Even in containerized software, which is basically the most modular and legible thing in the world, there is a ton of friction handing a project from one team to another team. “Hell is other people’s code.” ‘Off the shelf’ technology is code for TRL 9 and maybe 8.
Technology transfer makes sense in a world where if you can look at a drawing on a piece of paper and know everything about it, which was the case until the late 19th century.
We can abstractly think of you handing the “IP” - patent documentation to me but in order for me to actually do anything with it, so much more needs to be done. Patents aren’t even written by the people who created the technology anymore! Perhaps an extreme analogy but patents represent technology the way the dollar represents gold. The way in which a patent is insufficient to transfer a technology is analogous to how Just having the piece of literature doesn’t solve the problem
A patent describes the key new concept but usually doesn’t touch many implementation details. Since The creation of a heuretic has two steps - conception and implementation, it means that patents actually underdetermine a heuretic. This underdetermination means that raw “IP” is useful for suing someone who successfully copies the idea (because to copy the whole of it they need to copy the new concept piece), but it is not particularly useful for copying the heuretic itself. See Dupont was unable to replicate viscose rayon. The changing structure of American Innovation - some cautionary remarks for economic growth cites this phenomenon as one of the drivers of the creation of corporate labs. The implementation piece is (I suspect) one reason why large companies acquire startups for their technology instead of just buying the technology itself. A technology company is a package of human capital and IP.