It is interesting to see the coverage of things like this. The vast majority of financial market activity is now algorithmic trading... exact numbers across markets are a fiction, but believable numbers for the equity markets put it around 80% or more. The fact of the matter is that any time continuous activity -- or where the time increments are faster than "human speed" will naturally be dominated over time by different "bots"with differing levels of autonomy, "intelligence" and learning capabilities.
The interesting thing about the coverage is that the problem is always in the interaction of the algorithms. So when a new algorithm is released into the market or a unique set of conditions occurs, one doesn't know how the algorithms which are the market will interact -- just that they will interact at faster than human speed so that periodic flash crashes in markets will happen. And it isn't a problem which can be really be solved without fundamentally changing the market structure, which makes the regulatory discussion interesting -- or any discussion which pretends to address "flash crashes" without addressing the inherent causes of continuous trading and faster than human speed algorithms.
The answer to this problem is just that algos that lose money will fade away...
One systemic question is what if any governance structure should exist for these machines -- and as they become more and more intelligent? What currencies will they use for their faster than human speed transactions. Put differently -- when will the bots get "frustrated" with the current state of the global financial system?
A key driver of change in the global financial system is likely to be from those inside the system seeking a more efficient platform for their bots to operate on. This may be part of what is happening now with all the investment and interest in bitcoin/blockchain from the corporate world - but corporations seem to be very interested in private blockchains, which just makes me smile since it is an explicit implicit recognition of the lack of trust that people have for each other even within corporations.
Just one of the factors to consider now as "machine learning" and "AI" become cloud services:
[url]https://www.tensorflow.org/][http://alchemy.cs.washington.edu//url]
[url]https://www.tensorflow.org/https://aws.amazon.com/machine-learning/There are probably a variety of services which it might make sense to see implemented in a distributed, decentralized, trustless, ownerless platform over time.
Being able to do this sort of thing seems to be one of the goals at Crown, but that is many steps down the road.Just putting the idea out there and connecting it up with a few other things. If this sort of goal sounds interesting -- reach out to Chaositec or Stonehedge -- they are the ones with the technical plan. I'm just an unfrozen caveman from a time before cell phones and intertubes and don't understand many of these modern things...
19 left!
- Or I can point you at the repo if you can build yourself. 