Another month of data...in fact two months for the ENSO statistics:

These 3m centered means lag reality by quite a long way - the current index (27 Sept) is about -1.5. The red line should track the blue one quite closely for the rest of the year, I think. And here is the updated temperature, which is still rising (in terms of 12m lagged average) but not by much:

Here I'm using the
3v version of the data (which I prefer), the
non-v version also has August data in and these are clearly cooler, so things are likely to head downhill from here.
Chris R suggested that it might have been better to show year-to-date values rather than 12m lagged average. I chose the latter partly due to Hansen's discussion (pdf
here) where he argues that 12m rolling averages are more scientifically relevant, as there is nothing special about the year end that justifies focussing on this calendar period to the exclusion of (say) May-April. I suspect this issue only piqued his interest due to the fact that the 1998 calendar year temp was also so close to the max of any 12m period (as you can see from the above pic), so the calendar year 1998 shows up as particularly hot on the standard annual (calendar year) data plot.