Nanny State and Error Bars
X is 0.5 with an error margin of +-0.1
You can make a law that says it is illegal for X to be anything other than 0.5, you will prosecute a lot of people as a result. You may even pat yourself on the back, Nanny state protecting people from rogue Xs! Hurrah!
But there are error bars on that of +- 0.1, we're just not sure if X should be anything from 0.4 to 0.6. Someone closer to measuring each X may know better than us. We're working with the average of X's, not a particular X with a particular error value.
So at worst the law should be (X>0.4 AND X<0.6)
Worse still an error bar is itself an average, in extreme cases X may even be 0.7.
Prison sentences for people with rogue X's have a negative penalty too. Making a crime is not a neutral thing with no consequences!
So there is a downside to society of X being 0.7, lets called that downside Y.
And there is a downside to making it a crime for X to be greater or equal to 0.7, you criminalize people who don't need to be criminals, lets call that downside Z.
You make a law that X must be less than 0.7, you've made Britain (Y-Z) better, and you know that X=0.7 is a bad thing, so Hurrah, Nanny State has made a difference!. But if Y is less than Z you've made Britain worse not better.
So sure X being 0.7 is bad, but if Y<Z it's not as bad as criminalizing it. At some value of X there comes a net benefit. Perhaps X<0.85, or X<0.9
My point here, is a lot of the newer Blairist laws ignore the error bars, they treat the law as having no negative consequence (blindly ignoring prisons full of people and ever increasing violence as officers try to enforce ever more petty laws). So the laws typically define strong penalties at sharp cutoffs and the cutoff point is well within the error bounds where it encroaches on signal noise. Margins of error that are best defined by the individual in the individual circumstance are defined by a minister with only a bullet point powerpoint slide as guidance.
Then there's the rounding error. I set the motorway speed limit at 70mph, 71mph will be prosecuted with every increasing penalties until it stops. So I'm driving home to give my sick mother her medicine and I'm just too late and she's dead. The probability of this happening is 1e-8. i.e. 0.000000001, as near as damn it nothing. There may be a billion of these obscure scenarios,but a billion nothings is nothing so we can ignore it right? Hubble space telescope had millions of tiny rounding errors in the conversion between millimetres and inches and as a result it was myopic. You can't ignore the rounding error on millions of tiny scenarios because those errors add up. You have to estimate it.
I'd make this point again (quantizing one of the negative effects of strict enforcement of motorway speeding limits):