The fact that it is possible to randomly get some good outcomes despite low vetting standards does not make a cost-effective way to get good outcomes. The Hotel being hits-based approach does not preclude a better vetting policy.
That’s true, but I think it actually IS a cost-effective way to get good outcomes. There are already grant giving organizations that fill the next step in the pyramid with stronger vetting based on status, credentials, and results. I think organizations with stronger vetting need to exist as well, but a community that has organizations with very strong feedback loops (and weak vetting upfront) as well as organizations later in the pipeline with only strong vetting, will get more hits than one that only has organizations with very strong pre-vetting procedures.
About the plans to vet post-hoc: I predict bias towards keeping people because of sunken costs on both sides.
I agree this is a risk, but don’t agree it’s inevitable, especially if you’re looking for it.
I agree that tapping into the thought diversity of the larger community is good, I just think that you need some vetting—what I would like to see is a plan for a vetting process which both gets this diversity but also maximises quality. I don’t think you need a total open doors policy to get thought diversity, although acknowledge the trade off.
I don’t think this is the actual policy of the hotel, nor what I was advocating. It only makes sense when you have less people then slots available (which has been the case much of the time). I do think that a policy of “accept anyone if there’s space, vet a bit more if it’s competitive,” makes sense for something looking to fill the niche of the EA hotel.
The vetting also looks very different for an organization at that stage in the pyramid, more “have they actually thought through the idea/project?” than “do they have previous accomplishments that they makes me think they could pull the project off?” or “do I agree it’s a high value project?”.
I just think that is a quite a different goal to the one of being an incubator for high impact projects.
This is a good point. I do think there’s value in incubating both people and projects. People may find out they’re not cut out to run projects, or good people may find out their project isn’t viable, but either way they’ve gained valuable skills that allow them to get across the chasm and move up the pyramid.
don’t like the pyramid. In particular, the distinction drawn between someone who self-identifies strongly or weakly as EA seems irrelevant
I agree that could have been worded better. The point is if you’re “sort of into EA” there are ways to get you really into EA—meetup groups, conferences, etc. Once there, you might start thinking of creating a project that does as much good as possible… at which point you hit the chasm. That is, it’s trying to show that there are parts of the pipeline that exist before the Chasm.
I think you are doing the right thing here, and if anything I would say you could go further in the direction you are already going.
I observe that less vetting means fewer decisions and less costs for the Hotel. Further, if demand for slots is low enough that no vetting is required, this effectively makes the project zero-risk to the Hotel. A good member of the community is still helpful to all the other members of the community, even if their project goes nowhere.
Following on that point, I support coming down hard on the side of optimizing for people over projects. I can think of several reasons, but the simplest is that this was the explicit position for Xerox PARC, which birthed personal computing and is therefore a candidate for the highest-impact project of all time. A lot of relevant detail is in Alan Kay’s The Power of the Context, and a fuller history in The Dream Machine by M. Mitchell Waldrop.
Further, a person’s impact is probably spread over many projects so investing in them is usually a gain, whereas a failed project is a sunk cost. Lastly, the people optimization method will help with another way in which EA is constrained: unambiguous signals about how and where to apply the absurd surplus of talent available. I expect this last to work both ways, so people who have spent time at the Hotel will also have a better sense of where they can contribute next.
I observe that less vetting means fewer decisions and less costs for the Hotel. Further, if demand for slots is low enough that no vetting is required, this effectively makes the project zero-risk to the Hotel.
This seems to assume the marginal cost to the hotel of taking on a guest is negligible. That does seem plausible to me, but it’s worth highlighting explicitly.
This is close to what I was assuming, but you are right I should have been explicit.
My actual assumption is that the marginal difference in cost between taking on one or another guest is negligible. Based on this I make the claim that projects which are not evaluated by the Hotel are zero risk (to the Hotel).
I expect we should always prefer the case where we did not evaluate the project at all to the case where we evaluated it and were wrong. I don’t see any reason to expect that the cost of evaluating experimental projects will have a high enough success rate to be a net benefit, even before we consider the impact of taking time and money away from the focus on supporting people.
That’s true, but I think it actually IS a cost-effective way to get good outcomes. There are already grant giving organizations that fill the next step in the pyramid with stronger vetting based on status, credentials, and results. I think organizations with stronger vetting need to exist as well, but a community that has organizations with very strong feedback loops (and weak vetting upfront) as well as organizations later in the pipeline with only strong vetting, will get more hits than one that only has organizations with very strong pre-vetting procedures.
I agree this is a risk, but don’t agree it’s inevitable, especially if you’re looking for it.
I don’t think this is the actual policy of the hotel, nor what I was advocating. It only makes sense when you have less people then slots available (which has been the case much of the time). I do think that a policy of “accept anyone if there’s space, vet a bit more if it’s competitive,” makes sense for something looking to fill the niche of the EA hotel.
The vetting also looks very different for an organization at that stage in the pyramid, more “have they actually thought through the idea/project?” than “do they have previous accomplishments that they makes me think they could pull the project off?” or “do I agree it’s a high value project?”.
That was comparing the culture, not the goals.
This is a good point. I do think there’s value in incubating both people and projects. People may find out they’re not cut out to run projects, or good people may find out their project isn’t viable, but either way they’ve gained valuable skills that allow them to get across the chasm and move up the pyramid.
I agree that could have been worded better. The point is if you’re “sort of into EA” there are ways to get you really into EA—meetup groups, conferences, etc. Once there, you might start thinking of creating a project that does as much good as possible… at which point you hit the chasm. That is, it’s trying to show that there are parts of the pipeline that exist before the Chasm.
I think you are doing the right thing here, and if anything I would say you could go further in the direction you are already going.
I observe that less vetting means fewer decisions and less costs for the Hotel. Further, if demand for slots is low enough that no vetting is required, this effectively makes the project zero-risk to the Hotel. A good member of the community is still helpful to all the other members of the community, even if their project goes nowhere.
Following on that point, I support coming down hard on the side of optimizing for people over projects. I can think of several reasons, but the simplest is that this was the explicit position for Xerox PARC, which birthed personal computing and is therefore a candidate for the highest-impact project of all time. A lot of relevant detail is in Alan Kay’s The Power of the Context, and a fuller history in The Dream Machine by M. Mitchell Waldrop.
Further, a person’s impact is probably spread over many projects so investing in them is usually a gain, whereas a failed project is a sunk cost. Lastly, the people optimization method will help with another way in which EA is constrained: unambiguous signals about how and where to apply the absurd surplus of talent available. I expect this last to work both ways, so people who have spent time at the Hotel will also have a better sense of where they can contribute next.
This seems to assume the marginal cost to the hotel of taking on a guest is negligible. That does seem plausible to me, but it’s worth highlighting explicitly.
This is close to what I was assuming, but you are right I should have been explicit.
My actual assumption is that the marginal difference in cost between taking on one or another guest is negligible. Based on this I make the claim that projects which are not evaluated by the Hotel are zero risk (to the Hotel).
I expect we should always prefer the case where we did not evaluate the project at all to the case where we evaluated it and were wrong. I don’t see any reason to expect that the cost of evaluating experimental projects will have a high enough success rate to be a net benefit, even before we consider the impact of taking time and money away from the focus on supporting people.