I just found this on Darren Kuropatwa’s blog who got it from Gary Hayes. And before I turn something like this loose on my students, I figured I ought to see what you’d do. Obviously, we could do something with rates, but what else is here? 
Click for counter I would have embedded the thing, but I can’t figure out how to do it in WP. Got it on my class wiki though.
David Cox
Something about this seems wrong to me. I took screenshots at one second, one day, one week, one month, and one year, and ran them through Excel. I’ve gotta dig in deeper but it seems to me this guy has essentially chosen a linear or exponential model for each of these variables and set that model recklessly in motion.
I guess my preference would be to use the same statistics he does and have students build the models, get the values at a year, and discuss the error inherent to our statistical overreach.
Rather than treating his statistical overreach like best practice and working backwards to those statistics. At this point I can’t justify it past “just a feeling, really.”
I was thinking about the validity of the stats as well. But I wonder if it really matters. I mean, the lesson could start with figuring out what model is being used, ie. linear, exponential, etc., coming up with a line of best fit and/or equation and then deciding if the models work. I don’t want to tell them that it’s overreach, I want them to decide that it is.
If not, then we could use the same stats and build the model ourselves and decide if extending the model beyond tomorrow really makes sense.