It’s easy to focus on the researchers at Bell Labs didn’t have a unique model like Shocklee, Shannon and others. However, the majority of people there were engineers doing much more mundane things like assembling and testing equipment.
The things we think of as massive wins are hits that happened over ~50 years of operation. This is an absurd organizational time scale.
When Bell Labs began, only about 15% of its employees did research.
Of the two thousand technical experts, the vast majority worked on product development. About three hundred, including Clinton Davidson and Mervin Kelly, worked under Harold Arnold in basic and applied research.
Arguably, the things we lionize Bell Labs for probably not have been massive wins without people around who were deeply entrenched in manufacturing processes and products. Prototyping needs manufacturing in the room.
On top of all that there was pretty clear status inequality between the people working on sexy research and the people doing development. This inequality is one of the reasons <name redacted> argued that At the organizational level corporate R&D labs either cause brain drain or don’t have the best people . Bell Labs may have routed around this slightly by being a completely separate entity from AT+T. However, it’s not clear to me how it maintained a strongly hierarchical system for decades.
Some possibilities are that people could move from development into research if they were good enough. Another possibility is that there were legitimately two prestige tracks you could move through and while development was more salt-mine-y in the beginning, you could end up doing bigger things; Chuck Elmendorf ended up directing both the transatlantic cable project and the waveguide project. Another possibility is that alternatives were less good - now unsatisfied Bell Labs developers could just go off and start a startup. Yet another possibility is that, like Google today, the compensation was good enough that even if the work wasn’t the most prestigious within the organization, it was stable and the money was good and you might not need the crazy outliers in these development roles.
There are two key upshots here. One is that you need more than just researchers to get impactful technology out into the world. Either these people need to be part of the organization or you need to plan a clear way to interface with them. The other (related) upshot is that there is a lot of grunt work that needs to be done to actually build impactful technology. Incentivizing that is hard short of the money you would make starting a company. It’s worth thinking deeply about how to tackle that.