Talk:1124: Law of Drama

Explain xkcd: It's 'cause you're dumb.
Revision as of 17:44, 22 October 2012 by (talk)
Jump to: navigation, search

Regarding the transcript: I don't think you have enough data to characterize this short curve as exponential. What does "slightly exponential" mean, anyway? In any case, it looks like it becomes linear as the x values increase. --Prooffreader (talk) 11:21, 22 October 2012 (UTC)

It could be a shallow power function curve . . .--Joehammer79 (talk) 13:57, 22 October 2012 (UTC)
I took 26 data points, assumed the axes defined a (0-1,0-1) window, and tried an extrapolation (using Microsoft Excel; someone with a different tool can surely do better). An exponential model fits fairly well: y = 0.0782 * e^(2.7035*x) with R^2 = 0.9928. However, I agree about the linear end section -- the exponential trendline clearly starts to pull high. --BigMal27 // 13:57, 22 October 2012 (UTC)

I think Randall thought about the shape of this curve. You see how it becomes linear as both drama and anti-drama declaration increase? At low values, there is a residual amount of drama even when there is little anti-drama declaration, but the marginal increase eventually becomes constant. --Prooffreader (talk) 11:28, 22 October 2012 (UTC)

I think that may be it. Care to add it to the page? Davidy22 (talk) 11:31, 22 October 2012 (UTC)
I think that the upper limit for drama statements does indeed have an end-point, beyond which those declarations can't increase. At that point, I suppose, the drama-ridden person experiences a split state-change, either dropping to the original non-drama state by disavowing all the causers-of-drama in their lives, or by becoming a causer-of-drama.--Noni Mausa (talk) 13:11, 22 October 2012 (UTC)
At this point in the discourse, I'm reminded of a real-scientist friend who admonished me once for reading too much into some data, and it seems applicable here, too. To wit: the axes are not labeled with units -- no tick marks to be seen anywhere -- nor is it clear what sort of axes are in use: log, logit, probit? Randall, not being the naïve sort, likely understands this, and merely shows us a graph that suggest a slightly accelerating direct relationship between the two axes. If the axes are linear, the curve has the characteristic upward swing of an exponential, but we don't know that, and any conjecture beyond observable facts is inappropriate. To leap to application of, say Levenberg-Marquardt, seems folly. (As an aside, I'm reminded of the old Benny Hill skit, where he's a movie director being interviewed on some talking-heads show; says the interviewer: "I particularly enjoyed the poignancy of suddenly switching to black and white film right as..." Benny Hill: "Rubbish, we just ran out of film, and black and white was all we had left.") -- IronyChef (talk) 14:23, 22 October 2012 (UTC)