So what you're saying is that without domain knowledge you can make an accurate allocation decision ? I don't believe that. Can you demonstrate ? I'm making a

So what you're saying is that without domain knowledge you can make an accurate allocation decision ? I don't believe that. Can you demonstrate ? I'm making a device, it doesn't matter which kind, rocket or car, or what it does. Why don't you give me a plan to allocate production of this device. Of course, it has to apply to all possible devices, rockets, boats, cars (fisher-price and the regular kind), ... Since you say this decision can be made intelligently without domain knowledge, there must be a general solution. As an AI aficionado, I am EXTREMELY interested in this solution and I can personally guarantee this to you : If you have even a basic (but non-trivial) beginning of a plan that actually works, I'll make sure you get the Fields medal at least 5 times, or 1000 billion dollars, because a discovery of that magnitude absolutely merits such a reward, and the money would be easy pickings with such an algorithm. Hell it'd be rather easy to become the leader of a robot empire that easily defeats the US and every other nation on earth if you have such a solution. But frankly I expect that you're bullshitting. Out of the dozens of lunatic proposals I've received from various sources, from students to foul-smelling people I'd never seen before (or again), they at least mentioned the method they claimed would work. None were even remotely workable. You just claim that "some people have learned it". I disagree in the strongest possible manner. Please, please convince me wrong. If it can be learned, it can be taught : so teach us, right here in this thread if it does indeed exist. I claim these people who have learned "allocation" have not learned a thing. If they did, capitalism (a learning algorithm itself) could never possibly win against communism or dictatorships (because one of them would have had a leader that got allocation right). Instead, manager success is attributable, exclusively, to either the skills of the people they "allocated" and got things done (usually by undoing at least some allocation decisions), some external factor (it is no great feat for a microsoft product manager to build a program used by 500 million people due to monopoly), or shear dumb luck. Oh right, there's also bribery and nepotism to account for management positions (as opposed to success). This sort of arguments makes the ideas behind software development methodologies make much more sense. Especially scrum, but agile before it and others seem to be highly focused on taking allocation decisions out of the hands of people with this "allocation" "skillset", like managers and project managers and the like, and into engineer's hands by moving them into systems the engineers control. I would also like to add that, in the business article/book way of saying things, "most" successful companies, from Ford, over Microsoft to Google, became successful and grew when "allocation" of resources was effectively not done at all in these companies. Especially human resources that brought these companies their big wins were often going against management decisions (or management was giving them 20% time. Except of course, they spent a lot more than 20% on it).