Unavoidable failure

I recently wrote that one of the lazy assumptions of the political-media class  is the belief in perfectibility and inability to see that failure is very common. Since writing that, we've seen two examples of what I mean: reports that the cost of two new aircraft carriers is double its initial estimate; and the PAC's claim that the implementation of the new universal credit has been "extraordinarily poor".

Such episodes confirm my prior, that mismanagement of big projects should be regarded as the norm and success as the exception. This is simply because such things are more complex than any single individual can possibly manage; bounded knowledge isn't an individual weakness to be deprecated but rather an ineliminable fact about the human condition.

Our inability to see this leads to the planning fallacy. Imagine that, for a project to come in on time and on budget, 50 separate elements must come together, and imagine that each has a probability of success of 98%. Basic maths then tells us that the probability of the project succeeding is only 36.4%. Sadly, though, many politicians don't know basic maths.

You might think that this just shows us how inefficient the public sector is - which is only to be expected given that incentives are weak and departments are run by the sort of people who give mediocrity a bad name.

I fear, though, that this is too glib.As Paul Ormerod has said (pdf), "failure is pervasive" in the private sector too, and for the same reason - that managers' ability to know enough is severely limited (pdf). For example, around 10% of firms cease trading every year, which means that the chances of long-term survival are very slim.

And bosses are like politicians in at least two respects; they too misunderstand the basic maths which warns that planning is failure-prone. And they too are selected to be overconfident about their chances of success.

Perhaps, then, politicians, journalists and the media share a common ideology - a presumption that top-down management can competently oversee complex events to a greater extent than is the case. This presumption leads then to fail to see that what makes the private sector successful - insofar as it is - is not so much the quality of management as the existence of markets. The general public might have anti-market attitudes - but so too does the ruling elite.

Comments

[FromArseToElbow]

The planning fallacy (or optimism bias) is real, but so too is the bad news bias - i.e. failures get more publicity than successes. In fairness, failure is usually a lot more entertaining (except for the participants). Schadenfreude at hubris confounded never goes out of fashion.

In fact, large-scale projects can be planned successfully (i.e. approximately on-time and budget and delivering the promised benefits), using standard deterministic and probabilistic techniques (in a nutshell: plan from worst-case backwards). The common causes of project failure are well known. Their persistence is a feature, not a bug - i.e. a reflection of political constraints, vested interests (rent), or a cultural overhead ("the cost of doing business").

Common-sense tells us that major projects do succeed without pulling the world down around our ears. GDP growth can be interpreted as success exceeding failure in aggregate (and as failures often "cost" a mutliple of original estimates, there must be a lot of successes).

Pessimistic planning tends to be eschewed by organisations wedded to the heroic leader fallacy, where simply describing success is deemed to be 90% of implementation (Hitler in his bunker and IDS have a lot in common on this score). This is near mandatory for projects that are the result of political initiatives.

[Jim]

"what makes the private sector successful - insofar as it is - is not so much the quality of management as the existence of markets"

There is more joy in heaven over one sinner who repenteth.........hallelujah!

[rogerh]

Relying on chance to ensure project elements come together is known as the 'poke and hope' method - completely hopeless. Relying on chance or rigid specifications alone to ensure the elements come together is a recipe for failure. Good project work is a constant fight against chaos.

Error is common enough and some error is unavoidable but timely feedback systems will, with luck and skill, keep the project on track. The word being timely.

I generally avoided government work, the worst problem was that the requirements were usually internally inconsistent. In the private sector you point this out and get a straight(ish) and honest(ish) answer to fix the issue - not so on a government job. So government work attracts a special breed who are well used to the process and have billing systems to match. Failure is less common in the private sector IMHO, the targets are clearer.

[Phil Beesley]

The two aircraft carrier builds provided opportunity for an expensive but illuminating experiment. One carrier could have been built according to the original project outline, with appropriate technical mitigations and corrections, but without change of scope or fundamental design. The other could have been built using the project management which was adopted. At the end of it, we could see whether either meets the original or current functional spec.

[mutant_dog]

See the lead book review from the Economist magazine this week, subtitled "why a strategy is not a plan". I believe it is apposite.

http://www.economist.com/news/books-and-arts/21588834-strategies-too-often-fail-because-more-expected-them-they-can-deliver-why

[Guano]

I have done evaluations of a number of World Bank projects in Africa over the last 20 years. They have had a common failing. After the usual pre-project planning, a large contract is signed with a European or North American engineering contractor. After a few weeks or months, it is discovered that the initial conditions are different from those assumed in the initial planning (eg soil conditions are different, the new project is being linked into an old system that is in a worse state than was assumed). So the project has to be redesigned after a big contract has been signed (maybe 50 million dollars), which means de-scoping or robbing of other budget lines to pay for the cost over-run. Often the budget lines that lose out are small, self-contained add-ons that were running well.

I have always asked the World Bank: why can't you start with a smaller contract that will tease out the unknowns before committing to a big contract. The answer is always that international contractors will not sign up for contracts overseas unless they are in the 50 million dollar range. Local contractors might, but the World Bank insists that projects have to be open to international contractors.

[Metatone]

As FATE notes, there are plenty of large scale successes, we just hear less about them.

A useful division can be found in Dave Snowden's work between complex and complicated. Complicated can be project managed to success using traditional techniques. Complex is whole different world.

(Chris, your arguments in this area would be a lot more impressive if you had the hang of some of these distinctions.)