Doesn't justice recognise this by accepting that liability is on a sliding scale?

The law must reach some binary decision - guilty or innocent. The accidents are varied in their degree of blameworthiness. Hence there will always be borderline cases - decisions that could go either way.

In practice, maybe the law even errs on the side of letting drivers off?

Between 2010 and 2014, there were 3,069 crashes with pedestrians in the Twin Cities and its suburbs. 95 were killed. 28 drivers were charged. But many of the deaths weren't even judged worth a traffic ticket.

http://www.startribune.com/in-crashes-that-kill-pedestrians-the-majority-of-drivers-don-t-face-charges/380345481/

So the general moral take-home would be that society expects us to be in "reasonable" control of our actions. And we know that "reasonable" is then tough to define as there is always so much more we can do to prevent accidents or slip-ups. So at a social level, some kind of trade-off between the effort required and the potential damage that might be caused, has to be agreed.

There is a social norm when it comes to a duty of care, whether it be driving your car, doing heart surgery, or carrying a cup of coffee across your living room. Then justice is about making some black and white judgement on an individual instance. It has to be because there also needs to be a specific action that follows. You can't half lock a bad driver up.

The justice system of course has appeal courts and community service penalties, etc. But the principle would be that there has to be some social-level norm as a generality. And then to particularise this generality - apply it to some individual case - a line has to be drawn across the world. On one side is social responses - the system of penalties or sanctions that can be imposed. On the other is your personal response - your freedom to think about what you just did in anyway you like. You might well have rather a strong response to killing a pedestrian even if it was judged "a complete accident".

So essentially this is the rather abstract Enlightenment view of humanity in action. Social norms are encoded in laws. They are treated by society as the statement of absolute constraints. Then by the same token, what is not forbidden becomes your personal freedoms. They are also just as absolute. The Enlightenment machinery would also recognise some basic freedoms as rights. This goes further in saying society can't write laws that impinge on these freedoms.

Being human becomes a highly abstract affair on both levels. Actual humans are taken out of the equation as much as possible so that we become creatures of an abstract system.

Of course societies don't apply this model with complete rigour. Social networks and community standards mean who you know, what power and status you have, can affect outcomes. People get away with what others will let them get away with.

Abstract justice systems operate in a real human world. So a kind of meta-judgement needs to be made. Given the trade-off issue - the effort to enforce a strict rule of law vs the cost of that effort - it could be that a society is doing a "reasonable" job in being pragmatically relaxed. Or it could in fact be just socially corrupt.

You were asking more about the notion of personal responsibility. Getting back to that, my reply is that there is good reason for the judgements of justice to be binary - the Enlightenment model wants to draw a clear line between society's constraints and your freedoms. The norms it encodes in law then do recognise a sliding scale in terms of just how much effort we ought to have to put into regulating our own behaviour.

But deciding how general laws apply in a particular case is always going to encounter complex borderline instances. And a judgement still has to be made - either in favour of social sanction or individual freedom. However the effort and cost of applying "blind justice" is a meta-consideration for society. Is perfection ever itself a reasonable aim?