¶ 13 January 2005
I understand that forecasting the weather is difficult, and I hold no human being culpable for the fact that it was 35F today when the forecast had insisted it was going to be 55F. I assume there were clear, tangible indicators that suggested it was likely to be 55F today, perhaps even very likely, but something less-likely happened, and in the end it wasn't. The improbable will periodically occur.
But what I would hold someone responsible for, if I thought it weren't a pervasive cultural flaw, is the destructive precision with which uncertain predictions are communicated. Weather is merely the most obvious daily public manifestation of a fundamental reluctance, or perhaps an inability, to say what we really know, rather than what we wish we knew. We'd like to know whether it's going to rain, and how cold it's going to be. What we know, however, is not these things, it's what's on the radar and how it's been moving and what our computer models can extrapolate given the data we know how to supply. We know what stock prices have been, and how many we sold last year, and how long other projects that seem like they will turn out to have been similar took.
I brought the clothes I'd need for running in 55F, but since it was 35F when I left home, I also brought some others. If I'd gone out and felt too cold, I'd have come back inside. The forecasting failure of is little direct practical consequence, but that isn't an excuse for a forecasting grammar that obscures the essential natures of the activity and its results. "Today's high: 55F" is nonsense. What we should really see is a graph of the last 24 hours and the next 24, maybe, of past predictions and future ones as probability ranges, with a trendline of the actual temperature running through the past. Ditto for precipitation: past predictions, future guesses, the measurement history. This, after all, is what we actually know for sure: what we think, what we thought, what was.
And if we learned to talk about the weather, maybe it would help us understand how to talk about other systems. Maybe a daily reminder of the limits of certainty in one natural system patently out of our control would encourage us to acknowledge the limits of certainty in the human systems over which we exert only slightly more influence. Maybe we'd be a tiny bit less likely to manage by oversimplified dashboards and spurious charts and uncalibratable figures. Maybe we'd build communication tools better suited to representing hard-won understanding of vital complexity, rather than discarding it in favor of quantification of our wishful ignorance.
But what I would hold someone responsible for, if I thought it weren't a pervasive cultural flaw, is the destructive precision with which uncertain predictions are communicated. Weather is merely the most obvious daily public manifestation of a fundamental reluctance, or perhaps an inability, to say what we really know, rather than what we wish we knew. We'd like to know whether it's going to rain, and how cold it's going to be. What we know, however, is not these things, it's what's on the radar and how it's been moving and what our computer models can extrapolate given the data we know how to supply. We know what stock prices have been, and how many we sold last year, and how long other projects that seem like they will turn out to have been similar took.
I brought the clothes I'd need for running in 55F, but since it was 35F when I left home, I also brought some others. If I'd gone out and felt too cold, I'd have come back inside. The forecasting failure of is little direct practical consequence, but that isn't an excuse for a forecasting grammar that obscures the essential natures of the activity and its results. "Today's high: 55F" is nonsense. What we should really see is a graph of the last 24 hours and the next 24, maybe, of past predictions and future ones as probability ranges, with a trendline of the actual temperature running through the past. Ditto for precipitation: past predictions, future guesses, the measurement history. This, after all, is what we actually know for sure: what we think, what we thought, what was.
And if we learned to talk about the weather, maybe it would help us understand how to talk about other systems. Maybe a daily reminder of the limits of certainty in one natural system patently out of our control would encourage us to acknowledge the limits of certainty in the human systems over which we exert only slightly more influence. Maybe we'd be a tiny bit less likely to manage by oversimplified dashboards and spurious charts and uncalibratable figures. Maybe we'd build communication tools better suited to representing hard-won understanding of vital complexity, rather than discarding it in favor of quantification of our wishful ignorance.