I just looked back at the April thread from last year and it confirmed my suspicion ! I was 14 places lower with a better cumulative error. The standard really was unbelievable last year
I think it is unlikely that the standard was much higher last year as the contestants this year are much the same people although as you say the accuracy (as opposed to "standard") was higher in 2016
IMO there are two other factors which govern the overall accuracy of the whole collection of estimates:
(1) The accuracy of the "official" Met Office forecast at the start of that month which we often use as a guide
(2) The deviation of that months CET from the long term average
I don't know about number (1) but this year Jan-Apr there have benn 2 months with a CET of more than 2C outside the average whereas last year there were none. Us contestants generally cluster our estimates around the average especially if the month is tricky to call so months with big deviations tend to catch out a a lot of people