Venture Capital, Catastrophism, etc.

Here are the slides from my presentation yesterday in San Francisco:


  1. Any chance of getting the speaking notes? Some of these slides aren’t exactly intuitive :)

  2. Speaking notes? Speaking notes? I don’t need no steenking speaking notes!
    More seriously, went something like this:
    1) Title
    2) Spirit of this presentation
    3) Embed the discussion in current events. The current San Diego wildfires are a wildly anomalous event, a scant four years after equally anomalous events. Is the problem with our models, our knowledge, our history?
    3)- 5) Pictures from the fires
    6) A strange new thing is the first thing to happen in disasters: All the major related websites go down. In San Diego’s case, the site was up and down constantly for the first thirty hours.
    7) Screen shot of borked site.
    8) Local paper relocated to a blog
    9) Monitoring police/fire/CHP via web feed
    10) A fire wiki from SDSU
    11) Mashup of traffic data with road closures
    12) Twitter feed with continuous headlines
    13) Mashup of fire perimeters, refuge stations, and animal boarding sites
    14) User-driven weather stations gave better and faster data on changing wind conditions
    15) But step up a level. Why did this happen? Our inability to forecast extreme events, whether epochal wildfires, or traffic to sites during disasters?
    16) Part of the trouble is the nature of natural processes, and of infrequent events. Often tough to collect additional data without simply waiting.
    17)At the same time, some of the trouble is in dueling paradigms: catastrophism vs. uniformitarianism. The latter holds sway, and yet there is increasing evidence for the former. Catatastrophes matter more than we like to think.
    18) Consider the hedge fund meltdown this past August. A 16-sigma event shouldn’t have happened, not enough time since the earth cooled, and yet it did. Is the problem with the world, or with our models?
    19) In some sense, the trouble is that in far more cases than we like to think, we don’t know what normal is. We operate in cosmologically brief and quiescent periods, and extrapolate serenely in every direction.
    20) We lie to ourselves about the past. In case the wildfires, we convinced ourselves that current catalyclysms aren’t historically unusually, that they happened through history. There is growing evidence that the historical record is a fraud, or at least prone to wild exaggeration.
    21) All of this would give great comfort to Immanuel Velikovsky. The Austrian psychologist’s catastrophism theories, most known through his wildly controversial “Worlds in Collision”, were driven by Zionism and some ill-judged ideas about scientific acceptability, but they are provocative, ringing more true now than ever given recent events in cosmology, markets, and fire ecology.
    22) You can see something similar in capital markets, especially in private equity and venture capital. These are very young industries, with sorely inadequate data. We kid ourselves when we say we think we know what “normal” is: we have no idea.
    23) And, worse yet, there are few funds that matter, so the data problem is worse than mere history dredging would suggest.
    24) Further, we are losing funds at a frantic pace right now, so the data set is getting smaller yet.
    25) And the result is we have no idea if these are good returns, bad, or just a fluke.
    26) There is ample reason to expect that financial markets are constrained by some of the same limits to human performance that constrain us in sports and elsewhere.
    27) Tiger Woods’ historical accuracy/distance data.
    28) Pace Gould, the disappearance of the .400 hitter has to do with improving ability all around, not the disappearance of the Ted Williamses of yesteryear.
    29) The result is immense cognitive and performance blind spots, problems that lead to big misses.
    30) The fire regime in SoCal is fairly patterned, until recently, and then it goes through a jump change. Relying on historical data to predict current events would have led you to vastly underestimate the likelihood and location of the fires of 2003 and 2007 in San Diego County.
    31) The perhaps apocryphal story of Lancaster bombers in WWII and their armor plating.
    32) The lovely thing about errors and broken models. Strangely enough, us being badly wrong recently makes me wildly optimistic about the future.

  3. Doc McClenny says:

    31) The perhaps apocryphal story of Lancaster bombers in WWII and their armor plating.
    – which story? The only armor being the pilot’s seat? not entirely correct – the back of the engines were also armored.
    Brit’s made a different set of design tradeoffs than the Yanks – trading off increased speed and height for less armor protection.
    Each bomber force also ended up optimizing for their separate night and day time bombing roles.

  4. Those are some great notes — I think I almost like them better without the presentation :-)

  5. Hank Haney might disagree with your notes if you moved the date to year 2006 rather than 2000-2005. See the following article from the Golf Channel this week:

  6. John — In case I wasn’t clear, I actually argued that Woods is one of the more accurate drivers of the ball, at his distance. I’m way out in front of Haney on this one … literally.

  7. Sorry, but I missed that point from the Tiger Woods graph.

  8. Looks like someone’s been reading a lot of Nassim Taleb these days.