The variable on the vertical axis -- calculations per second per $1,000 -- is a clever one (assuming they've adjusted for inflation). It's not totally obvious to me that the data points create a curve. It could easily be a line, but the editors clearly wanted to project a curve. Regardless, it appears that computing power is increasing exponentially, and we will soon approach The Singularity, whatever the hell that is.
I'm often surprised just how bad our forecasts for technology are. I recently watched "Back to the Future II," where Marty visits the distant future of October 2015. It's pretty funny, really. It was a world of flying cars and hoverboards, but no cell phones or Internet, as far as we could see. At one point, Marty Jr. actually walks past a payphone! So I don't know whether the next fifty years will give rise to a Singularity or Skynet or what, but as long as I can someday run probit on my smartphone, I'll be happy.
Good thoughts on forecasting the future from Patton Oswalt:
| Jokes.com | ||||
| DVD - Exclusive Patton Oswalt - The Year 2009 | ||||
| ||||
2 comments:
I believe the relationship in the graph is curvilinear because the Yaxis scale is logarithmic...
Yes, the scale is logarithmic, but that's no reason that the data points in the graph need to follow a curve. Indeed, researchers often use logarithmic scales in graphs to straighten out curves.
Post a Comment