I read your comment on Open Mind/Teachable Moment thread. I posted the following in reply, but with the inconsistent editing of my posts on that site, I now repeat my posts here. I do hope you read it and understand where I’m coming from. Here is my complete post as submitted.
Search the original article for the word acceleration. Tamino states emphatically that the data shows acceleration at the end. He bases this on the cubic being the best fit (via AIC scores) to the data and that the cubic accelerates at the endpoint. I say that while the cubic is the best fit via the AIC method of ranking, the claim that the data shows acceleration is not accurate and that the shown acceleration in the cubic fit is due to the nature of cubic fits.
To show this, I looked at the data from 1928-2000. I used Tamino’s method to determine the best fit via AIC scoring – it was a cubic. The cubic fit for 1928-2000 shows an acceleration toward the end of the series just like the 1928-2013 cubic curve fit does. At this point, using the full data set, Tamino declares emphatically that there is acceleration in the data and that using a linear fit is foolish because it doesn’t match the data as well and it doesn’t show the acceleration. Therefore, it’s logically consistent that the same claim can be made if we had looked at the data back in 2000.
I then compared the 1928-2000 cubic fit to a linear fit and as we extrapolated both lines out to 2013. The lines diverge as one would expect. Then I overlaid the data from Sewell Point on top of the two curve fits – think of this as waiting around and gathering data for 13 years. Which curve fit agreed with the “newly gathered” data better? The linear! This clearly shows that any acceleration seen in the cubic model at the end of the series (even without extrapolation) is a function of the model and not necessarily a true physical acceleration.
As for the 1960 timeframe mentioned above, Bernd asked (Comment 86654) what happened if you look at the data from 1928-1960. The original trendline shows a decreasing acceleration and he asked if fitting a cubic to this short timeframe would pick up the decreased acceleration. I said, as shown above, that it wouldn’t. Tamino then lambasts me for not finding the best fit – but I wasn’t looking for the best fit, i was answering Bernd’s question!
That said, I doubt Tamino will publish this as he’s not published about half of what I have written – including a benign comment on how engineers think – but allows a post insinuating all engineers are creationists. Not publishing that post is why I have been copying my submissions here to my own blog.