There are very different purposes to GCSEs and driving tests

Policy's all back to front in certain areas: where things should be relative, they're absolute; and where they should be absolute, they're relative.

Consider two cases: GCSEs and alcohol limits for driving.

What do we really care about in student GCSE results?  If what were most relevant were whether students had passed some competence threshold, or if exams were mainly a matter of assessing the progress of teaching techniques over decades, then we might want an absolute/ threshold standard.  The consequence would be that if teaching techniques progressed over time, the proportion of students passing exams and getting higher grades would increase.  But of course that not what GCSE results are about.  The vast, vast majority of people taking some account of an individual's GCSE results couldn't give a fig how her results compared with those of the average student twenty years ago.  And they certainly have zero interest in what her GCSE results tell us about her competence at, say, Chemistry or French.  No-one is going to hire you for anything on the basis that your GCSE chemistry means you have gained some knowledge crucial to doing that job.

No.  What we care about in GCSE results is how well students do relative to one another.  So for politicians and teachers to proclaim that more students are getting higher grades because of all their hard work and because teaching is so much better today totally misses the point.  People aren't doing GCSEs so we can assess their teachers.  They study subjects so as to acquire various forms of knowledge and to improve various other things about themselves, but they take the actual GCSE exams solely and only so we can test how good they are relative to one another.  So if you devise a marking scheme that assesses them by an absolute standard instead of a relative one and far too many people get 'A' grades for us to work out who's any good and far too few fail for us to work out who isn't useless, then you've horlicksed the whole thing up and need to start again.

By contrast, when someone does a driving test, I'm interested in whether she's competent to drive.  I want to know whether she's going to crash into me and cause me to die.  So her competence really matters.  I couldn't give a fig how enormously much better she is than the average one of her peers or how enormously much better she is at driving today than she is normally is she is, nonetheless, not competent.  Likewise, when we consider whether the car is roadworthy I don't care how much less unroadworthy it is than similar cars of its age.  And if I'm thinking about things you shouldn't be doing when driving, what I fundamentally care about with regards to them is whether they threaten your competence to drive - not whether they make you better or worse at it.

We have no obligation to be at our peak in order to drive a car.  If I have a cold, or my children are in the back, or I am singing along to the radio, or I am distracted with worry about getting a new job, or there is a particularly dishy man sitting next to me, I may not be at my absolute peak driving skill.  But provided that none of those things threatens my competence to drive, I should be allowed on the road.  (You doubt this?  Consider Stirling Moss.  After his famous accident he wasn't as good at driving as before - after all, that was why he quite racing.  Do you think that means he should have been banned thereafter from driving a car on the road?)

But when we hear that there are plans under consideration to reduce the alcohol limit such that even a pint of beer or a couple of glasses of wine would put you over, what do you suppose is the reason?  Is it that studies now suggest that a significant number of people would not be competent to drive after one pint?  No!  It is that studies now suggest that many people's ability to drive is measurably impaired by even one pint.  But that totally misses the point - the role of an alcohol limit is not to attempt to force everyone to be at their tip-top best for driving; it is that consumption of a certain amount of alcohol renders a significant number of people incompetent to drive under fairly common road conditions when they are otherwise fairly normally hale and hearty.

So when we should have an absolute standard - driving competence after drinking alcohol - we have a relative standard (relative to one's own normal competence); and when we should have a relative standard - official school exam results - we have an absolute standard.

As I say: all back to front.