Context: used to work for CFAR, currently work for MIRI.
“CFAR exists much less than it used to” feels true. “MIRI exists much less than it used to” feels true, but false-by-implication when lumped in with the comment about CFAR, because that makes them feel like similar reductions and that’s not at all the case.
CFAR is essentially nonexistent/non-recognizable from the perspective of someone who attended a workshop 2014 − 2019. There are (if I understand correctly) 2-3 employees at the moment, and projects are spun up one at a time. There might be more ambitious or enduring things in the future, but I think that’s all up in the air?
Whereas MIRI still has multiple functioning research groups that have been pursuing their research directions more-or-less uninterrupted the whole time. We did wind down a large project that had caused us to hire a bunch of engineers, and those engineers don’t work for us anymore, and also it’s quite true that Nate and Eliezer (our two seniormost researchers) do not have a concrete angle of attack on the acute risk problem.
But if CFAR is something like 10% of its 2018 self, MIRI is more like 70-90% of its 2018 self. It swelled and shrank 2019-2021, but is still a functioning org in a way that, if I understand correctly, CFAR is not.
Reminder: do not currently work for CFAR, do not have super special insider knowledge. I think I’m closer to both orgs than Ben, so it felt worth chiming in, but Anna or Jack or someone else close to CFAR may have corrections for my narrative there.
Context: used to work for CFAR, currently work for MIRI.
“CFAR exists much less than it used to” feels true. “MIRI exists much less than it used to” feels true, but false-by-implication when lumped in with the comment about CFAR, because that makes them feel like similar reductions and that’s not at all the case.
CFAR is essentially nonexistent/non-recognizable from the perspective of someone who attended a workshop 2014 − 2019. There are (if I understand correctly) 2-3 employees at the moment, and projects are spun up one at a time. There might be more ambitious or enduring things in the future, but I think that’s all up in the air?
Whereas MIRI still has multiple functioning research groups that have been pursuing their research directions more-or-less uninterrupted the whole time. We did wind down a large project that had caused us to hire a bunch of engineers, and those engineers don’t work for us anymore, and also it’s quite true that Nate and Eliezer (our two seniormost researchers) do not have a concrete angle of attack on the acute risk problem.
But if CFAR is something like 10% of its 2018 self, MIRI is more like 70-90% of its 2018 self. It swelled and shrank 2019-2021, but is still a functioning org in a way that, if I understand correctly, CFAR is not.
Reminder: do not currently work for CFAR, do not have super special insider knowledge. I think I’m closer to both orgs than Ben, so it felt worth chiming in, but Anna or Jack or someone else close to CFAR may have corrections for my narrative there.
(For the record these comments by Duncan seemed to me both helpful and closer to the ground than mine, I don’t know why Duncan retracted them.)
Oh, Anna chimed in and they seemed to have been in-many-places directly contradicted by Anna’s statements.