GIGO / by Katherine Hajer

This blog post is going to seem like it's all about math, but really it's about writing and editing. More to the point, it's about some of the stupid mind tricks we can pull on ourselves when we're planning things like writing and editing, and trying to get a schedule together.

Last Friday I wrote a story about a near-future corporation who replaces their office workers with robots. They then discover that they didn't do a very good job of projecting some of the consequences. True to Friday Flash, the comments were the best part, including a chilling real-life example which Sonya Clark provided.

The effects of outsourcing and automation were certainly a big part of the story, but today I wanted to delve more into the math behind a line I gave to one of the executives:

"We bought these robots expecting 24/7 productivity out of them, or one robot for every 3 FTEs, but we're only seeing about 23 hours of work for every 24-hour cycle. That's a 4-hour lag 3 weeks into the launch."

The executive is claiming that a human worker puts in 8 hours a day, that each robot does the work of 3 humans, and, therefore, unless a robot is working 24/7 like she assumed it would (because she forgot to calculate in maintenance time), the company will be less productive and losing money.

Here's a quick spreadsheet I made to show how that assumption works out versus the actuals:


Even leaving out human-worker variables like overtime, vacation, and sick days, the executive's math is off. She assumes humans work 7 days a week, for one thing. Okay, a lot of us do, but it's not assumed to be the norm when figuring out FTEs. Nor did she include vacations, which should have been a no-brainer. Probably if she'd worked out the numbers on a per annum basis instead of per week she would have noticed something was off.

Instead of getting 3 FTEs from a robot, the company is actually getting 4.29 FTEs. They are ahead in terms of productivity, not behind, even with that one hour of maintenance mode per day. Even if you factor in vacation time for 3 FTEs, you still don't lose a whole FTE's worth of hours over the year. There are still too many robots to replace the humans, too much productivity for which there is as yet no measurable demand. Yet perception is reality, and the executives believe productivity goals won't be reached because of the maintenance hour.

But that's the thing about spreadsheets, or any other "what if?" math. Humans tend to simplify the variables as much as they can to eliminate the fuzzy, unmeasured parts of a problem, and in turn come up with off-base predictions.

While I was writing this blog, I Googled the term "how efficient are office workers?". I was trying to find some quick and dirty numbers on how much time the typical office worker gets to, you know, do work, instead of handling interruptions or creating their own. (Yeah, I know, "quick and dirty" numbers. I've been trying to write this post for four days and instead been spending it on overtime. Please understand.)

Check out the results. Loads of tips on how to become "more" efficient, sure. How to measure how efficient you are right now, or how much more efficient you've become after following those fabulous tips? Not so much.

So: writing. Next time you tell yourself you're going to get more done by getting up half an hour earlier, or staying up later, or writing with a fountain pen on paper, or whatever the heck scheme you come up with.... get some numbers. Find out how much different practices improve your game. Okay, don't get to the point where you're spending all your time measuring yourself and no time writing, but get something together.

You may well be surprised what you find out.