I was one of the people who edited interview notes and sent other feedback to Triplebyte candidates; I certify that everything John said here is correct, even re: the parts of the process he wasn’t directly involved in, and I endorse his takeaways. This comment is more a response to John than it is a response to the OP, but hopefully/maybe people might still find it useful.
Feedback emails were about 25% of my job. As a team, we sent maybe 50 feedback emails on an average day (not totally sure here, numbers fluctuated a lot and also it was two years ago).
One of the things that made it possible to give good feedback at scale was that Triplebyte had a well-oiled, standardized process. Every candidate took much the same interview, which meant that we could largely build our emails out of pre-existing blocks — e.g., telling a candidate that we were impressed with their code quality but they could have been faster, or mentioning specific knowledge areas where they could improve and linking to relevant resources. I doubt the same could be done at most EA orgs? Maybe if they’re hiring programmers.
The process of editing interviewers’ raw feedback became pretty quick and easy after a while (edit out the swearing and anything mean, change some keywords, bam), although sometimes one of us would slip up and that wasn’t great, lol. So yeah I agree that this is a job that could pretty easily be offloaded to a decent communicator who was familiar with the interview process. We did write some of our own content if we felt it was needed (e.g. writing nice things if we knew the candidate was struggling personally), and we used our judgment to tone down the harshness (e.g. if someone needed improvement in every single area we tested, we would focus on just a few areas rather than sending a super long email telling them they were bad at everything).
There was also huge variation in quality between the notes of different interviewers; some would write long, detailed, encouraging messages to the candidates, while others would write like one sentence. So if EA orgs choose to go down this road, they need to make sure to give the feedback-giver enough to work with.
Another thing is that we were explicitly assessing only candidates’ technical abilities, and not whether we wanted them to join our team. That meant that all rejections were of the form “you need to brush up on X skills”, and we never had to reference a candidate’s personality or energy levels or whatever. That probably helps a ton re: protection from lawsuits. (I had never thought of that before, huh.)
The address on this event is the address for the Lodge, not the Burrow