What has happened to employment? The official figures seem clear. In the three months to October, it fell by 63,000 and unemployment rose by 128,000.

This does NOT mean that 63,000 people lost their jobs in the latest three months or that 128,000 became unemployed. These refer to net changes, which are much smaller than gross ones. The ONS's experimental data show that, in Q3, 903,000 people left employment: 513,000 entered economic inactivity (say because they retired) and 390,000 entered unemployment.

Even this net change, though, is only an estimate. It's drawn from a survey with large sampling error. The ONS says that the 95% confidence interval around its estimated change in employment is 134,000. This means there's a 95% chance that the net change in employment was somewhere between minus 197,000 and plus 71,000.

This, though, is not the only evidence we have about jobs. There's also a measure based on a survey of employers, rather than individuals. This shows that jobs increased by a net 150,000 between June and September; there are about 30 reasons why the two measures diverge.

And there is sampling error around this as well - though the ONS, with its infinite capacity to infuriate the hell out of us, says there is 'no on-going measure' of how big this is.

The picture here, then, is murky. The worst case reading of the numbers is that the labour market is doing terribly; 197,000 is almost 0.7% of all employment. The best case is that jobs are (net) being created. Taken on their own - which they are not and should not be - today's numbers are consistent either with the view that macroeconomic policy is far too tight and is causing a disastrous fall in employment, or with the view that it is about right and in consistent with rising employment.

This confirms what all economic forecasters have known along. Not only do we not know where we're going, we don't even know where we are.

In one sense, this uncertainty doesn't matter as much as you might think. Lots of good policies do not depend upon noisy quarterly moves in jobs. The merits or not of tax and welfare policies, supply-side changes or active labour market policies can be considered independently of these uncertain data.

In another sense, though, this uncertainty does matter. Because the media almost never discuss it - see for example these otherwise reasonable write-ups of the figures - they give the impression that hard knowledge of the economy is possible when in fact it is not. They mistake precision for accuracy. And this serves a nasty ideological function. It helps to increase the credibility of those politicians, technocrats and cranks who claim to possess precise knowledge of the macroeconomic present and future when such knowledge is in fact elusive.