what if?" feeling when the train thumps over some bit of particularly uneven track. But judgement and blame - these are surely for later.

It is not, however, too early to start to think through the context of this accident, to wonder what we are doing wrong, and to lay the ground for perhaps doing rather better.

Safety in all walks of life is a complex mix of human, technical and practical considerations. Some technologies are inherently dangerous and have to be made safe by the complex building-up - over many years - of a mixture of rules, technology and culture. Air travel is a good example. What was once a truly dangerous method of travel has been transformed into, on some measures, the safest, by the meticulous application of aircraft technology, navigation and other systems, and rigorous training.

Other technologies are inherently dangerous, and remain so, but we accept an element of danger because of the other benefits they bring. I suppose road transport is the best example here. With Sweden, the UK has the best record for road safety in the world but we still have an average of 10 deaths a day, and many more injuries, on our roads. Each death is a tragedy - yet we put up with this situation. We do not insist, for example, that every car driver has an annual health check-up as we do with airline pilots. We are making the roads safer, but not as quickly as we might.

And railways? Well, railways are indeed safe. They are inherently safe, despite the mass and stresses involved, because it is possible to control most of the sources of danger. In that sense they differ from aircraft, which are still at risk from weather, and from road traffic, where the variables are too complex to be controlled. Add to that the fact that the world has had 170 years of knowledge on how to make trains safer: better signalling, track improvements and the generation of a culture of safety.

Recently here in the UK we are making the railways safer still. In the last five years of the 1970s, an average of 25 people died each year. In the past five years, it has been an average of 11 people. But that is far too high. It is too high because we have higher expectations of safety than we did a generation ago, partly perhaps because we require higher standards from privatised companies than we expected from the old, nationalised British Rail. It is also too high because we are more aware of world standards, where Britain is in the bottom half of the European league table, though not at the bottom. All Europe, it should be noted, is far below Japan in its rail safety. Accidents on Japanese local lines are extremely rare and no one has ever died as a result of an accident on a Japanese high-speed train.

So we have to do better; and we have to do better against the background of a shrill and politicised climate of debate. When there is a tragedy in the air - the Concorde crash is a recent example - it is met with an orderly, painstaking inquiry, carried out by professionals seeking the truth. No one blamed the fact that Air France was in the process of being privatised, or accused it of "putting profits before safety". The same sort of professional, orderly process ought now to take place, looking not just at this particular rail accident, but how to learn the bigger lessons from all accidents in recent years.

It would be presumptuous to try and do more than give a loose sketch of how this process might run, but here are five suggestions.

First, we have to look at the structure of the railways. It has been widely argued that the privatisation was botched, and it is probably not ideal. But a sub-optimal structure need not necessarily damage safety. Airlines are haphazardly owned, with a complex series of alliances, sub-contracts, leases and so on. Yet they deliver safety. In Japan, the railways are owned by a mixture of private and publicly-owned companies, with the nationalised industry, JNR, being privatised. But this hodge-podge works.

I suspect that the structure is not actually the core of the problem and that the disruption of trying to change it would do more than harm than good. But it should certainly be considered.

Second, we need to look at other countries. Japan manages its safety record by a mixture of investment and culture. As anyone who has travelled much on Japanese railways will testify, there is an extraordinary sense of order about the passengers as well as the staff. Each reinforces the other: you do not sit in the wrong seat, even if the carriage is half empty.

Third, we need to look at other industries. Standards of customer care may be rising, but remain below that of the airlines; I suspect that standards of safety are similarly lower, though they need to be assessed and measured in an orderly way.

Fourth, the monitoring of safety should, as far as practicable, be separated from the operating companies and from Railtrack. Monitoring safety should not be a top-down, officious bureaucratic process, but it does need to be independent, both of the rail industry and of politicians.

Finally, government should have as little to do with this process as possible. Improving safety anywhere is mostly a matter of extreme attention to detail. Politicians are dreadful at detail: the disorder of their lives, the fact that they are rarely in the same job for more than a year or two, the fact that, in general, they have little experience of the commercial world, the long time-lags between decision and result - all this makes them uniquely unsuited to giving practical help in lifting rail safety.

Yes, government is ultimately responsible. Yes, it can be an influence for good. But it can also make enormous mistakes. The hard work of improving safety standards will be done by orderly, committed individuals who know something about railways. It is profoundly in the self-interest of the industry that they are allowed to do so, and hence to rebuild public trust. The victims of this and other disasters deserve no less.