Going to get a bit meta, here, and talk about risk.
I have spent my life working mostly at small companies and start-ups, where projects often operate at a very high level of risk. (As in, if the project fails, the company might cease to exist.) While in the past I have sometimes certainly got things wrong, I did learn.
You learn to factor your risks. There are lots of folk who have written on this topic, so it is very possible to learn.
In my current foolish exercise, I am intentionally running a high level of risk - trying lots of things that might not work - of purpose. This is more fun. 
My last work (software) project ran at an extreme level of risk, and delivered on time and on target. Might just be that I learned.
You learn to factor your risks.
In my current experiment, I moved my risks to the front, to discover problems, but moved revisions back, to batch iterations.
In the pictured frame, I have a long list of queued revisions, but am proceeding with the present for best progress. Every last bit of pictured plastic will be re-printed before the final iteration. The CAD designs are updated, but the present iteration does not require those changes.
To be clear, projects (in software or hardware) often get lost when they spend too much time in iterations. You have to batch your iterations, and judge when an iteration can be deferred, or is needed.
Examples?
All the braces are printed at 20% infill. That saves time when iterating, but means weaker prints. If the current prints are adequate, prints at 40-50% infill should be superbe.
The motor-mounts are PLA. The motors might get warm enough to soften PLA, so they could be printed from ABS. There is some suggestion my chosen Trinamic stepper-drivers might warm motors less.
The 2020 extrusions chosen for this exercise are (intentionally) a bit light. The rails in Z have an unbraced span that acts like a spring, and resonates at about 100Hz. (The rest is remarkably dead to resonance - as hoped for from the plastic/metal junctions.)
My next major iteration (if I get so far), will use 3030 for the frame.
The singular stepper driving the print-bed in Z is (naturally) positioned in the center of the rail. The load on that stepper is low, but enough for the rail to flex in torsion. As the bed only steps smoothly and slowly down in Z, this is probably not an issue. But I do not like the flex.
The diagonal top braces will get a bit thicker. The current braces work, though they are ridiculously light, but need to thicken so to use fewer screw lengths.
And onwards…
