Nancy, I don’t mean to jump on you specifically here, but this does seem to me to be a special instance of a general online forum disease, where people {prefer to use | view as authoritative} online sources of information (blogs, wikipedia, even tvtropes, etc.) vs mainstream sources (books, academic papers, professionals). Vinge calls it “the net of a million lies” for a reason!
The common example I go on about is any situation where a system generally succeeds at achieving a goal. This is a very large class. In such situations there will tend to be an absence of correlation between the effort made and the success at achieving it. The effort will correlate instead with the difficulties in the way. Effort and difficulty together cause the result; result and goal together cause effort.
A few concrete examples. If my central heating system works properly and I am willing to spend what it takes to keep warm, the indoor temperature of my house will be independent of both fuel consumption and external temperature, although it is caused by them.
If a government’s actions in support of some policy target are actually effective, there may appear to be little correlation between actions and outcome, creating the appearance that their actions are irrelevant.
An automatic pilot will keep an aircraft at a constant heading, speed, and altitude. Movements of the flight controls will closely respond to external air currents, even if those currents are not being sensed. Neither need correlate with such variations as remain in the trajectory of the plane, although these are caused by the flight controls and the external conditions.
Yes, this is completely wrong. There is frequently no correlation but strong causation due to effect cancellation (homeostasis, etc.)
Here’s a recent paper making this point in the context of mediation analysis in social science (I could post many more):
http://www.quantpsy.org/pubs/rucker_preacher_tormala_petty_2011.pdf
Nancy, I don’t mean to jump on you specifically here, but this does seem to me to be a special instance of a general online forum disease, where people {prefer to use | view as authoritative} online sources of information (blogs, wikipedia, even tvtropes, etc.) vs mainstream sources (books, academic papers, professionals). Vinge calls it “the net of a million lies” for a reason!
I didn’t feel jumped on, though I still don’t have a feeling for how common causation without corelation is.
The common example I go on about is any situation where a system generally succeeds at achieving a goal. This is a very large class. In such situations there will tend to be an absence of correlation between the effort made and the success at achieving it. The effort will correlate instead with the difficulties in the way. Effort and difficulty together cause the result; result and goal together cause effort.
A few concrete examples. If my central heating system works properly and I am willing to spend what it takes to keep warm, the indoor temperature of my house will be independent of both fuel consumption and external temperature, although it is caused by them.
If a government’s actions in support of some policy target are actually effective, there may appear to be little correlation between actions and outcome, creating the appearance that their actions are irrelevant.
An automatic pilot will keep an aircraft at a constant heading, speed, and altitude. Movements of the flight controls will closely respond to external air currents, even if those currents are not being sensed. Neither need correlate with such variations as remain in the trajectory of the plane, although these are caused by the flight controls and the external conditions.
“The carpets are so clean, we don’t need janitors!”
“When you do things right, people won’t be sure you’ve done anything at all.”