What would this even look like in software?

I always wonder if software is different than physical construction, or if software is just less mature of a discipline.

In software, we can’t estimate projects accurately and consistently. We have to build a few to throw away just to get a better (yet still incomplete) picture of the problem we’re trying to solve.

Imagine if the people building your house had to build half of it and then start over. Maybe twice.

That never happens in physical construction. Maybe something has to be redone because someone made a mistake, but almost never due to not understanding the problem. So what’s different about software?

Correctness checking in software already exists for mission critical applications and is seen in sectors like spacecraft or avionics and to a degree in core finance etc.

Development in those fields are lot slower and very conservative and is by no means perfect, it is matter of culture and regulation and what you are ready to spend .

Project management challenges should not be conflated with product quality. Take JWST, a notoriously hard project to manage costs or timelines, but their product quality was perfect, if we can launch a telescope like that accurately the first time we can build software well, if we can afford it.

> Imagine if the people building your house had to build half of it and then start over. Maybe twice.

No need to imagine, I have seen people do that all the time, if you have the money you can afford endless remodeling and some people actually do that.

It all comes down to appetite to spend on good quality and culture to do so, in the era of low/no/AI code or off-shoring before that, there is constant downward pressure on costs, quality and security are the trade-offs.