Comment by amluto

Surely there’s room for a middle ground. There are plenty of 1990s-era engines that were excellent designs, had no meaningful connectivity to anything except their own ECUs, and could be produced new for not very much money. Some of them were quite modular, too — I know someone who took the drivetrain out of a salvaged Honda Civic and built an entire car (with no resemblance whatsoever to a Civc) around it.

If a tractor with a clean-burning, efficient $7500k engine could be purchased and were designed around the theory that, in 20 years or so, the owner could reasonably quickly replace the entire engine (with a first-party or aftermarket solution), would that be a good solution?

The common tech that has solved these problems nicely (IMO) is network transceivers: SFP and similar modules are built according to multi-source agreements. They contain all kinds of exotic tech, and they are not intended to be serviced at all, but (unless your switch or NIC has an utterly stupid lockout) you can pull it out and replace it with an equivalent part from a different vendor in seconds, and those parts can be unbelievably inexpensive considering what’s in them. (Single-mode bidirectional 1Gbps transceivers are $11 or less, retail, in qty 2. This is INSANE compared the the first time I lit up a 1Gbps SMF link. To be fair, this particular tech may require one to replace both ends if one fails, but if you can spare a second fiber, the fully IEEE-spec-compliant interoperable ones are even less expensive.)