Posts Tagged ‘prediction’

IACCM, Regression to the mean, and other topics of the week

May 17, 2013

NASA Sunspot Number Predictions for Solar cycl...

  • The humble founder of TPS planet – Captain TPS?…*running to the shower* – will be making an appearance on the, in this case inappropriately named, ‘Ask the Expert’ IACCM webinar series on June 6. I’ll be blathering on about prediction markets, crowdsourcing, and how they can be applied to procurement and the supply chain. You can register here. You have to be a member, but if you sign up close enough to the event, you can attend this one for free. Log in, listen, add your own thoughts, heckle me…just make sure you are there. You’ll either learn something, or watch me publicly embarrass myself, but either way, it’s a win for you.
  • While I’m on the topic, friend of the blog (and author) and CEO of IACCM, Tim Cummins writes a very informative blog on contracting that can be found here.
  • Least surprising, yet concerning, news of the week; the IRS scandal. So you mean that people with a substantial amount of power over other people’s lives have used that in a power in a biased way? Someone get Kahneman on the phone…
  • ‘The Un’ (of the Kim Family) has either gone quiet or the world’s news media has moved on to something more interesting. I wouldn’t be totally shocked to learn that his dad left him a script…’During the annual South Korean and USA joint military exercises, you shall take the following steps…’
  • Key quote from Thinking, Fast and Slow ‘We will not learn to understand regression [to the mean] from experience.’ (My note: among many other areas in which creative thought/problem solving/rationality is required). To put the quote in context, Kahneman is talking about the fact that most people base predictions of the future on a relatively small sample of readily available evidence without accounting for the fact that the correlation between that small piece of evidence and predictions of an uncertain future are less than 100%. We then take that evidence, and develop causal interpretations that are completely inaccurate.
  • Regarding job performance, I have witnessed this phenomenon more times than I care to remember. A couple random (or worse, influenced by selection bias) comments are taken as a proxy for overall performance (both good and bad). I would even go as far as to say that our visceral feelings about someone then bias overconfident people even further to selectively look for evidence of strong or weak performance. The good news is, if you are in a business environment, you can focus on looking good when an overconfident exec is watching and don’t worry much about the day to day (unless it results in a problem that then gains attention that can be pinned on you). The bad news? Well, if you do a great job everyday, but a random poor performance happens to be noted by someone influential, then you have a steep hill to climb. Ridiculous? Uh-huh.
  • Re-reading the stuff you have written in the past is always scary, but I went ahead and gave it a shot. I did not make the connection at the time between my article in The Journal of Prediction Markets and behavioral economics, but I now believe there is a link.  Our tendency is to predict more extreme outcomes than are truly likely based on our lack of understanding of regression to the mean (see discussion above). Prediction markets remove those errors by combining them with the errors of lots of other people, thus mitigating the extremeness of individual predictions. (If you can tell me what assumption I’m making here in the comments, you will win a free annual subscription to the TPS Report.)  Still think this is not having an effect on the information flow in your supply chain?

Advertisements

Facebook and the past’s past

January 23, 2011

The past has a past. Ponder that for a second and then continue reading. In spite of the conscious knowledge that conditions and circumstances change in ways that we never could have predicted prior to a major shift, there are those that insist on making linear projections of the current state of things into the future.

Actor Jimmy Stewart

Is it a wonderful life for Facebook?

The latest example that I’ve come across is James B. Stewart who writes for SmartMoney magazine (I’d love to know his stock picking track record to see just how smart his money is) and has made a celebrity guest appearance in the Wall Street Journal. His article is here.

The basic idea is that Facebook is worth the $50 billion value that the current level of equity investment implies and more. While I won’t pretend to have any idea whether or not this is the proper value, Stewart’s reasoning is a bit flawed. The main thrust of his argument is that since it has achieved dominance in terms of the amount of users, it has a kind of natural monopoly based on its ability to exploit this data to adapt to the needs of users better than anyone else could. He cites Google as an example, which is currently under siege from competitors such as Bing.

There are a few assumptions that I believe he is making that open up the article for criticism. One is that Facebook will continue to make all the right moves to keep their customers satisfied based on all the “information” that Facebook can collect. If there is anything that I have learned about the human mind is that more information does not necessarily lead to better decisions. We can’t process all this information and neither can the management team as a collective whole. Zuckerberg’s belief system will enter into the equation on major strategic decisions, potentially going in the wrong direction. Past good decisions may have been lucky, or may have worked in the environment in which they were made…either way, the hot streak could end on any big decision.

He assumes that there will be no major paradigm change (I must be intelligent…I used ‘paradigm’ in place of ‘model’!) in the way we interact online. What about the potential for second life to take over? Something else not yet invented? Many possibilities here. In a world of continuing decentralization of content, any platform to deliver that content is subject to abrupt obsolescence.


%d bloggers like this: