Copernicus has September as hottest on record. It holds the record by 0.05 C, the same as May.
35% is far, far too high - try 1%
Fit a line to the last 10 years of RyanBeck's data and treat errors as normally distributed.
Compare to this range question, where the community gives an 87% chance that any of the top 25 will file by Mar 2024.
Note that fires 6, 7, and 8 were very similarly destructive and this is most likely due to chance, not some magic of that number of buildings being destroyed. Hence a new fire is quite unlikely to stop at places 7 or 8. Quantitatively, fire 7 was only 1.8% more destructive than fire 8, while fire 5 was 18% more destructive than fire 6.
The same goes for fires 3 and 4.
Fit a logistic with Jun 13 as midpoint.
Looks like I can't collect tachyons (yes, I'm below 50).
Community seems too high and narrow. Since 1919 the average growth has been 0.24% and since 2007 it's been 0%, with slow declines at times that don't look like recessions.
At the same time, the standard deviation since the report on Sep 15, 2020 (chosen to postdate covid volatility) is 0.97%, double the community estimate.
I just input the cdf for the 20 historical maxima, adjusted for inflation.
Quadratic fit to SimonM's numbers shows 2020 is a huge outlier. Annual increase is projected to be only 0.078. With a standard error of 0.108 I'm predicting 78.34 - 78.56.
Taking @RyanBeck 's sheet and filling in the last 2 months of data, I get a mean of 4411 and sd of 77.
In 2017 solar was about 9% of all renewable power, according to wikipedia. https://en.wikipedia.org/wiki/Renewable_energy
However, solar grows by about 40% per year - much faster than the 2.5-3% growth of renewable on a whole. That means we can expect 17% of renewable to be solar in 2019 and 28% in 2021. I expect this to push renewable growth way over 3%.
— edited by PepeS
The current sum is 813,230 and the sum for the last 3.5 months (since data is only available up to Sep 11) is 102,528.
Public figure predictions are incorrectly matching a number of search queries. E.g. either
not_guessed_by seem to match all public predictions, even if the corresponding Metaculus question doesn't match.