I'd even argue that the burden of proof is reversed. There might be billion dollar companies which bring good, but I'll be skeptical of that fact at first.

I would argue that there is plenty of large(above billion dollars) companies that bring net good to world. Manufacturing is good field. Producing things like machinery improves quality of live and productivity.

Not that these often doesn't have negative aspects when they strive for even higher value extraction, looking at something like John Deere. Still it does allow massive efficiency gains in farming.

I wonder if there's a general rule that the larger an organisation the more pronounced the diffusion of responsibility and the higher the likelihood of shady behaviour? It would be interesting to see some studies into this either way.

I would imagine this gets progressively harder the bigger you get.

Thus why one should rather prefer slow growth. Fast growth only means you’ll be more likely to reach the shitty state faster.

But slow growth brings less money to top management, and top management decides if they want to go for slow or fast growth, right?

I think the simplest way to measure that is by asking 'what would we truly lose if this company disappeared tomorrow?'. With one follow-up question of 'how easily could the important bits actually be replaced?'

No; that kind of thought experiment heavily biases in favor of large organizations.

It's much more helpful to ask, "What would we lose if this company were broken up tomorrow into parts no larger than (say) the median company size?"

I think it does the opposite, it makes you think about whether the company is big because of its products/services or other reasons. It's not a yes/no answer, it's a prompt.

> 'what would we truly lose if this company disappeared tomorrow?'

I guess such a big company is here because they managed to create a need. The question may be: "were we really worse off before we had what this company built?".

Example: if GitHub disappears tomorrow, that's likely a pretty big problem. But we were fine before GitHub, we just had different (not worse) workflows. GitHub created a dependency.