Best Of

@euan_dykes I do think it can be taught, but I do think some people are better at starting things and some are better at finishing them, and that's okay. Roll with it.

I also hold onto a concept I helped establish at my last company of "levels of CAD" I had a video about it at one point that's been taken down and I plan to re-create it, but the gist is this: The level of rigor in your CAD should be proportional to the clarity of design vision. Rigor means a model that's robust to change, navigable by guests, and lightweight to compute.

To break that down imagine a design starts as a sketch on a napkin. There's the spirit of an idea and not much else. There's not confidence it is desirable (people want it), feasible (it can be done), and viable (makes business sense). At this early stage it makes no sense to make incredibly robust "good" CAD. Your CAD should suck, but be quick to do (Level 1). This CAD is just for the CADder, and no one should care how messy it is anymore than they'd care what messy napkin sketches they have stuffed in a drawer. Iterate until you and your team agree it's believable that it could tick the three boxes of desirability, feasibility, and viability (DFV). You don't have total proof, but it seems worth investing time in proving. Here you begin iterating over the same concept, but with little variations. The core idea is pretty locked in, but there's work needed to actually prove DFV. You're probably trashing all of the models you've made so far, and beginning from near scratch on this Level 2 model. This is when your team might begin sharing models and making prototypes you want to keep track of, so you need just enough rigor to support that. Things should be named, feature folders used, old irrelevant tabs deleted, etc. Once you've iterated enough that you have confidence in DFV and a clear design vision (a very concrete idea what you want to make), then the business is ready to commit to the real product development. That's when it's time to do Level 3 CAD, which is the very robust kind of CAD you were asking about here.

You mentioned risk reduction, and that's exactly right, but the risk at the beginning of the project is over-investing in an idea that isn't going to work, and the risk at the end, is that you do a bad job making it. The beginning is about making the right thing and the end is about making the thing right.

Without everyone agreeing where the project should be sitting on the speed-rigor continuum, you won't agree on what kind of CAD is appropriate. Your ID friend might actually be right that it doesn't make sense to make a really stable model because the entire design vision is also not stabilized. But if the company is committed to the product and it's production CAD I think it is worth the time making it very robust. I find that people (let alone teams) are not clear with themselves about this continuum, and therefore overdo it on Level 1 CAD and under do it on Level 3 CAD, when really their worst CAD should be even "worse" and their best CAD should be better.

Wow that got longer than I meant. I'm getting excited to re-capture it in video form!