A Bug's Life

The long walk in the link is quite a decent analogy. The biggest problem with it is that the terrain is very much harder, and the real route longer, than the original estimate was based on. If the journey was on straight roads, the estimate would have been reasonably valid.

So how do you learn the terrain? Mostly by traversing it, and getting a feel for how hard it really is. But when you're estimating software development, it's always for a new route, because you've already written the code for the routes you've done. Except you didn't do it perfectly; this is where the analogy breaks down, because there are always bugs in the code you've already created.

Fixing those is maintenance, rather than development. The trick is to make things better as you re-work the routes, building chunks of paved road to fix bugs, rather than just chipping a step into a rock that's precariously balanced in position. To be able to do that, you need maintenance to be an accepted part of the way the business works, and a worthwhile activity, rather than "something that shouldn't be needed", to be bodged hastily. Making that happen is a problem in structuring the business, not a problem with estimating.

Enjoyed that, and passed it on. Thanks.

Oh, yeah, so true. And if I triple or quadruple my time estimates to try and give people a more realistic projection, they accuse me of padding things on the grounds that it can't possibly take that much time.

On a recent project, the actual thing requested took me less than an hour to run, but it took three sixteen-hour days of preparation (plus a 20-hour "overnight" unattended run for the software to crank through about 20 million combinations of things to find the 60,000 or so combinations that I needed) to get it set up. My managers see the run time of the final product but don't understand the grinding tedium of the set-up.

Not just software - many major civil engineering projects in the last few decades have been late and over budget (e.g. the Humber Bridge, Channel Tunnel etc.) In some cases this is indeed because new ground is being covered, but in others I suspect that cannot be the whole story, so I think it's a lot more complicated than the analogy given there implies. The Cambridge Guided Bus looks like it ought to be an object lesson in how not to run a project, although we don't have access to all the details at the moment as everyone is still suing each other.