E135: Why Big Game Dev Plans Never Succeed

I was once on a project with a couple of hundred people.

They were split into several orgs, each of which had a bunch of teams inside of them.

And I was on one of those teams and helping coordinate some of the between-team and between-org things.

And I asked the leader above me, what chance do you want my team to have of succeeding as part of this larger effort? They said 80%.

And I said, well, wait, what do you want the overall project chance of success to be with all the teams and all the orgs? And he said, 80%.

And I thought, oh, we are not taking math seriously as we're doing our percentages.

Because the reality is, if the overall project needed to be 80% likely to succeed, then The right way to get there is not to average all of the individual team's efforts.

And if they're all at 80%, it means we have an 80% chance of success.

In fact, we have much worse chance of success.

So I walked him through the math of it.

As he realized what was happening, he said, oh my goodness, I hadn't thought of that.

I hadn't looked at it that way.

I didn't understand the way that that worked.

And it helped him go do some things to hopefully have a higher chance of success.

But the reality is, If all those teams inside of that giant project had an 80% chance of success, the overall project probably had a chance of success well under 5%.

I've seen other organizations do this.

They tell everyone to be 80% confident and assume that that will mean that the overall project averages to be 80% confident.

but you don't average your chance of success.

You multiply it against itself.

What ends up happening is you have a set of teams that are all 80% confident on their own, and then the total confidence is quite a bit less.

This is one of the reasons projects often fail their deadlines.

People build in buffers, they ask about things like confidence, they think that they're taking that seriously, and in actuality, what they've basically just proven is that it's almost impossible for their project to succeed.

And this isn't something people discuss a lot.

So what's the problem? The core problem, everybody seems reasonably confident in their little piece, and yet the whole is far more likely to fail than to succeed.

Let me walk you through this.