As a Marxist, I'm supposed to be a woolly-brained utopian whilst centrists are hard-headed realists. However, three things this morning make me suspect the opposite is the case: the Today programme's discussion of the disturbance at Oakwood prison; reports that more people die of heart attacks in the UK than Sweden; and news that NHS waiting time data are unreliable.

These stories seem to be framed as if they were deplorable departures from a norm of perfect organization. My reaction, however, is different. I remember Adam Smith's line - that there is a great deal of ruin in a nation. Any large organization will naturally contain some failures. There are, at least, two reasons for this.

One is the existence of trade-offs. Maybe prisons could prevent riots by being better staffed. But this would be expensive, and would trigger reports of "scandal of idle prison officers." Likewise, it would be expensive for the NHS to emulate best practice around the world, and efforts to do so would lead to headlines: "fury at NHS bosses free holidays". If you want lowish-cost public services, you must accept some imperfections.

One overlooked trade-off here is between efficiency improvements and consistency. Improvements - in hospitals and schools as in business - require experiments, to discover what works and what doesn't. But this would lead to complaints either about "postcode lotteries" (where successful methods aren't immediately available nationally) or about failing schools, where experiments don't work. You can avoid these problems by avoiding experiments - but the resulting consistency will be a mediocre one.

A second problem is bounded knowledge; management simply cannot know everything. It cannot predict where there'll be riots, as these are a classic case of emergent behaviour, and it is vulnerable to underlings manipulating data.

Sometimes, these two problems interact. For example, social workers cannot tell with 100% accuracy which children are in grave danger and which aren't. They must therefore choose between two errors: leaving children with parents where they might be mistreated; or breaking up families unnecessarily. Whichever they do risks the anger of the media mob - even though occasional errors are inevitable.

I stress that these failings are entirely compatible with hierarchic organization working in many/most cases. My point is merely that the best organizations are inevitably inherently imperfect and prone to error. We should not pretend - as the media (and bosses!) do - that management can be perfect and so failures could be eliminated if only people were smart enough; the outcome and hindsight biases, of course, contribute to this myth of perfectibility.

This myth, though, has pernicious effects. It encourages the belief that there can be a few heroic "leaders" who can achieve such perfectibility and who therefore deserve multi-million pound salaries and disproportionate esteem. If instead we saw small-scale failures as inevitable, we might be less inclined to pay big money for a job that cannot be done.

In saying all this I'm not making a Marxian point: my influences here are Smith, Berlin, Hayek and Oakeshott. It's insufficiently realized that scepticism about the powers of management isn't a Marxian view but rather a properly liberal-conservative one.