Let's assume for the moment that the stock market is an efficient way to distribute capital in endeavours that will produce surpluses, for the pie-embigenning benefit of all [and I am becoming increasingly skeptical about that, but that's the premise that capitalism has been sold to us on]. Is this level of sub-second trading really necessary? I can see who benefits from each individual action, but who benefits from the system being set up like that in the first place? Is there evidence that making price discovery faster by seconds or minutes benefits the economy overall? I very strongly doubt it. Once the stock market is efficiently allocating capital in response to information issued by the company and news, then that's enough, and the rest of stock trading is a zero-sum game.
[And let's not forget that day trading does not in fact distribute capital to companies. The point of stock exchanges is to provide a liquid market for company ownership, so as to encourage investment at flotation/new share issuance time - which is when the actual company gets the benefit - so it's already an abstraction away.]
There was a story recently about a new cable being laid down between London and Toyko, to bring the ping time down from 230ms to 168ms. They spent actual resources in the real world, with actual engineers and an actual boat, putting down a cable to let people conduct high-frequency trades that bit more efficiently than their competition. Of course, their competition will just put down their own cable, and so on. Eventually, you're left with nobody having a timing advantage over anybody, and the system is exactly the same overall as it was. It's like the Red Queen said - "it takes all the running you can do, to keep in the same place." Who loses out? There's the opportunity cost of what all the cable-layers and programmers could have been doing otherwise. And crucially it increases the barriers to entry for participating in the stock market.
So why have we have a society bothered doing that? They could achieve exactly the same effect by building in an artificial latency, and then levying a 'latency tax' to bring it down, the proceeds of which would be used for some socially beneficial purpose.
If they really must have automated agents doing stock trades like that, can we not find a more efficient way of doing this, perhaps using a central arbiter system and VMs running bytecode? Then we won't get people running transarctic cables down so they can get a slightly better ping time unless it's for something actually important to humans like voice communications latency or Quake.
This entry was originally posted at http://morwen.dreamwidth.org/387553.html. Please comment there using OpenID.