There is an old saying in the building trade “Measure twice, cut once”. This I have found is very useful advice and my son kept repeating this to me when we were installing his kitchen, laminate flooring and bedroom furniture.
Recently we had some blinds fitted to our front room bay window – the salesman came and measured with what looked like a very expensive telescopic measuring rod (I’ve found one for £150 on e-bay) – much more accurate than my 3m metal tape measure bought from Woolworths many years ago.
He went away and returned a couple of weeks later to fit the blinds. There was only one problem – the main blind did not fit the window.
The problem wasn’t with the measuring it was the fact that he swapped the height and width measurements when entering the values into his tablet.
This is what we call in the Market Research industry a data-entry error.
So how do we get around these problems?
We run checks on our data as follows:
- Check the logic of the data and cross-check responses at different questions
- Double enter the data and cross check if necessary – done by two different members of staff
- Report out the outliners and manually check those that do not fit in
If the blind company had checked they would have seen that we had two windows of different heights – unlikely in a bay window and maybe worth double checking before manufacturing the blinds.
The salesman is fined by the company for making mistakes – he had commissioned a bespoke blind for a window that didn’t exist, so that was an expensive day out for him.
The same theory of “Measure twice, cut once” can be applied to programming. If you have a complicated definition that you need to use more than once, then set it into a variable – that way if you are asked to update the definition you only need to change one line of code. It may take a little bit longer to setup, but it could save problems in the future.
I recently ran a complicated calculation to get at an average figure and in order to check the output I also ran a check analysis a different way to see if it came out with the same answer.
The two runs didn’t agree and it took me about 2 hours to sort it out – it turns out that the problem was with the check analysis rather than the original calculation. I still believe it was worth doing and is similar to the “Cut once” theory.
When I Googled “Measure twice” some websites say it originated in Russia and over there they say “Measure seven times, cut once” – Seven times – sounds like something Snow White would do.