mcastillon

Before you know it, we're going to have a real world proof of the infinite monkey theorem https://en.wikipedia.org/wiki/Infinite_monkey_theorem

I think this underrates how many emails are literally just replies of "sounds good". Small snippet replies seem to be the vast majority of automatically suggested responses in gmail

538 actually did a terrific job accounting for those polling phenomena, finding that in spite of that narrative around right-leaning polls arising, polls that cycle were pretty spot on https://fivethirtyeight.com/features/2022-election-polling-a...

Nate got his start in this world doing baseball forecasting (https://en.wikipedia.org/wiki/PECOTA), well before he delved into politics. Jay Boice definitely doing much of the heavy lifting these days, and deserves more personal recognition than he currently gets. But I wouldn't discount the amount of work Nate personally put in on the sports front.

Reading some more, it does look like Zillow was in fact leveraging humans in the loop https://venturebeat.com/2019/12/11/why-explainable-ai-is-ind.... What seems like a likely explanation is that the proper balance between machine and human judgement was not calibrated to this current market.

"Part of the challenge of Zillow Offers’ human-in-the-loop system — any such system, really — is finding the balance between humans and machines. “In order to optimize this human-in-the-loop system, we’d like to figure out when the human is best, when an assistive situation is best, and […] when a machine is best,” said Fagnan."

I'm surprised that nothing here has been written about human in the loop. All time series models are flawed in some capacity, and are going to have difficulties in the current market given how it won't necessarily match historical seasonalities. Just having more human safeguards to flag wacky outputs seems to be a more reasonable explanation than not handling Prophet's shortcomings properly. Zillow would have likely faced similar issues with other time series packages as well

I think in general just limiting your world view to singular topics is also just not a great way to learn about the world. We all have our blindspots, and it's exacerbated by that sort of curation

It could be argued that by focusing on the wide release, they inherently doomed the product. Hard to build a cohesive community of 300M when the stickiness factor doesn't exist yet. Arguably would have been better to focus on limited releases on niche communities to create that stickiness/increase retention.