What do bicycle helmets, moral hazard and regression testing have in common?
Killer head gear?
Common sense would suggest that wearing a helmet would increase the safety for cyclists. The evidence, however, is less clear. A number of studies show no overall benefit from wearing a helmet. At least one study suggests that wearing a helmet is associated with an increase in cyclist fatalities.
How can having protection increase the risk of injury?
Banks wear helmets
No-one has chosen "Fractional-reserve banking" as their specialist subject on mastermind. There are no "shout-outs" to the Financial Services Compensation Scheme on Radio 1. However,the scenes of depositors queuing outside Northern Rock in 2008 illustrated the importance of both and brought home the realities of what happens when banks go bust, or even suspected of going bust.
As Oscar Wilde did not say, one Bank Run is a misfortune; more than one is careless. In fact multiple bank runs (aka systemic banking crisis) is more than careless; it is associated with some of the major economic collapses of the last 100 years, including the great depression in the USA and 1930's Germany.
For this reason we make banks wear helmets.
The helmets provide protection in the event of accident. The Financial Services Compensation Scheme in the UK, and the Federal Deposit Insurance Corporation in the USA provide protection for depositors in the event of bank failures.
The protection does not stop there. We do not want banks to bump their precious heads, so we essentially wrap them in cotton wool and promise not to let them fall.
Didn't this use to be a blog about testing?
We have helmets in the world of testing. We call them regression tests. You can develop (cycle) with gay abandon, safe in the knowledge that if you slip up your regression tests (helmet) will save you from serious injury.
Unfortunately cyclists and bank have unwittingly discovered the downside of helmets; they encourage you to take more risks.
We are less safe, because we think are more safe.
It is a truth not universally acknowledged, but it seems that strapping on a helmet makes us feel invincible, makes us take more risks. We are less safe, because we think are more safe.
That bank that is "too big to fail"... suddenly looks like a bank that is taking too many risks. We are less safe, because we think we are more safe.
Moral Hazard
Moral Hazard is "a situation in which one party gets involved in a risky event knowing that it is protected against the risk and the other party will incur the cost."
Still wondering what this has to do with testing?
If you have a great set of regression tests.... well done you. You have strapped a nice padded helmet to the head of your organisation.
But, as a result, is the organisation now cycling faster? Is it jumping red lights? Does it have the swagger of an invincible bicycle courier?
If so, then you may be less safe, because you think you are more safe.
Regression testing adds safety only if you are diligent in enforcing safe practices throughout the software development cycle. A helmet will not save you in a frontal collision with a truck. You still need to look both ways before you cross.
Regression tests are not a substitute for proper testing. They are what we do after all the other testing has completed.
Personally, I still wear a helmet... despite the studies. But I cycle with the caution of someone not wearing one. I try to test the same way.
I am a little bit safer, because I feel less safe.
-----
For further reading, see "What evidence is there that cycle helmets save lives?", "Published evidence sceptical of helmet effectiveness or promotion", "Moral Hazard" and "Bank runs".