Human Computation in Capital Markets & Elsewhere

Great post by Tim O’Reilly in a Money:Tech conference context about the never-ending race between humans and machines in markets. The former’s algorithms are constantly thwarted by the latter’s wilful misbehaviors, and so is there a solution in computation? This is a theme at Money:Tech, and it’s a vital subject.

Here is Bill Janeway of Warburg Pincus, as quoted by Tim:

…as driven by the web more generally, the frontier between human and machine-decision making has become radically problematic. First, quantitative approaches in trading, pricing, valuation, asset definition vastly expanded the domain for machine decision-making. But then the humans struck back, by refusing to act like the mindless molecules that the models driving machine decison-making required. The self-reflective, behavioral attributes of human market participants is now driving back that frontier, requiring innovations in every aspect of financial market processes, beginning with techniques of risk measurement and risk management. When price is an inverse function of liquidity and liquidity is an inverse function of price certainty, the recursive loop can only be broken by human intervention and action.

Bill neatly captures the tug-of-war between humans and machines in dynamic markets. Thought-provoking stuff, even if you think this particular genie has, you know, left the bottle and broken it.


  1. It’s not so much humans are misbehaving as it is “all models are wrong, but some are useful”.

  2. Bleh. He’s just doing a sleight of hand on the fact that the-map-is-not-the-territory, in order to pundit on Man Vs. Machine. Read one of these, you’ve read them all. They run roughly:
    “The worship of machines will rob us of our humanity. But the sheer unpredictability of the human condition can never be captured by the cold equations of an algorithm …”
    Machines vs. Machines is far more interesting anyway.

  3. There is the fundamental math/CS notion of “formal undecidability”, and that is what is at play here. The self-reflective nature of the problem rules out algorithms that will always compute correctly.
    A similar point is made in Wolfram’s NKS.

  4. Wolframs point, I believe, is precisely that “formal math” *cannot* model complex systems like financial markets.
    Even a cursory knowledge of systems like these will show you that they cannot be modeled by typical quantative methods. Yet, look at the types of people being hired for modeling financial systems and they seem to be almost exclusively from the formal math/statistics fields. Exactly the type of people that would create models that would invariably end in financial meltdown, oh wait…
    Statistics (or what I technically term “guessing”) is easy prey for smart human managers (or smart software for that matter), it does not adapt. The financial services companies need to destroy their monoculture and bring in new blood to evolve and adapt, or continue marching toward extinction. imho.

  5. I was re-reading my comment and thought it sounded eerily familiar, like deja-vu. It seems I wrote almost the same thing a while back, on this very blog. Guess this point seems to rub me the wrong way!
    Ok, I should add a bit of value(?), one of Wolfram’s [NKS] central points is that randomness and unpredictability occur at points of inflection (my interpretation). I think we are at such a point right now in the financial markets with the inflection being assisted by the shift from human to computerized modeling/trading.
    This shift wont last forever of course but an entrepreneur or smart fund manager can see this inflection point as presenting opportunities for radical market disruption.
    The inflection comes about through lack of momentum (or conversely inflections are avoided by sustained momentum, again my interpretation). The current liquidity crunch has curtailed recent market momentum and provides the context for a large inflection and short term chaos/opportunity, perhaps.
    btw, Wolframs thesis is stunning (although some parts are now dated) but requires you read pretty much his whole opus to digest. Highly recommended.