[jbondeson]

I'm sorry, I wasn't very precise in my language, I meant the industries working together to generate a more generic process that isn't tied to a specific industry. This would be done with the intention of building a Professional Software Engineer program.

Anything relying on the physical world obviously cannot be mathematically verified, however the mathematical rigor is still there in the modelling and testing.

[tjr]

Ah, I think I follow then. Mathematical proof of very large software systems may presently be impractical; the verification procedures currently in place for avionics systems are pretty daunting as they stand.

I'm not sure what we would gain from an overall software engineering professional qualification. While I am quick to point to avionics development as (in my opinion) being true software engineering, I also acknowledge that a great deal of software is not built using any methods even remotely as rigorous. But then, it also probably doesn't need to be.

You put a lot more rigor into building a house for people than you do a house for dogs, and you put a lot more rigor into building a skyscraper office building than you do a house. And rightly so. Each needs to withstand different levels of pressure. I see no reason to apply rigorous engineering processes to iPhone games; it would be a waste of time and resources.

As it is, the industries that require rigorous software development have produced their own standards. I suppose if you could ascertain the similarities across industries and codify that into general software engineering practices, that could be a good thing, but you'd probably still need industry-specific regulations to make sure nothing was missed.

I dunno. Maybe if more software needed extensive rigor it would be more obvious what needs to be done across industries.

[jbondeson]

There are tons of industries that should require it. The issue is that we focus in on the software that can directly kill us, rather than all the software that can indirectly do it. We instantly reason that all software we use to control things like cars or airplanes should be highly controlled, but what about financial software, phone systems, medical records software?

One of the sloppiest and haphazard programmers I ever worked with is now writing software for non-critical medical monitoring devices (external monitors like O2 sensors, and blood pressure monitors) and it scares the living daylights out of me. When I'm in a hospital room I look around to see if his company's hardware is in use. It might not kill me, but it could contribute to that end.

I'll come full circle around to my initial assertion that I don't know if we can even get to the point of a true Professional Software Engineer. Many of the tasks we deal with are highly abstract. How do we define an acceptable failure rate for a TCP stack? And even if we do, how do we take that into account for a monitoring application? There are many hard questions we have to tackle prior to our industry being taken seriously as a true engineering discipline.

What troubles me is that very few of the leaders in the software community even care about these questions.