So a day or 2 back FivethirtyEight got hit by a divide by zero bug that led to them briefly showing Trump with a greater than 99% chance of victory. From what I can make out I think fivethirtyeight are using STATA. So I just wondered if Scala users had more information on the Tech Stacks of the major players in this election: the pollsters, aggregators, networks and parties.
Not nearly as bad as the racecar that crashed due to NaN…
I certainly don’t claim to be an expert on political polling, but I can’t imagine that the most important computations are particularly complicated.
The challenge of polling is to get a statistically representative sampling. Some polling organizations deliberately try to bias the results by oversampling a particular party, but that’s another matter. The challenge for honest polling organizations is just getting people to cooperate. I vaguely recall reading a while back that something like 40% of those contacted hang up, but don’t quote me on that. The problem is that they have no way of knowing whether supporters of a particular candidate or party are more likely to hang up, thereby skewing the results.
Let’s stick to tech stacks only, please, and not let this turn into an election discussion.
On second thought, I’m just going to close this topic, as any potential connection to Scala seems hypothetical and tenuous and it just seems to open the door to off-topic discussion.
Agreed. There are some interesting tech issues regarding elections (e.g. issues of reliability, exactly-once transactions, and so on), but if we don’t actually have such a question, let’s not make one up, even if it seems timely in the United States.