Tuesday, May 04, 2010

Why Predictions Fail

Rob Cosgrave has a useful post on his blog, Tertiary21, on why predictions fail. He cites six reasons why predictions of the evolution of (higher) education systems fail:
  1. Failure to account for economics as a key driver, rather than technology.
  2. Failure to consider human factors and rates of change.
  3. Predicting out of area of expertise
  4. Failure to account for changes out of area of expertise
  5. Wishful thinking.
  6. Predicting the Weather, not the Climate
I would add "anchoring", the tendency to be too conservative in projection of radical changes from the status quo. I don't know how many people predicted the growth of a number of very large open universities, but I suspect they were very few. Comparably, I suspect few people predicted much in advance that MIT would put its curricula freely available online.


Robert Cosgrave said...

A Good point.
I'm currently reflecting on that, while reading Talib's 'Black Swans' book, which addresses the topic of unexpected changes driven by processes out of the scope of the original predictions.
Currently, I think there is a system inertia which needs to be considered. Volative systems, like localised weather, stock markets and Lebanese politics, can change violently for no 'good' reason. They are chaotic.
High inertia systems, like demographics, geopolitics and (usually) climate, tend to change slowly, and are much less susceptible to out of scope 'noise', loud as it may be. They are not so chaotic, 'sudden' shifts may take a century to unfold. For Consider that in terms of the overall global trajectory of economic and population growth in the 20th century, the two world wars could arguably be dismissed as 'noise' irrelevant to the overall trend.
The trick is to pick out which trends are noisy, and which have high inertia. In the long run, noisy trends won't matter as much, I think.

John Daly said...

I agree, but it seems to me that ones view of inertia depends on the time scale of the prediction. Birth and death rates change slowly if one is thinking of predictions a year or two in advance, but they changed quite a bit over the last century -- more than Malthus would have predicted.

I think educators who are not expert in technology have in the past failed to predict the rate of technological change and the impact that technological change would have on educational practice.

People also sometimes believe change will be dominated by one set of institutions and find later that another set actually predominated. I looked at one project in a developing country that assumed computers would be introduced in schools primarily as a top-down planned process, while in many countries the dominant process has been "viral", advanced by local action from the grass roots upward, and driven by outside forces such as commercial development of courseware or penetration of the technology into the home and family.

John Daly said...

This is an interesting power point presentation on decision making and biases from Maurice R. Masliah course on "Cognitive Ergonomics":