Schedule estimates, take 1

[wshaffer]

At work, we're planning a big project. I handed a coworker a batch of specs and asked him to estimate how long it would take a writer to document the features described. I made my own very crude calculation, trying to pad the numbers a bit to compensate for my known tendency towards optimism.

I came up with an estimate of x writer-weeks. My coworker came up with an estimate of 3x writer-weeks. Head, I'd like you to meet Desk. I know you'll get along famously.

I'm tempted to split the difference and turn in an estimate of 2x writer-weeks. But what I probably need to do is go back and re-examine the assumptions I used in my back-of-the-envelope calculation.

Comments

[kate-schaefer.livejournal.com]

3X is a good estimate. A study I read years back found that no matter what the field of endeavor, whatever people were trying to do took, on average, 2.7 times whatever they thought it would do.

With that piece of useful info, I would make my estimates, then multiply times 3. Perfect estimates, except that whatever I was trying to do still took 2.7 times the original estimate times 3...

I no longer estimate anything. It just takes as much time as it takes, dammit.

[wshaffer.livejournal.com]

The real kicker here is that we're not really trying to figure out how long the project is going to take. We have a deadline. What we're really trying to figure out is how many writers need to be assigned to the project in order for us to meet the deadline with a high-quality end product without anyone dropping dead of overwork.

We all know the folly of assuming that a piece of work that takes one writer 6 months will take 2 writers 3 months. But the person-week is the only established metric we've got.