Ok, if I understand well, you distinguish
(a) EoS, which are leading to increase returns of scale at the beginning but eventually experience diminishing returns to scale
(b) what you call Southern Side, composed by NE+ AoP + ML, which would be leading to never-ending increase returns to scale.
And that explains the shift of paradigm:
> Old world: lowering the price (to kill the competition) but still capturing the value before reaching the point of decreased return to scale
> New world: lowering the net profit by increasing capex (not lowering prices), because you don’t need the Northern Side to capture value but you need to ensure the permanent growth so the Southern Side can generate profits.
I would totally agree with that.
Yet the Southern Side (I love the image btw) have also some limitation/diminishing returns to scale, like on the Northern Side (but with different time frame):
- network effect undergo local network effect
- the architecture of participation (which are based at least partially on network effect as well) suffer from noise generation — due to bots or trolls in user-generated content/filters/rating systems
- machine learning face computational limitations (which imply learning limitations, but might be resolved over time)
Eventually, I believe everything is about finding the right balances between the business models & time frames -which requires a good business model engineering strategy, as you highlight in p2.