pull down to refresh
883 sats \ 9 replies \ @elvismercury OP 15 Nov 2023 \ on: _Broken Money_ book club, part 4 bitcoin
In a famous paper, Hayek laid out the argument for the decentralized intelligence of the market more clearly and succinctly than anywhere I've seen. Lyn cites this paper, and asks:
This critique made sense in the era in which Hayek was writing. However, given that the availability of information, the speed at which it propagates, and the computational and analytical power that can be used to process it have increased to almost unimaginable degree since the Soviets attempted to centrally plan their economy, is it possible that Hayek's assumptions about decentralized intelligence no longer hold?
A good example of why Hayek's quote is as apt as ever is the U.S. central bank. Of course Powell famously got "transitory inflation" wrong, but that bad call was just one example of how central banks have no ability to predict the economy, much less intervene in a useful way. The forecasts are never accurate, despite all of the information available to them in our technologically advanced world. Yet the MSM covers these stabs in the dark as if they are papal decrees.
reply
Solid point. Hard to rebut.
An interesting topic to me is what the "right" level of organization is. The ecosystem is not just a bunch of single-cell organisms transacting with (or against) each other; it aggregates at multiple levels. Multiple-cell organisms abound. Coase's theorem frames it in terms of transaction costs. And I think we've seen that tech increases the allowable complexity / the carrying capacity, even if bullshit jobs emerge.
So whatever the size and scope of government should be, clearly it's different now than 50 years ago, right? But what is it?
reply
is it possible that Hayek's assumptions about decentralized intelligence no longer hold?
It still holds. It's possible it doesn't hold forever though. I can imagine a centralized decision maker that's smarter than the entirety of the market. You can kind of naively model the decision ability of a system as
decision_ability = sum(intelligence of agent(s)) + sum(value of information gathered) - sum(information transaction costs) - sum(information gathering costs).Parents of small children are an example of this at play. It implies that if individual market participants are extremely dumb or extremely inefficient at information gathering relative to a centralized decision maker, the centralized decision maker will allocate resources better.
reply
I like your equation. I also think there's something hiding in your "value of information gathered" having to do with orthogonality of information that isn't made enough of. Or perhaps, we could just talk about "information" in the Shannon sense, where these redundancies are formalized.
In any event, you can imagine a sensor-rich world where there's simply more, and more useful, info avail to some centralized mega-brains, enough that allows them to outcompete the more numerous multi-agent model.
reply
Or perhaps, we could just talk about "information" in the Shannon sense, where these redundancies are formalized.
Ah great insight - we could quantify information novelty/scarcity/surprise. I wonder if we could use Shannon to model how fast our infinite aliens would get bored.
In any event, you can imagine a sensor-rich world where there's simply more, and more useful, info avail to some centralized mega-brains, enough that allows them to outcompete the more numerous multi-agent model.
You just wrote the plot to at least half of the next decade's dystopian scifi movies.
reply
However, given that the availability of information, the speed at which it propagates, and the computational and analytical power that can be used to process it have increased to almost unimaginable degree since the Soviets attempted to centrally plan their economy, is it possible that Hayek's assumptions about decentralized intelligence no longer hold?
The piece of the puzzle that you're missing is that value is also subjective.
It can only be done at the individual level effectively, since every individual is different. When you attempt to centrally plan, it simply means that you're not capturing those differences at all and therefore you allocate resources sub-optimally when compared to what each individual, with his differences, would have done.
The state can only choose "1 thing for many people". "Many people" can choose "many number of things", all different from each other.
reply
I'll grant that value is subjective, but it's also the case that, despite the fact that people are precious snowflakes, people have lots of things in common and there's overwhelming agreement on many of them, like a central limit theorem analogue of subjective value.
When dealing with hundreds of millions of people, a lot of these things smooth out.
reply
Not at all.
Even something like a chair, that everyone might need could not be centrally planned.
Every house is different, each table height might be different, climates across the country might be different which would require different preferred materials, etc.
Absolutely no personal good or service is better when chosen by somebody else than you, the only one who has perfect information about your personal needs.
reply
The issue is less whether a chair, or your underwear, could better be chosen by some faraway superbrain than by you.
The issue is more whether locating a highway, or coming up with an industrial policy, or making the assessment to not allow straws in NYC, is a decision that can better maximize human flourishing than aggregated individual market choices.
reply