mollygh

@johnnycaffeine on base rates: https://www.infer-pub.com/comments/97840

OTOH @belikewater notes situation seems to have deteriorated more recently, including US embassy in Haiti urging US citizens to leave Haiti asap: https://www.infer-pub.com/comments/97210

Ultimately coming up a little bit.

July 7 marked two years since Moïse's assassination [1].

Old (February) news: "[Henry] has pledged to leave office by Feb. 7, 2024, after postponing elections indefinitely in the wake of a devastating earthquake that affected the Southern Peninsula, along with gang violence." https://www.reuters.com/world/americas/haitian-pm-installs-transition-council-prepare-long-awaited-elections-2023-02-07/

I doubt that there will actually be elections before Feb. 7, 2024. But "vigilante justice" seems to be working to stabilize Port-au-Prince and quell gang violence [2]. This makes me think an irregular transition is less likely than I thought previous.

[1] https://www.voanews.com/a/haiti-marks-2nd-anniversary-of-moise-s-assassination-/7172302.html

[2] https://www.nytimes.com/2023/06/03/world/americas/haiti-crime-gangs-vigilantes.html

After only an hour of reading, here are some numbers I've come up with. These are very tenuous -- I bet they'll move a lot as I learn more:

Ways that Ariel Henry could cease to be president (or acting president) through an irregular transition ("yes" resolution):

• (A1) Coup: 15%
• (A2) Assassination: 5%
• (A3) Ambiguous / U.S. govt sours on Henry: 8%
• (A4) Gangs form quasi-state? (It seems like gangs "control" ~60% of the country right now? Fuzzy on what this would mean, but video at [1] mentioned it): 10%

(A3) and (A4) aren't mutually exclusive.

Ways that Henry could stay president OR leave office in a normal way ("no" resolution):

• (B1) Election (widely accepted as legitimate): 3%
• (B2) Resignation: 2%
• (B3) ~No change from what's happening currently, i.e. still president: 25%

That leaves ~30% for outcomes I haven't thought of!

----

The Fragile [read: Failed] State Index from Fund for Peace ranks Haiti 11th, with Myanmar on one side (10th) and Guinea on the other. Its score has gotten higher (worse) for three years in a row. [2]

[1] https://www.cfr.org/backgrounder/haitis-troubled-path-development

[2] https://en.wikipedia.org/wiki/List_of_countries_by_Fragile_States_Index

"The economic sanctions have been extended successively for six months at a time since July 2016." [1]  c/o @sebawi

They renewed on July 26 and January 13 [2] this year. I suppose that they'll do the same ~January 13 this year. It seems there's some question of whether Hungary will stop the renewal (see @A19 rationale).

I suppose sanctions could be left to expire because of something silly and bureaucratic -- they might let it expire and then renew it, like, in February, and that would still result in a "yes" resolution.

[1] Economic sanctions: https://www.consilium.europa.eu/en/policies/sanctions/restrictive-measures-against-russia-over-ukraine/#economic

[2] Timeline: https://www.consilium.europa.eu/en/policies/sanctions/restrictive-measures-against-russia-over-ukraine/history-restrictive-measures-against-russia-over-ukraine/

There are September data now (thanks again @RyanBeck for making your code public): $4.77 billion for Jan through Sep 2022, 10.9% less than Jan-Sep 2021. If that just were to keep up, there would be ~$6.46 billion (2021 was $7.25 billion). I think things have gotten worse for China since September, but third bin still looks pretty good. 1:2 odds of landing below $6 billion seems right to me.

I wish I'd quit this question in September (comment) but since it's almost over...

Right now it's 4.917395411 commits (somehow - not sure how you have fractions of commits). Percentage is ~10.7%. Uncheck "Estimate 2022 values" and download as CSV. In 2021 EU27 had 23.02 commits (10.8%).

> Using the 439 topic labels identified by Gonzalez et al. – as well as the topics “machine learning”, “deep learning”, and “artificial intelligence” – GitHub provides with a list of public projects containing AI code. GitHub updates the list of public AI projects on a quarterly basis, which allows to capture trends in AI software development over time. (https://oecd.ai/en/github)

I suspect that's part of why all the numbers for 2022 are fairly low right now; they won't have gotten the updated list of projects and forks from this quarter.

I think this question is not getting at the thing we want to get at. I think the topic labels they use are outdated (Gonzalez et al paper they cite was published in 2020). I don't believe it's the case that everyone peaked in 2017.