Hacker News Reader

viceconsoleon The US stops sharing air quality data from embassies worldwide

I worked in a US consulate in a country with bad air pollution and developed asthma for the first time in my life during my two year tour there. I've since left that country (and the job) but still suffer the effects of the exposure. I used both the State Department provided AQI numbers as well as the host country's numbers to plan when and whether to spend time outside.

None of the reporting I've seen on the issue has investigated the cost of the program. On the subreddit for State Department Foreign Service Officers, which I still frequent, the most plausible estimates for the installation of this equipment at a single foreign post range from $180k to $250k, plus ongoing maintenance that would require flying in private contractors from the U.S. and putting them up in a 5 star hotel (in countries where you need this type of monitoring, hotels less than 5 stars generally don't meet Western standards). [1]

There has to be some kind of middle path between take a chainsaw to anything that has the word "environment" in it, and spending the cost of four years at a private college to install equipment that is available off the shelf for a few hundred dollars. (Yes, I know consumer grade equipment won't cut it, there are major network security concerns, etc., but surely it could be done at 10x or 20x the cost of a consumer solution, not 500x).

[1] https://www.reddit.com/r/foreignservice/comments/1j3owmk/com...

viceconsoleon Mathematical Optimization for Cargo Ships

I work in a restricted environment where I can't install or-tools. In order to solve our scheduling problem, I wrote a solution in Prolog that runs in a purely web-based "notebook" environment. It works well and has the benefit that non-technical users only have to deal with specifying simple "facts", then running a pre-defined query.

I also have access to Google's OR API as a trusted tester and have been toying with the solveShiftScheduling endpoint. Out of the box, I don't think I'll be able to easily represent some of our constraints to work with the API. Also, our management frequently want to change the scheduling behavior, so while I can reformulate the problem to make it work with the API now, I never know what's coming down the pipe.