Tuesday, September 7, 2010

Summarizing with Relative Measures

"It spewed a lot of data."
"I found many bugs."
"That took a long time."

I know I've said every single one of these things. Most testers, developers, support engineers, product managers, and pretty much everyone else involved in creating software has said something similar more than once. It's a common way to summarize something.

So what's the problem?

Every time I've said this (or heard someone say this), the first question is identical:
"How much is a lot?"
"How many bugs?"
"How long, specifically?"

It's about as common as saying, "How are you?" followed by "I'm well. How are you?".

The problem is that our measure is relative. "A lot" to you might be 100, but to someone else, it might be "10 million" before it's a lot. "A long time" to that kernel developer might be 10 milliseconds, but to the web developer, 10 milliseconds is just about nothing. When we use these relative measures, our audience needs to calibrate those measures to their own relative measures. So they ask for the real number, pull out their internal yardstick, and say, "yup, that's a lot" or "wow, that's not a lot".

Don't be afraid of relative measures; they're just fine to use. Just be sure that when you're providing a summary with a relative measure that you have the actual number handy - odds are you'll need it to make sure your relative is the same as your audience's relative.

No comments:

Post a Comment