I've been wrestling with this for some time and have formed some strong opinions on this topic. I would love to back this up with data. To preface, my industry is A&D, low volume, highly complex, systems. We might build one unit for a customer and sell it (like selling the first prototype). High ratio of tooling to deliverable hardware. The common ask is for faster change processes, less reviews, get to floor faster so we can start building. The common solution is prototype process or new product introduction where the revision series (typically numerics) indicates prototype, not for production and that switches to MilSpec once we cross to a production review, typically with an advanced lifecycle.
Does this enable more revisions, less quality reviews and more chaotic design output? I am also a firm believer in "Don't track status by revision" so having an item at 3.1 should not mean anything more than B.1. This process also forces all items to go through prototype processes once and then again to get to "production" revision letters. Seems like all items need two revisions minimum. I would argue that most items can be designed and built with one revision and be correct (that's the goal isn't it?).
As for quality of design, this rapid change change change could lead to skipping critical design checks that would cause more changes to be required later in build. Again in my case, where that prototype could be the one you ship, it could lead to a case where its built and done, now comes the formal detailed review to move ALL components and parts to production revisions before ship.
And isn't the real only prototype we are taking about is the top level assembly or assemblies? That is the main item that people are concerned with. "We have to release so we can order parts or we won't know if this works till we start building and testing". I would advocate for delaying release of designs until we reach that part of the assembly and manufacture. So in schedule drive organizations, we release up the structure holding open the top level to absorb the changes occurring lower in the assembly structure. Once all components are released and close to being completed, we can complete the assembly and then starting putting it together.
As far as design reviews, we can always use methods to tailor signoff as needed, reducing the reviews to the critical ones. No need to create complex and manual version series and lifecycles. There is also the option of segregating those true science projects as separate part numbers, risk reduction units, true prototypes or engineering models. Again, same simple version series and process, just different review levels. So does anyone have true data that shows one is better then another (versions per part number averages)?
