Hi folks. I am a CS major taking a 3rd year course in relational databases. The example DBs we study are pretty much all either a school or a company. On the bright side we get to do a project of our own design with C++ and Oracle DB. Has to be some kind of program that makes use of a reasonably sophisticated schema.
I was thinking I could make a DB program that does economic planning, but I don't know what direction to go with it, really. Maybe the kernel of it, the usefulness could be, computing everything down to hours of human effort using the LTV. Labour time accounting. For example, we create a profile for what we want the living standard to be, like private and shared square feet per person, food choices, clothing choices, level of convenience of transport etc. Then the program could use a database containing information about the SNLT to produce different products and services to compute what professions would be needed and how much we all need to work, basically.
But like any idea this is starting out huge. So does anybody have ideas for how to make this small but extendable? Or different directions go with it, or totally different ideas that you have?
While I haven't read it yet myself, I've heard that Paul Cockshot's Towards a New Socialism goes into detail about algorithms that could be used for central planning.
I read it years ago, and I should definitely dig in again and review. Big part of why I want to do everything in labour time as much as possible. However I think he suggests the use of a neural network at one point which is a little over my head for now. I am thinking simpler like the pen and paper material balance planning the Gosplan cdes used to do...
I haven't read the book yet, but Cockshott seems pretty active on twitter. You could probably hit him up with questions either through twitter or his university email, and I bet he'd be more than willing to give you some pointers for reading material.
Regarding the neural network, could the whole thing be represented by a linear regression? I'm shooting blind here, but depending on the complexity of your project you could use simpler but easier models that do what the neural net is supposed to do.