The thing about news is that it has to be unusual, otherwise it's not news, it's just life. That's one theory at least. But it presents all sorts of problems, says Michael Blastland in his regular column.
Imagine a land where people wear spectacles that turn danger upside down - what's deadly appears trivial, what's unlikely chills the blood.
They lurch into national panic over a mouse dropping - recall Parliament, phone the army! - but regard everyday killers with indifference. It's a place of swivel-eyed disproportion.
But is this fiction, or real? Enter the strange media world of Harrabin's Law, and decide for yourself. The media does sometimes seem to be a place where what you hear most is often what happens least.
I've named the phenomenon after the BBC's Roger Harrabin, who puts it like this: "When considering societal problems over the long term, news-worthiness is often in inverse proportion to frequency. If problems become commonplace, they are not new - so do not qualify as 'news'. This means the media often guides politicians to focus on less serious acute problems at the expense of more serious systemic problems."
Man, dog, biting
To find out if it was true, he counted how often stories about various risks were covered by a sample of media outlets, and compared that with the data. The law seemed to apply.
I'll offer a new example that extends the idea - in a moment.
"So what?" you say. That's what news is - the unusual. As an old journalistic saying has it:
"Dog bites man" - not news.
"Man bites dog" - that's news.
So the more rare an event, the more it stands out. The more it stands out, the more attention we give it.
And that's when we hit trouble. Because the more attention it's given, the more everyone reads and hears about it. The more we hear about it, the more frequent we think it is - and the bigger we think the problem. There are the spectacles.
As Harrabin says: "In areas of public policy where government is asked to intervene there is often a perverse pressure on politicians through the media to act on issues which appear more immediate but are ultimately of lesser public significance."
An example is the comparison between the reporting of road and rail deaths. Rail deaths are far more infrequent and because of that - not in spite of it - they gain more coverage.
Here's a topical test-case for you to judge that stretches the principle a little. Has Harrabin's Law struck again - on strikes? Seems to me there's been an obsession with an imminent "winter of discontent", and indeed spring, summer and autumn of discontent, more or less continuously for the last three years.
The data, however, suggests that this is a period - comparatively speaking - of breathless tranquility, with strike days lost about 50 times higher in 1979 than now and fewer strike days in the past 20 years put together than in 1979, despite 4.5 million more people in the workforce today.
If Harrabin's Law applies, maybe the most interesting and typical thing about workplace relations throughout this recession has been mostly ignored. That is that they seem to have been marked by weary resignation in the workforce or, dare we say it, co-operation during hard times on both sides. Meanwhile effort goes instead to finding proof of that elusive strike-mayhem.
Well, up to now. It's conceivable that we will see more strikes as the cuts bite. Even if the warnings of mass strikes have rung out wrongly for the past three years, it might be true one day. Even a stopped clock eventually tells the correct time.
Unions are far weaker today, but still a force. Pay and conditions are under genuine pressure. And the new boss of Unite, Britain's biggest union, Len McCluskey, has warned of a "wave of industrial action in the spring".
But true in realistic comparison with 1979? The interesting question is whether this is a dog that hasn't yet bitten because people have changed in the UK. That's a question not much asked.
But what about strikes at the BBC? What about British Airways? That's exactly the reaction that bears out Harrabin's Law. We seize on a few prominent examples and assume they equal a trend. There are often problems with data, that's true, but I quite like to know what the trend data is, and on strikes it's not much reported.
There have to be limits to Harrabin's Law. If strikes never happened at all, it would be surprising if that led to an increase in media comment on winters of discontent. But short of that, what do you think? Harrabin's Law? Or am I exaggerating?