Map of Syria with Aleppo highlighted (Photo credit: Wikipedia)
We have seven new IARPA questions on the market! The topics are:
- Israeli elections date
- Aleppo‘s fate in the Syrian conflict (ordered/multi-choice)
- The number of Syrian conflict refugees reported by the UNHCR
- Kuwaiti parliamentary elections date
- Durability of the UK Coalition government
- Korean bilateral talks (CONDITIONAL — Q.217 & Q.272)
Conditional questions are formulated as “If A happens by this date, then…” and “If A does not happen by this date, then…”. Therefore these questions come in pairs and add one more layer of judgement and interpretation to forecasting.
Note that refugee distributions can be power-law distributed. This iRevolution blog article looks at a FlowMinder study on predicting refugee flow from disasters, and may have some useful ideas.
We are looking forward for your participation and comments! Remember, these are IARPA questions so we’re scored on them.
Related articles
- Syrian refugees near 150000 as exodus grows: UN – Reuters (reuters.com)
- Syrian Soldiers Battle Rebel Forces as U.S. Warns of Terrorism (bloomberg.com)
- Aleppo cleared of terrorists (english.ruvr.ru)
- Aleppo: Conflicting stories and the truth (english.pravda.ru)
- Syrian Fight Uproots 1.5 Million as Iran Urges Cease-Fire (bloomberg.com)
- Syrian refugees take to begging on Hamra street (dailystar.com.lb)
- New questions on the market! (daggre.org)
Good questions. I did notice what looks like an extra option on the Aleppo question #267: “31 October 2012 , (d) Event will not occur before 1 November 2012″.
Sorry for the delay, this comment slipped by me.
Argh. I thought I fixed that when creating the question, but it made it through to the user side. I’ve “fixed” it now by naming Option #4 “[ INVALID ]” and driving its probability to 1%.
If everyone refrains from editing that option directly, it will head ever further to 0%, which is what we want.
Unfortunately, the software does not support cleanly resolving part of a question. (Solutions that work here wouldn’t work when we go combinatorial, so we’re stuck with this method for now.)
Thanks for the heads-up.