First of all, I must say that I’d be pretty damn impressive if a 93 year old man weighing 700 pounds could run a mile in negative time and lift half a ton. I’d argue that the mile thing could become possible if we ever figure out time travel though. It is pretty clear that there are dangers with extrapolation, but there are perks too…
I live with two guys and I am (unfortunately) quite messy and they are pretty messy and as a result, our apartment is constantly messy. However, as Vickers so awesomely highlights in his messy kitchen example, “the increase in reward per unit effort decrease with increasing effort.” Therefore it really doesn’t make sense to keep cleaning after the basics are complete because the returns for your efforts begin to plateau (and who wants to clean without a reward?). I am thankful that this relationship is not linear, and I’m glad I have statistical ammunition to back up my lack of cleaning efforts if the guys ever say anything (which they won’t).
So extrapolation appears to occur when we improperly fit our data and we inappropriately look beyond the data that we have. So why do people do this? I can’t really answer that question, but it is interesting to think about. I mean, say for example the scientists involved in projecting climate change for the IPCC chose to violate these rules (if any of you remember climate gate from 2009 then you know accusations of this did exist). If the accusations were proven true, this violation would kill the credibility of science and also make the public even more skeptical of something that is so direly important.
How are we going to change environmental policy if we violate these rules? What is the perk in data misrepresentation? In inappropriately representing and extrapolating your data, you are likely creating a much more dramatic (maybe interesting) story, but it isn’t true and you lose credibility, even your field could lose credibility once that is unveiled. Yeah it may be more difficult to lobby for carbon sequestration efforts if you didn’t have a hockey stick curve (and I am not a climate gate believer), but you should let you data speak for itself and your interpretation speak for what steps should be taken after the fact. So what did we learn from climate gate? Well for starters, hackers suck. Be careful what you say, but in the end there will still be people who will see your words out of context and draw false accusations. Lastly, while this incident didn’t set the best tone for the climate summit in Copenhagen, one really positive thing did come out of it in that there was new value in data sharing and data openness.
After the climate gate investigations concluded that anthropogenic induced climate change really did exist and the science was sound and honestly conducted, I doubt that the climate change skeptics were ready to even be in the path of the bandwagon nor that politicians were eager to push pro-environmental legislature. But how do you fix a misunderstanding and regain public trust after something like this? From what I’ve read, scientists looked for better outlets of making their data publicly available or available on request, and made methods more accessible.
So, I think there is a lesson to be learned from improper data usage like extrapolation, or even accusations of data misconduct. The more open you are with your methods and data, the less likely your email will get hacked and the more likely you are to help science progress and cover you’re a$$ in the process.