I’m confused why you wouldn’t feel concerned about EA potentially wasting 15M pounds (talking about your hypothetical example, not the real purchase). I feel that would mean that EA is not living up to its own standards of using evidence and reasoning to help others in the best possible way.
Since EA isn’t optimizing the goal “flip houses to make a profit”, I expect us to often be willing to pay more for properties than we’d expect to sell them for. Paying 2x is surprising, but it doesn’t shock me if that sort of thing is worth it for some reason I’m not currently tracking.
MIRI recently spent a year scouring tens of thousands of properties in the US, trying to find a single one that met conditions like “has enough room to fit a few dozen people”, “it’s legal to modify the buildings or construct a new one on the land if we want to”, and “near but not within an urban center”. We ultimately failed to find a single property that we were happy with, and gave up.
Things might be easier outside the US, but the whole experience updated me a lot about how hard it is to find properties that are both big and flexible / likely to satisfy more than 2-3 criteria at once.
At a high level, seems to me like EA has spent a lot more than 15M£ on bets that are vastly more uncertain and dependent-on-contested-models than “will we want space to house researchers or host meetings?”. Whether discussion and colocation is useful is one of the only things I expect EAs to not disagree about; most other categories of activity depend on much more complicated stories, and are heavily about placing bets on more specific models of how the future is likely to go, what object-level actions to prioritize over other actions, etc.
I’m confused why you wouldn’t feel concerned about EA potentially wasting 15M pounds (talking about your hypothetical example, not the real purchase). I feel that would mean that EA is not living up to its own standards of using evidence and reasoning to help others in the best possible way.
Since EA isn’t optimizing the goal “flip houses to make a profit”, I expect us to often be willing to pay more for properties than we’d expect to sell them for. Paying 2x is surprising, but it doesn’t shock me if that sort of thing is worth it for some reason I’m not currently tracking.
MIRI recently spent a year scouring tens of thousands of properties in the US, trying to find a single one that met conditions like “has enough room to fit a few dozen people”, “it’s legal to modify the buildings or construct a new one on the land if we want to”, and “near but not within an urban center”. We ultimately failed to find a single property that we were happy with, and gave up.
Things might be easier outside the US, but the whole experience updated me a lot about how hard it is to find properties that are both big and flexible / likely to satisfy more than 2-3 criteria at once.
At a high level, seems to me like EA has spent a lot more than 15M£ on bets that are vastly more uncertain and dependent-on-contested-models than “will we want space to house researchers or host meetings?”. Whether discussion and colocation is useful is one of the only things I expect EAs to not disagree about; most other categories of activity depend on much more complicated stories, and are heavily about placing bets on more specific models of how the future is likely to go, what object-level actions to prioritize over other actions, etc.