While Bell Labs didn’t have a unique model gave researchers a ton of leeway, they weren’t given complete freedom. Instead, they were generally given high level directions that would potentially be useful to AT&T - “figure out how to get more information in this copper wire” “figure out how to make switching speeds faster and more automatic” and then given freedom. They also just biased towards hiring people whose research interest lay in areas that seemed vaguely useful to AT&T. Remember, AT+T benefitted significantly from creating better technology.
All the stories of great bell outputs do seem to start with “so I was working on this practical high level directive.” However, unlike most goal-directed research now, it does seem like researchers were given the freedom to pursue tangent research based on ‘huh that’s funny moments’ like The ability to say ‘huh that’s funny’ is important for discovering new phenomena.
There’s a strain of thought that says “researchers should be given complete freedom to investigate what they want” independent of any consideration of use. This line of thinking is the main theme of brabenScientificFreedomElixir2008 for example. It’s important to call out that this is not what Bell Labs did because in conjunction with Braben’s arguments the contrast suggests that both approaches can lead to paradigm-shifting discoveries. To use the Quadrant Model of Scientific Research, Bell Labs operated from Pasteur’s Quadrant. The subtlety is that while they started with a consideration of use, sometimes it ended up with currently-useless fundamental understanding (like Cosmic Microwave Background.)
Like many things, I suspect that hybridizing the two approaches will lead to failure. An organization needs to pick an approach and commit to it.