Modes of provisioning and (ahem) sci fi about procurement

This might be a bit leftfield, and I think it’s largely addressed indirectly in science fiction novels like New York 2150 (in their normalised use of shared kitchens for cooking and so on), but I really enjoyed @alberto 's piece about modes of provisioning here - as a way to think a a systems level how to meet a given need:

It’s a new concept to me, in my day job, I’m finding it helpful to think through an alternative framing for explaining why massive competing build outs of AI infrastructure to deliver largely same kinds of Generative AI models is a bad idea and also why relying on large companies who are incentivised to aim for more than 30% net profits year on year might not be the most efficient way to decarbonise the digital sector.

It also seems like it would be a good fit to introduce into discussions about procurement and contracts, and in the context of a new European commission how you might go about hitting some of the specific goals like 55% drops in carbon emissions by 2030, because a lot of the time, a tender feels very much like setting out a bunch of needs you want to have met, and then (ideally) optimising for the most efficient mode of provision - assumign you’re not in a massively concentrated market, obvs.

Looking for left field examples of procurement as as plot devices in sci fi to refer to

Anyway, my understanding is that legislation like the CSDDD sort of requires affected companies to demonstrate how their own climate action plans are broadly inline with goals like the 55% by 2030, as well as mitigating against a bunch of harms (see this explainer from Normative).

And in that context, I found out that there is a whole piece of work around creating model clauses to use in commercial agreements, with an focus on shared responsibility for mitigating environmental and social harms. This slide deck is a good summary.

Has anyone come across stories where people experiment with these ideas to say… aim for the most efficient mode of provision along various criteria to hit an given outcome?

1 Like

Hi Chris, just wanted to flag that I am reading you. I need to think about this, but I will get back to you.

Related is Brendan Ballou’s Plunder: Private Equity’s Plan to Pillage America.

Ballou is talking to Hanauer here: Stop the Steal: Revisiting Pri - Pitchfork Economics with Nick Hanauer - Apple Podcasts.

I couldn’t bear to listen to the whole thing . . .

1 Like

@mrchrisadams , I concluded that, if you want this:

Then game theory serves you perfectly. Basically, you have markets with large fixed costs and very small, non-increasing marginal costs. Just from static partial equilibrium theory, these markets you can serve most efficiently (lowest costs) with only one firm. To avoid monopolies, you might want to encourage more, but still only 2-4. Oligopoly equilbrium still allows for extra profits: moreover, these markets are hard to contest (once you are the 800-pounds gorilla of AI, you buy any contender, etc.). Given cost structure and netwrok externalities, this is a game of speed: first three firms to launch take it all. Hence the inefficiency you notice.

Here I think you need a few more moving parts.