Starting with a universal wood/metal working machine tool like a Smithy Super Shop, for instance, with the goal of reproducing the tool, I see a few ways open for progress:
One, starting with raw metal stock, reverse engineer blueprints for the machine tool and then build a copy from the blueprints.
Two, Starting with two of the machine tool, disassemble one copy, using each piece as a model, and then copy each piece.
Three, Order all the parts as replacement parts from the manufacturer, and assemble a copy.
Four, Order all the parts as replacement parts, and set about reproducing each part to build a copy.
Five, If the machine has N parts, order N machines and assign to each machine an operator capable of reproducing one of the N parts on their machine, then assemble a copy. Seems simple enough: you need to pick a machine, hire N operators, order N machines, and just have at it.
The reason I say wood/metal working machine tool is that I am pretty sure any universal self-reproducing machine tool has to have this character: Woodworking operations usually let the part guide itself through the machine under operator control, and this means such a machine has a work envelope that is potentially of infinite size; it certainly isn't closed or of fixed size. Metal working operations such as millling usually hold the work in a fixture of finite size and capacity, and so have a closed work envelope. Best of both worlds?
I have an email out to sales at Smithy asking for a part count and pricing for the set of replacement parts needed to assemble a copy of my Super Shop.
Doug Goncz Replikon Research Seven Corners, VA 22044-0394