When would it make sense for someone to leave a successful earning-to-give tech job to try to work on creating a start up (with the goal of earning to give)? Which considerations are at play? How might someone create a framework for making this decision?
I’m just going to state my views not fully back them up. I hope to cover this topic in more depth later.
I’ll go through a couple of key considerations and how to settle them.
1) Expected earnings
We have a bunch of unpublished research on this. I think the earnings as a startup founder are considerably higher in expectation (say at least 5x) BUT only if you make it into a top 5 accelerator or similar. More broadly, top accelerators seem to have very strong selection power—almost no-one who’s rejected by YC succeeds, but the odds are pretty good if you get in.
So I think reframe the question as “is it worth me spending a couple of months seeing if I can get into a top accelerator?”.
That depends on your situation. Do you have a cofounder? Do you have a problem in mind? Other entrepreneurs in the EA community can probably advise you on this.
Also since the earnings of founding are delayed 5-10 years, they’re worth about 50% less.
2) Career capital
You’ll may improve your technical skills more as a SWE, but as a founder you’ll learn more entrepreneurial skills and I think those are more valuable (in particular, within EA, there are loads of SWE, but a real shortage of entrepreneurs).
It’s also better for networking, and you’ll have to try out all types of work, making it good for exploration. The greater intensity also means you’ll probably push yourself more.
If you fail as an entrepreneur in the US, I think you can go back to SWE with little or no penalty.
Being a startup founder is far harder work and far more stressful. Most founders say they went through really high highs and really low lows. Even the most successful startups are often close to failure several times. All startup founders work very long hours. If you can’t hack this, then don’t do it.
Startups also take over your life and become your sole focus. If you succeed, you’ll be locked in for 5-10 years. If you’re unwilling to commit to this one thing for so long, don’t do it.
This means you’ll also have little time for side projects.
4) Direct impact
Entrepreneurship probably generates more consumer surplus and positive externalities than being a SWE because it’s more innovative. It’ll depend on the companies though.
How strong a consideration this is depends on what cost-benefit ratio you put on your donations. If it’s around 10, then it’s probably important. If it’s more like 100-1000, the donations dominate.
5) Flexibility
The startup founder path involves more ‘lock-in’, so if you’re highly unsure, lean against.
6) The right reasons
In the startup world, people don’t recommend starting unless you’re extremely driven by the problem you’re trying to solve. Going into it because you expect to earn more probably isn’t good enough.
Overall, of those 2 options, I’d lean towards encouraging you to experiment with the startup path—try to put something together that you could pitch to an accelerator.
If you don’t have a good idea, try to join an early stage startup instead (especially one via your network that’s got top backers). This is also much better for lifestyle, in particular, it’s much easier to quit, so you can try several times before hitting on a success.
Finally, this is a bigger question, but I don’t think you should optimise for earning to give. I think you should work towards working at an EA org or potentially exploring other areas neglected by EAs.
Re: number two: one of the reasons I dislike going into entrepreneurship right away for career capital is just a basic principle that, if you want to learn something, it’s almost never optimal to rediscover that thing independently. You almost always want to have someone else teach you, if possible.
With this more precise statement, it’s more of a criticism of the “standard model” of startups (two friends from college with similar backgrounds starting a company) versus a critique of startups per se. If you start a company with someone who is significantly more experience than you you might be able to get the best of both worlds.
Sorry this is slightly off topic. Considering Givewell’s recommendations are between 1x and 10-20x and there are questions about room for funding etc. where are you thinking that people might conceivably get a 100-1000x return on their donations?
And you could think AMF is 5x that, getting you to over 100.
Then you could think GiveWell itself (or other EA orgs) get you another 10x due to the multiplier effect, getting you to an overall CBR of 1000x.
Of course there’s lots of ways you could object to this too!
2) Another way you might get led to CBR of over 100 is that if you care mainly about long-run effects, and you think there’s a relatively narrow range of activities that clearly benefit the long-run. e.g. if you think preventing existential risk is the key proxy for a good long-run future, you might think most ways of spending money have about zero effect on that, while some opportunities have very positive effects (e.g. donating to the FHI), creating a very high-return on well targeted donations.
Those are multipliers in terms of social value created for the community compared to cost. It is a common (but not universal) position that a dollar has a higher value in a poor society than a rich one. If your work creates social value mostly in rich countries but your donations create value mostly in poor ones, this could add one to two orders of magnitude to the relevant ratio.
This is perhaps easiest to think about with GiveDirectly. At one level it’s easy to see that the return is 1x. The reason people think it’s a better target than just spending money yourself is that a dollar goes much further in the target community.
Some people may also or instead think there are more speculative interventions with bigger returns ratios.
As a distant observer, your .impact EA work seems unlikely to get replaced, whereas there are plenty of people doing startups—are you worried that adding a startup to your earning to give might squeeze EA work out?
I’m not sure doing a startup should really be thought of as replaceable.
But it is true that the .impact work is EA movement building, whereas doing a start-up isn’t, so if you think EA movement building is more pressing, then that’s reason against the startup.
True, the chances of someone doing a very similar startup aren’t 100% (though they’re not 0%). I was more thinking that there’s more of a shortage of people doing EA movement building than of people doing startups, and partly as a result of this a marginal extra person doing EA work is more valuable.
When would it make sense for someone to leave a successful earning-to-give tech job to try to work on creating a start up (with the goal of earning to give)? Which considerations are at play? How might someone create a framework for making this decision?
I’m just going to state my views not fully back them up. I hope to cover this topic in more depth later.
I’ll go through a couple of key considerations and how to settle them.
1) Expected earnings
We have a bunch of unpublished research on this. I think the earnings as a startup founder are considerably higher in expectation (say at least 5x) BUT only if you make it into a top 5 accelerator or similar. More broadly, top accelerators seem to have very strong selection power—almost no-one who’s rejected by YC succeeds, but the odds are pretty good if you get in.
