Why you should NEVER ask for metrics ?
At the risk of sounding pretentious I can say, very humbly, that if you are being asked , by powers that be, to generate a set of metrics ... for any project ... please don't jump at it with headfirst.
I have been living in the world of metrics for more than a decade now and every time I hear some one proposing -- this or that -- metric should be captured, I feel a bit .. well !! I don't know how to word that ...
Let me share a little story which is related to this ...
Long time ago ... there lived a good man in a far remote village in the mid lands. One day he started praying for the lord God to fulfill his wish... and he prayed and prayed for so long with so much dedication that one day, the God relented and traveled from heavens to the earth.
The God spoke to the good man "my dear child !! I am pleased with your prayer and I am here to grant you one wish. Speak !"
The good man bows to the God and says " Dear God .. please grow hair on the ear lobes of my mother-in-law ... "
The God gets puzzled! but he grants the wish any way. Before disappearing, the God asks the good man... "my child ! I am curious to know why you prayed for so long and so hard to ask for such a wish !! "
The good man bows again ... speaks very obediently, "God !! long time ago I read a book that said -- the son-in-law will become very lucky if he gets a mother-in-law with hair on ear lobes. I always wanted to be lucky and so I asked for that boon !! "
And then the God shakes his head,smiles kindly... and speaks "If you wanted to be lucky ... all you needed to do was to ask me for that!! " and the God disappears.
The moral of the story is : What you think you are asking for may not be the right solution to what you want to achieve!
Asking for metrics is just like that !! A metric is a solution that answers a question. Instead of asking the right question, many people ask for a specific solution like a metric with the implicit assumption that they know what they are asking for is the right solution.
Imagine going to a doctor and asking for acetylsalicylic acid instead of telling him that you have an ear pain and you need a medicine to fix it !!
Comments
It's a lightening post of various measures around quality. It's dangerous to measure an activity with out clear idea on what we are trying to achieve. However, we can't live with out any measures in the software world.
I am curious to know on how do you release any product with out any metrics around the same.
Venkat.
The good man didn't ask the god for data about luck. He didn't say "God tell me how hairy my mother-in-law's ears are."
The problem in the story is that the good man made solving the metric the goal, as opposed to solving the actual problem. However this doesn't make metrics, or collecting metrics bad.
I agree with you. All I am saying is that you would be better of articulating the goals to be met before releasing the product instead of masking a metric as a goal. As Deming says, making numbers as goals [ example ... a goal of having less than 10% of regression bugs before beta 3 and trying to track it as Bo Bayles said above ] is what I am advocating against.
Too often we ask (not just metrics) but anybody for specifics without any context.
We should ask for help in solving problems, not for specific solutions to problems.
-d