. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

I COLLECT YOUR ISSUES

LIKE A MAGAZINE

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .


Thursday, July 16, 2009

More Reasons I Hate the Business Whirled

One of the projects that I'm working on right now is to help the company develop age curves and gender curves. Basically what that means is we want to take a male at birth and a female and birth and see how much they cost to insure for each year of their life. We decided to do this by looking at the premiums that different sexed age groups would have to pay in 6 * 2 * 3 * 3 types of plans. And there were 27 age groups. And I needed to find the cost for different parts of the plan, as well. And initially, I had to do it individually, setting up a plan type, and then going through each of the 27 age groups. So altogether I would be filling out 6 * 2 * 3 * 3 * 27 * 2 cells, which is 5832 cells total. That's a long time. I mean, even if you could do six every minute (which you couldn't), that's still 100 hours roughly. Which would cost them $2100 in labor costs to me alone.


So I wrote a program to do it for me, and it basically reduced the number of cells that I had to work on to 120. Because I'm awesome. Which meant that I could do it in approximately 2% of the time that they had initially planned to set aside for me. One fiftieth. And of course, I finished it way ahead of time, with way less hours than expected.

So I got to the meeting on Wednesday or whatever, which had been severely pushed back, and I bring out the age curves I'd found. All 5832 cells were filled out, which was a pretty impressive feat in and of itself.

Didn't mention it. Instead they look at one of the headings that my team leader had come up with. "Poor plan vs. Rich plan? No no, I think... no, you mean lean plan rather than rich plan, right? Like, one with reduced benefits and such? Yeah, typically it's rich vs. lean in a situation like that." As if it meant anything at all what we called the plans. We could have called them a happyface plan and a sadface plan. Or X plan and Y plan. Or squiggly plan and squirrely plan. It didn't make any difference at all. It's not even like we were showing this to a client who would have to understand precisely what we were talking about.

And then she asked about the standard plan. "Why isn't the plan definition on this page?"

"Hm?"

"Why isn't the plan definition printed out here?"

"What? I mean, I... on everyone else's I'm sure I wrote it in on the bottom - it should be written in on yours, too. It is, isn't it? [I look; it is written in] Yeah, okay. I mean, that was just how it printed. I mean, it just could not fit one more line on the printer, so I wrote it in."

"Yeah well, you should have come to me to see how to print it."

No, I shouldn't have. She never has time for anything, first of all, and she wouldn't have time for this. Not to mention that it doesn't make a bit of difference, since the plan definition was there for her to see.

Not one mention of the fact that I completed everything in 1/50th the time they expected.

Just formatting issues.

Later we were talking about filling in the gaps between the years. Of course we knew how much it cost to insure someone exactly at the boundry of the age group - we had some data points - but what about inbetween those data points? She suggested doing linear interpolation. I couldn't believe what I just heard. Quietly,

"Linear interpolation?"

She looked at me. "We could do cubic splines, whatever."

"Why stop there? You could get a 10th degree polynomial..." She had already moved on.

So she really really cared about getting lean plan correct instead of poor plan, and making sure it was printed on the page rather than written in, but once it came to actual math - you know, getting something perfectly accurate to the 10th derivative rather than, oh, maybe getting the 1st derivative right - no no, don't worry about being exact. It's no big. Not like that's actually going to make any difference, or make our work less accurate, or potentially cost our clients hundreds or thousands of dollars. Just draw a straight line, like a four-year-old. It's not even like polynomial interpolation is hard.

//

"So do you know what you're doing next summer, err... you think you'll be coming back here?"

No.

"I don't know, I was thinking about maybe trying plain old insurance..."

//

Peace out.

No comments: