Thanks for this, it’s an interesting point and made me slightly more optimistic about the potential impact of direct for-profit work. My prior is sceptical of this however, for three main reasons:
In general, an organisation optimises for one thing. That’s normally profit in for-profits, and (ideally) impact for non-profits. My expectation is that an organisation that tries to optimise both at once will not do either as well as it could if it just focused on one.
The tight loop between success (revenue) and growth (investing that revenue) make a powerful engine for for-profits. But in general we should expect that such an engine will be systematically misaligned with providing services to the global poor because you cannot make as much money off them as by providing services to the global rich. A similar (and stronger) argument can be made for animals and future people, both groups which are entirely unable to pay.
The most impactful interventions are unlikely to be the most profitable ones. We should expect this on prior just from there existing an imperfect correlation between these metrics, and from an expectation that both distributions are heavy-tailed (cf. ‘Why the tails come apart’). The likelihood of an intervention being at the extremes of both impact and profitability is low.
Any of these considerations are just general patterns, and so may not apply to a particular case. You’ve pointed out some very promising exceptions, and so even if a sceptical prior is right, the expected value of direct for-profit work may still be underrated, like you say.
It is true that there will be divergence in recourses if profit and impact don’t correlate strongly. In the other resources section for more about profitability, see this here. I would generally advise against the optimization of 2 goals at the same time, but when the mission of the company is aligned either by selling to the beneficiaries or when a technology correlates with a better product and good outcomes (Tesla with EVs), or customers internalize the cost of current products (Tesla customers want green cars), the 2 become aligned. SpaceX is finding how to establish a mars base by selling launch services and internet with the same rockets.
Not all organizations can be for-profit, but when it is possible, it can lead to much bigger growth. One can innovate from high cost to lower cost (such as Tesla). Also middle income countries may have enough for moderate spending. Each business will face the problem and have the learn about the specifics of the situation to know where they can make money and make sure it is aligned with their mission. Not all direct-impact for-profit companies are selling to the poor. Take lab-grown meat, there is potentially a large group of the global rich who would be the first customers.
There should still be non-profits, but my thesis was that direct-impact for-profits are underrated, not that all non-profits should start for-profits. If they have the possibility and it would lead to higher impact, then they should. If one can start something like one of the companies mentioned, or Tesla or SpaceX, then there should be a good bit of attention paid to it in EA, rather than just small non-profits. We can then compare starting/joining a startup that serves a smaller problem but grows naturally vs a non-profit that is serving a large problem but isn’t guaranteed to grow and relient on donors. In the FTX case, I don’t know how many non-profits went out of business because of it, but it could happen again (due to fraud or otherwise).
Thanks for this, it’s an interesting point and made me slightly more optimistic about the potential impact of direct for-profit work. My prior is sceptical of this however, for three main reasons:
In general, an organisation optimises for one thing. That’s normally profit in for-profits, and (ideally) impact for non-profits. My expectation is that an organisation that tries to optimise both at once will not do either as well as it could if it just focused on one.
The tight loop between success (revenue) and growth (investing that revenue) make a powerful engine for for-profits. But in general we should expect that such an engine will be systematically misaligned with providing services to the global poor because you cannot make as much money off them as by providing services to the global rich. A similar (and stronger) argument can be made for animals and future people, both groups which are entirely unable to pay.
The most impactful interventions are unlikely to be the most profitable ones. We should expect this on prior just from there existing an imperfect correlation between these metrics, and from an expectation that both distributions are heavy-tailed (cf. ‘Why the tails come apart’). The likelihood of an intervention being at the extremes of both impact and profitability is low.
Any of these considerations are just general patterns, and so may not apply to a particular case. You’ve pointed out some very promising exceptions, and so even if a sceptical prior is right, the expected value of direct for-profit work may still be underrated, like you say.
It is true that there will be divergence in recourses if profit and impact don’t correlate strongly. In the other resources section for more about profitability, see this here. I would generally advise against the optimization of 2 goals at the same time, but when the mission of the company is aligned either by selling to the beneficiaries or when a technology correlates with a better product and good outcomes (Tesla with EVs), or customers internalize the cost of current products (Tesla customers want green cars), the 2 become aligned. SpaceX is finding how to establish a mars base by selling launch services and internet with the same rockets.
Not all organizations can be for-profit, but when it is possible, it can lead to much bigger growth. One can innovate from high cost to lower cost (such as Tesla). Also middle income countries may have enough for moderate spending. Each business will face the problem and have the learn about the specifics of the situation to know where they can make money and make sure it is aligned with their mission. Not all direct-impact for-profit companies are selling to the poor. Take lab-grown meat, there is potentially a large group of the global rich who would be the first customers.
There should still be non-profits, but my thesis was that direct-impact for-profits are underrated, not that all non-profits should start for-profits. If they have the possibility and it would lead to higher impact, then they should. If one can start something like one of the companies mentioned, or Tesla or SpaceX, then there should be a good bit of attention paid to it in EA, rather than just small non-profits. We can then compare starting/joining a startup that serves a smaller problem but grows naturally vs a non-profit that is serving a large problem but isn’t guaranteed to grow and relient on donors. In the FTX case, I don’t know how many non-profits went out of business because of it, but it could happen again (due to fraud or otherwise).
Thanks for the comment!