Speaking as a down-in-the-dirt circuit-slinger who works alongside a lot of ivory-tower engineers who couldn't diagnose a loose wire if their life depended on it, I think that's the whole point.

* Tens of hours and under $1000 is an extremely cheap route to this level of understanding if compared against university courses. I'd give my left nut to have my coworkers go through this process.

* Cutting and laying out wires is time-consuming and error-prone. Yup, hardware is like that. Sweeping out the dojo is menial but important. Developing an intuition around when a weird behavior might actually be a hardware problem, is priceless and absolutely essential for any embedded engineer. And you will never, ever get that from a simulator. (As much as I love the ideas behind nand2tetris, it's entirely done in simulation and that misses half the point, IMHO.)

* Sad to say, the outdated components thing is only getting worse as more basic stuff goes out of production in favor of more highly integrated components. (Even the epic VULCAN-74 caved and used modern RAM.) It might be possible to rework this project to use newer parts but keep the educational value. I suspect there's enough people who've pursued this project and done bits and pieces of that, someone just needs to unify it.