And we know from research, tons and tons of research, that it is exponentially more expensive to fix a problem that has reached production than it has reached, than it's only reached like a developer or a test instance or something along those lines.

One of the things we need to realize is that software delivery is really powered by those feedback loops.

You build really kind of short feedback loops using things like automated tests, where you can get an almost instant answer to the question of is this software any good? And then you build out from there in layers more and more feedback loops.

Um, and that should include a feedback loop from the software's users, um, you know, uh, to find out if it's working for them and solving their real problem.

The more of those loops that are missing, the worse it is.

Using Bob's example of this rainfall, um, I know in the UK we have this watertight house checklist.

It's a list of things that have to be in good order in order for rain to not get inside your house, and it starts with, um, I think it starts with the roof and the guttering, and then the windows, and then the damp course, and then there are these little trays that sit, um, uh, as part of the window frames that direct water out if it manages to blow in.

Do we have it ready yet or is it still a work in progress?