So I think reframe the question as “is it worth me spending a couple of months seeing if I can get into a top accelerator?”.
That depends on your situation. Do you have a cofounder? Do you have a problem in mind? Other entrepreneurs in the EA community can probably advise you on this.
Also since the earnings of founding are delayed 5-10 years, they’re worth about 50% less.
2) Career capital
You’ll may improve your technical skills more as a SWE, but as a founder you’ll learn more entrepreneurial skills and I think those are more valuable (in particular, within EA, there are loads of SWE, but a real shortage of entrepreneurs).
It’s also better for networking, and you’ll have to try out all types of work, making it good for exploration. The greater intensity also means you’ll probably push yourself more.
If you fail as an entrepreneur in the US, I think you can go back to SWE with little or no penalty.
Overall I’d say the startup path is better for career capital, though there are counterpoints: http://philosophyforprogrammers.blogspot.com/2014/09/if-you-want-to-start-startup-go-work.html
3) Lifestyle
Being a startup founder is far harder work and far more stressful. Most founders say they went through really high highs and really low lows. Even the most successful startups are often close to failure several times. All startup founders work very long hours. If you can’t hack this, then don’t do it.
Startups also take over your life and become your sole focus. If you succeed, you’ll be locked in for 5-10 years. If you’re unwilling to commit to this one thing for so long, don’t do it.
This means you’ll also have little time for side projects.
4) Direct impact
Entrepreneurship probably generates more consumer surplus and positive externalities than being a SWE because it’s more innovative. It’ll depend on the companies though.
How strong a consideration this is depends on what cost-benefit ratio you put on your donations. If it’s around 10, then it’s probably important. If it’s more like 100-1000, the donations dominate.
5) Flexibility
The startup founder path involves more ‘lock-in’, so if you’re highly unsure, lean against.
6) The right reasons
In the startup world, people don’t recommend starting unless you’re extremely driven by the problem you’re trying to solve. Going into it because you expect to earn more probably isn’t good enough.
http://startupclass.samaltman.com/courses/lec01/
Conclusion
Overall, of those 2 options, I’d lean towards encouraging you to experiment with the startup path—try to put something together that you could pitch to an accelerator.
If you don’t have a good idea, try to join an early stage startup instead (especially one via your network that’s got top backers). This is also much better for lifestyle, in particular, it’s much easier to quit, so you can try several times before hitting on a success.
Finally, this is a bigger question, but I don’t think you should optimise for earning to give. I think you should work towards working at an EA org or potentially exploring other areas neglected by EAs.
And talk a lot to Ben West!
Re: number two: one of the reasons I dislike going into entrepreneurship right away for career capital is just a basic principle that, if you want to learn something, it’s almost never optimal to rediscover that thing independently. You almost always want to have someone else teach you, if possible.
With this more precise statement, it’s more of a criticism of the “standard model” of startups (two friends from college with similar backgrounds starting a company) versus a critique of startups per se. If you start a company with someone who is significantly more experience than you you might be able to get the best of both worlds.
Going into an accelerator could also help a lot.
Thanks Ben, that’s a helpful summary.
Sorry this is slightly off topic. Considering Givewell’s recommendations are between 1x and 10-20x and there are questions about room for funding etc. where are you thinking that people might conceivably get a 100-1000x return on their donations?
Just to clarify, I’m not claiming you can get 1000x returns, just that your view about the maximum spread of returns will effect the conclusion.
Some examples of how you could reason towards CBR of over 100:
1) It’s plausible Give Directly has a CBR of ~30 (just focusing on the short-run welfare effects). http://reflectivedisequilibrium.blogspot.com/2014/03/givedirectly-happiness-and-log-income.html
And you could think AMF is 5x that, getting you to over 100.
Then you could think GiveWell itself (or other EA orgs) get you another 10x due to the multiplier effect, getting you to an overall CBR of 1000x.
Of course there’s lots of ways you could object to this too!
2) Another way you might get led to CBR of over 100 is that if you care mainly about long-run effects, and you think there’s a relatively narrow range of activities that clearly benefit the long-run. e.g. if you think preventing existential risk is the key proxy for a good long-run future, you might think most ways of spending money have about zero effect on that, while some opportunities have very positive effects (e.g. donating to the FHI), creating a very high-return on well targeted donations.
You can see a great review of the overall debate about how large typical differences in cost-effectiveness are here: http://reducing-suffering.org/why-charities-dont-differ-astronomically-in-cost-effectiveness/
Those are multipliers in terms of social value created for the community compared to cost. It is a common (but not universal) position that a dollar has a higher value in a poor society than a rich one. If your work creates social value mostly in rich countries but your donations create value mostly in poor ones, this could add one to two orders of magnitude to the relevant ratio.
This is perhaps easiest to think about with GiveDirectly. At one level it’s easy to see that the return is 1x. The reason people think it’s a better target than just spending money yourself is that a dollar goes much further in the target community.
Some people may also or instead think there are more speculative interventions with bigger returns ratios.
Great responses, thanks both for taking the time.
As a distant observer, your .impact EA work seems unlikely to get replaced, whereas there are plenty of people doing startups—are you worried that adding a startup to your earning to give might squeeze EA work out?
I’m not sure doing a startup should really be thought of as replaceable.
But it is true that the .impact work is EA movement building, whereas doing a start-up isn’t, so if you think EA movement building is more pressing, then that’s reason against the startup.
True, the chances of someone doing a very similar startup aren’t 100% (though they’re not 0%). I was more thinking that there’s more of a shortage of people doing EA movement building than of people doing startups, and partly as a result of this a marginal extra person doing EA work is more valuable.