Experiment without the wait: Speeding up the iteration cycle with Offline Replay Experimentation. Awesome work by my colleague Maxine @PinterestEng https://t.co/hTSj1XoVd1
@TinaLBarton We've moved to a directly responsible individual (DRI) system to simplify things. Every project has a DRI who makes the final call. If it's cross-functional, there can be a DRI from each team. Pros: speed and clear ownership. Cons: up to the DRI to loop in others.
Despite being on Twitter right now, I think social media is wrecking our brains. Especially bad during COVID, with fewer options to go out and do things. Sharing my thoughts on how I'm trying to better utilize my free time. https://t.co/jAGeZfnE6k
I've been using this internal tool at Pinterest for years and it's indispensable to my work. Congrats to the team launching this! Open sourcing Querybook, Pinterest’s collaborative big data hub by @PinterestEng https://t.co/EZsmYm1iHC
First blog post in ~2 years, about what's been top of mind for me lately: ways I'm practicing essentialism while working in tech
https://t.co/i9d8w0WjTC
Wow. Two Rep. members of Board of Canvassers refused to certify election in Wayne County, MI if votes from Detroit were included. Trump tweeted his support of this. Board has since voted to certify, but fellow board member @NedStaebler has some strong words for his colleagues
@2plus2make5 Have to coordinate sampling across counties. Need a sampling frame in the form of "ballot manifest," a list of ballot IDs and storage locations. Different pools of ballots (mail, in person, provisional). Wrote a paper on these issues (audits, not polling) https://t.co/MVC4eoNjqv
@willafriedman This paper I worked on during grad school uses two unique sets of student evaluations of teaching to show evidence of gender bias in these evals. https://t.co/eQjqvSin50
Election commission orders top voting machine vendor to correct misleading claims. "DS200 optical scan machines *with modems* are in use in 11 states and the District of Columbia" -- scary. https://t.co/sljFmZvLRY.
@sastoudt "Best" procedure: use a long (10-20 digits) random number from another source as the seed. Bad procedure: seed with a lot of zeros. See section 4 (this came out AFTER my dissertation of course) https://t.co/PwtHAvkkvW
I, like she, got into a lucky situation with a great advisor who supported me and taught me a lot. I got to work on projects I cared about and touch many topics. Not everyone has that experience.
It's time for statistics departments to start supporting their applied students. @rlbarter shares what I've been thinking for years. It's BEEN time for the field to change, or lose a lot of great people. https://t.co/BVbDisYSIN