There is no such thing as zero risk. Accidents happen, however unlikely the chain of circumstances.
What really matters is the magnitude of the risk, and who decides if that magnitude is acceptable or not.
The disaster at Great Heck last week raises several vital questions which need speedy answers. Ten people died after a train crashed into a Range Rover on the tracks and then into another train.
However, there is absolutely no point in asking how we prevent such a tragedy occurring ever again. To reduce what statistics already show is a very low risk by a factor of 10 or more (see News) would involve building thousands of kilometres of high walls alongside rail lines, installing high barriers for hundreds of metres each side of every rail overbridge, and eliminating all pedestrian crossings.
And this would only save maybe a handful of lives a year.
One of the many unusual features of the Great Heck disaster is that the consequences were so much less than could have been expected. Ten dead are 10 too many, but it could have been many more.
Sunday's bridge collapse in Portugal, which killed at least 70 people, shows what can happen in a worse case scenario.
Had fate been kinder there might have been only minimal traffic on the crossing when scour finally undermined a central pier. Or, as at Inverness in 1989, the fatal loss of support might have occurred slowly enough to give plenty of warning, and no lives would have been lost at all.
So, in the aftermath of both tragedies, the most important question is: was there anything within reason that could have been done, which was for some reason not done? In the case of the Entre os Rios bridge it seems that the authorities had, sensibly, tried to put an end to the sort of uncontrolled sand and gravel extraction which can contribute significantly to scour problems.
A responsible authority would also have carried out underwater inspections of the pier foundations and, if necessary, dumped rock protection around them.
It remains to be seen if aggregate extraction was a factor in the collapse, if any inspections took place, or if any rock was dumped. The spate of political resignations suggests its own conclusion.
Investigations into the Great Heck crash are still at an early stage. The preliminary Health & Safety Executive report revealed little that was not already known. However, the same question still has to be answered.
Should the Highways Agency have done more? Were there any cost effective measures that could have been taken, given the obvious potential consequences of a vehicle crashing on to a high speed rail line?
At first sight, one could point out that it would have cost very little to extend the safety barrier much further from the bridge itself. But how far is far enough?
After all, if the throttle on a Range Rover jams wide open, and the driver doesn't switch off the engine, it could travel a very long way off road. And extending the barrier is no guarantee that a vehicle and trailer would never somersault over it a few metres from the bridge and crash on to the line below.
But these are unlikely events.
In the final analysis the level of protection at each crossing or intersection has to be a matter of judgment. And the statistics confirm that over the years the judgment of the highway engineers in this country has been almost flawless.