 |
February 15, 2014, 10:25:17 AM |
|
This topic is just what I needed.
I'm a computational economist currently writing a paper on how certain innovations, including blockchain-based contracts (of which Bitcoin is an example, but not the only possible one), may impact the way we think about economics. The paper is far from finished, but so far I am under the impression that two very different worldviews are coming together in a way that has few precedents.
Vitalik Buterin's concept of DACs incorprorates, albeit implicitly, the neoclassical idea of homo economicus. There could be no DACs without the idea of optimizing, utility-seeking agents who solve maximization problems. Here I am purposedly mentioning "utility" as opposed to "profit", as DACs might have social/charitable goals etc; what counts here is not the greed motive per se, but rather the underlying idea that economic behavior can be modeled in a highly deterministic fashion once goals are known.
Looking at the way economic literature evolved in the past, say, 20 years, this is an interesting step... backwards. Please note that there is no element of good/bad judgment here. I'm just saying that the behavioral economics revolution championed by Nobel prize Daniel Kahneman, and the increasing relevance of psychology/neurosciences in the explanation of economic choices, point to phenomena such as bounded rationality, preference inconsistency etc. that suggest deterministic modeling has little chance of working in any real-life context. That's not because factors other than money are at play, that would just lead to a need for more variables in the models. That's because some of these factors are very hard to represent mathematically, either because our knowledge of the human brain is still insufficient, or because their nature is intrinsically stochastic.
Of course with DACs one is not trying to create a laboratory version of existing markets/firms so as to forecast their behavior; the firms and markets themselves are what's being created, in an experiment of artificial economy that will be welcomed by those who embraced methods such as agent-based modeling as an operational tool as opposed to an instrument exclusively dedicated to representing reality. DACs may be, in other words, an implementation of what the world would look like if traditional economic modeling was methodologically correct.
I'm saying that this brings two very different approaches together because the use of artificial intelligence in economics (decision theory/microeconomics, not algorithmic trading!), mostly machine learning at first and then anything to do with emerging complexity, was going in lockstep with brain/mind sciences as a way to bypass the limitations of closed-form mathematical representations of utility-maximizing rational agents. Now, on the other hand, artificial intelligence might be one of the constituent parts in the creation of a world that is populated by such agents.
How would such an economy look like? One first conclusion I have is that it would not look very similar to what we have right now. In the DAC world, humans are supposed to be present at the level where the carrying out of specialized tasks is required, but the rest is self-sustaining and, well, decentralized. This looks nothing like real firms in terms of decision-making processes, but the bounded rationality biases etc are going to be at work right when contracts are designed. For a different paper, I'm trying to model the differences in outcomes between traditional firms and DACs based on this.
If you found this stimulating/useful please consider a tiny donation (as in 0.001 is great already!) to 1KQNcjSTmPx5j7atMWUAk26NZm7XpfExFY, I'm setting up a bitcoin-based behavior experiment and I can't be bothered wiring money to an exchange to buy a grand total of about six dollars worth of bitcoins...
And by all means I am looking forward to exchanging ideas on this subject with anyone who would like to contribute their own point of view. IMHO DACs could be one of the next big things in economics.
|
