Conglomerates Didn’t Die, They Look Like Amazon

The are a meta store, anyone can set up a store and do the hard work of marketing and fulfilling orders and Amazon gets a slice. Then Amazon started hanging on to other people's inventory too, now you just need to buy the stuff and Amazon will hold it while you go out and do the hard work of marketing and selling and Amazon gets a slice. Their know there is scarcity of products and try to connect sellers with buyers.

With AWS they know new businesses will use their infrastructure but they don't want to take the risk with how it will be used. Any startup with any idea can take a real existential risk on their business idea and Amazon will take a slice. They know that for all its ephemerality information in the correct place is scarce, but they don't know how to solve all the problems with information. They made good tools and lease them to those the tackle information scarcity.

Where it doesn't make sense to have other people do it, they try to reduce costs by literally moving things from a to b. Their drones are the their most vivid example but there are countless mundane Amazon warehouses strategically placed. If Amazon doesn't know if the risk/reward is good enough they will contract it out and let the other assume the risk.

They are automating or distributing the ways of reducing scarcity and leaving themselves with minimal risk. As their risk and cost gets lower it makes sense for them to tackle ever smaller amounts of scarcity.

They are so efficient at dealing with scarcity that it makes sense for them, a multi-billion dollar company, to give you a button with the sole purpose of ordering you more laundry detergent by pushing that single purpose button. They are leveraging the scarcity of one empty self in one tiny room of your home. If that isn't the story of a company with a deep relationship with scarcity then I don't know what else it might be.

Decision theory generally proceeds as follows: map out all possible outcomes, all possible courses of action, estimate probabilities of each action leading to each outcome, and choose the action with best outcome distribution.

So far I haven't seen anybody argue against it directly... the problem instead is availability, most people aren't aware of it and can't or don't use it on a daily basis.