FORECASTING COMMENTS SUMMARY:

Overall it seems that Afghanistan will dominate the public conversation but it is unclear if it will just fade away or create more tensions in the region. Seems like China will slowly take over the conversation.
Update 29 Septiembre: Seems like there are now more mentions to China, but not much new.

Particular comments:
- @vpsays believes that "Also anticipate conflicts in SCS" [South China Sea] (https://www.cset-foretell.com/comments/21105)
- @MullenAustin mentions that "I do expect the GPR index to tick up at least temporarily. Not sure if the uptick will be sustained for the rest of the period though." (https://www.cset-foretell.com/comments/21144)
- @AABoyles notes that he believes this things change somewhat smoothly, but doesn't know where it will be headed. https://www.cset-foretell.com/comments/21165
- @cafebedouin provides some projected mean difference model models based on data since the 80s (https://www.cset-foretell.com/comments/21280)
- @Wooster has a nice comment (https://www.cset-foretell.com/comments/21284) directing the possible causes of increase for each of the 6 word groups.
- @JJMLP mentions (as others do) that "The drop of the GPR index could be linked to the covid situation" (https://www.cset-foretell.com/comments/21386)
- @shaun-ee mentions that the late Biden presidency will mostly be dominated by China-US tensions (https://www.cset-foretell.com/comments/21935)
- @RyanBeck (https://www.cset-foretell.com/comments/22692) gives three main points.The first is that perception over time may have changed. The second is that the decrease during Trump administration was due to Covid. Finally "the index hasn't really seen a sustained increase over 200 for long"
- @Loki (https://www.cset-foretell.com/comments/23673) wonders how the perception has changed over time and what are the probabilities of new high impact events (mentions war with China)

username-deleted 1575
made a comment:

I'm generating the probability bands by creating a projected mean percentage difference for each half based on the historical average over the forecast period, projecting standard deviation based on this difference and then using the value of the projected mean and projected standard deviation to draw a standard distribution curve and taking the points for 10% and 90% from that distribution. So, no Monte Carlo involved on this one (and I may, or may not, know what I'm doing (looks right though)).

Tip: Mention someone by typing @username