This is similar to federalism -- central governments don't work very well for everything, but scaler local governments are still effective for local issues.

Pour one out to anyone who has ever done a sprint retrospective, or a company 360 review, where all the important problems get skipped over because 'there's nothing we can do about this, this, and that.'

Line workers and managers are given an incredibly narrow window in which they are permitted to operate. Anything beyond that needs beurocratic approval. Much like in the Soviet Union, sometimes that beurocracy is staffed by intelligent human beings, but usually it is staffed by donkeys. People who care enough about doing their work find ways to work around the latter, or just give up and punch in their 9-5.

Hypothetically, A firm like Tesla or SpaceX maybe be so successful in engineer feedback because they are high demand places to work and they have not exhausted the pool of individuals who can provide great work. However as the company needs to scale, or as the company becomes less glamourous you have people who either (1) just aren't as capable or (2) don't care to expend any creativity on solving the needs of the company (either because there's no reward or they are interested in other areas).

Furthermore, a company that is not centrally planned will most likely move slower than a centrally planned one. Just like how China is able to forcibly take your land to build railways; an action like that in a free society has many gatekeepers that would slow down such a process. Any public company, that is measured on a quartely basis would find themselves grappling with showing change to wall street.

For R&D departments and startups, plan economies can still work because the complexity is still small enough that humans can comprehend it.