It’s giving season and I want to finally get around to publishing some of my thoughts and experiences around funding. I haven’t written anything yet because I feel like I am mostly just revisiting painful experiences and will end up writing some angry rant. I have ideas for how things could be better so hopefully this can lead to positive change not just more complaining.
All my experiences are in AI Safety.
On Timing:
Certainty is more important than speed.
The total decision time is less important than the overdue time. Expecting a decision in 30 days and getting it in 35 days is worse than if I expect the decision in 90 days and I get it in 85 days.
Grantmakers providing statistics about timing expectations makes things worse. If the mean or median response time is N days it is now N+5 days is it appropriate for me to send a follow-up email to check on the status? Technically it’s not late yet. It could come tomorrow or in N more days. Imagine if the Uber app showed you the global mean wait time for the last 12 months and there was no map to track your driver’s arrival.
“It doesn’t have to reduce the waiting time it just has to reduce the uncertainty”—Rory Sutherland
My conversations about people’s expectations and experiences with people in Berkeley are at times very different to those outside of Berkeley.
After I posted my announcement about shutting down AISS and my comment on the LTFF update several people reached out to me about their experiences. Some people I already knew well, some I had met and others I didn’t know before.
Some of them had received funding a couple of times but their negative experiences led them to not reapply and walk away from their work or the ecosystem entirely. At least one mentioned having a draft post about their experience that they did not feel comfortable publishing.
There was definitely a point for me where I had already given up but just not realised it. I had already run out of funding, I was still waiting for a funding decision but even if I had gotten the funding at that stage there was no way that I would still be working on the project when it next needed funding.
I lost a lot of sleep waiting for a decision from someone on the other side of the planet. Aware that they are likely overwhelmed themselves. I stressed about when it was appropriate so send another follow-up email. What difference would it really make? I don’t want to be rude and pushy, particularly to the person who decides if I will be keeping my job
Funding renewals should be treated differently from initial funding.
I know someone who got funding for a year to work on their project. They then reapplied for another year and got rejected.
Imagine applying for a full time job. The hiring manager asks you one follow-up question via email but never interviews you for the job. Almost a year goes by and you have been working hard but still never even met your manager. You are up for a review, and a different manager reviews your written report. This new manager again never meets with you but decides to fire you with no severance payout.
This may not be what is technically happening, but the emotional experience is far better expressed with “I got fired by someone I have never met” than “my tuning application was not accepted”
Region beta paradox.
Fully fund or reject. Don’t under fund to see how it goes and then be disappointed it didn’t go well.
I was supposed to give some ideas that might lead to some positive change…
Grantmakers should have a hard due date.
Prioritise communication over reducing total decision time.
Prioritise reducing late-time over reducing total decision time.
If I were to be an applicant again I would set my own hard decision date. If I had not heard back from the grantmaker by that date then the next day I would send an email to the grantmaker to withdraw my application. Eliminating the uncertainty and inability to take actions.
Fixed not rolling application rounds.
Fund raising rounds tied to application rounds. So on day 1 you open applications and donations for that round. On day 90 you close applications and donation. You now have a fixed budget and a fixed pool of applications. No need to reject incase we get a better application next week. No need to delay an acceptance while hoping to get more funding next week.
Fixed due date for decisions. This date should be embarrassingly achievable.
At the end of the round we update all applicants and donors of the $$$ in applications and only $ funded projects. Somewhere between these two numbers is the funding gap.
Rejected applicants now know they were rejected because of funding shortage and become donors or fundraisers themselves.
Applicants only need to be ranked relative to each other not against a funding bar.
I do agree the EA Funds made a mistake not returning to fixed grant rounds after the Mega Money era was over. It’s so much easier to organize, coordinate, and compare.
I think it’s actually better for applicant to have a deadline too. Plenty of people procrastinate on applying, some to the point of eventually not bothering at all.
Also, if you are rejected but encouraged to reapply it’s clearer when it’s ok to apply. SFF has two rounds a year and I’ve applied, been rejected and then applied again the next round. I’ve probably applied more times to SFF than LTFF and this mostly comes down to there being an application deadline.
Also probably helps when getting rejected and comparing to other things you see get funded. If you see something a year ago that is just like what you were doing or seems much less value got funded it’s hard to understand that the funding bar might have been different then. It’s a much cleaner comparison when you are all in one group.
I think even for myself getting funded by SFF one round and not a later round. I think “oh, look at all these new projects that didn’t exist before that are so clearly awesome. Fair enough they don’t have funding for me any more”
A non-trivial fraction of our most valuable grants require very short turn-around times, and more broadly, there is a huge amount of variance in how much time it takes to evaluate different kinds of applications. This makes a round model hard, since you both end up getting back much later than necessary to applications that were easy to evaluate, and have to reject applications that could be good but are difficult to evaluate.
I’m interested to know why we haven’t seen Lightspeed Grants again?
Some of my thoughts on Lightspeed Grants from what I remember:
I don’t think it’s ever a good idea to name something after the key feature everyone else in the market is failing at. It leads to particularly high expectations and is really hard to get away from. (Eg OpenAI)
The S-process seemed like a strange thing to include for something intended to be fast. As far as I know the S-process has never been done quickly. It also seems to be dependent on every parameter being in place in order to be run so can easily be held up by something small.
At the time of applications it had a clear date when decisions were expected, this is much better than everyone else’s vague expectations. It ended up taking much longer than expected for decisions but still pretty quick overall. This was managed really well though. Certainly much better than LTFFs handing of a large volume of slow decisions last year. I judge the timing of decisions more on how late they are than the total time, but this is a proxy for the real problem which is uncertainty. Lightspeed got the basics right when it comes to comms.
I think it was really good over all and I’d expect that the major issues not accounted for in the first round can be managed.
I would really like to see this run again or a variation of it.
Some of my thoughts on Lightspeed Grants from what I remember: I don’t think it’s ever a good idea to name something after the key feature everyone else in the market is failing at. It leads to particularly high expectations and is really hard to get away from. (Eg OpenAI) The S-process seemed like a strange thing to include for something intended to be fast. As far as I know the S-process has never been done quickly.
You seem to be misunderstanding both Lightspeed Grants and the S-Process. The S-Process and Lightspeed Grants both feature speculation/venture grants which enable a large group of people to make fast unilateral grants. They are by far the fastest grant-decision mechanism that I know out there, and it’s been going strong for multilple years now. If you need funding quickly, an SFF speculation grant is by far the best bet, I think.
It ended up taking much longer than expected for decisions but still pretty quick overall.
I think we generally stayed within our communicated timelines, or only mildly extended them. We did also end up getting more money which caused us to reverse some rejections afterwards, but we did get back to everyone within 2 weeks on whether they would get a speculation grant, and communicated the round decisions at the deadline (or maybe a week or two later, I remember there was a small hiccup).
I’m interested to know why we haven’t seen Lightspeed Grants again?
Ironically one of the big bottlenecks was funding. OpenPhil was one funder who told us they wouldn’t fund us for anything but our LW work (and that also soon after disappeared) and ironically funding coordination work doesn’t seem to pay well. Distributing millions of dollars also didn’t combine very well with being sued by FTX.
I am interested in picking it back up again, but it is also not clear to me how sustainable working on that is.
What kind of numbers are we talking about needing here???
How much did it cost for the last Lightspeed round?
How were the operations funded?
How much distributed? $5mm?
How much would you need for operations in order to run it again?
If you had the operations funding would the grants funding still be a problem?
How much would you need for operations just for the grant decisions but not the distribution of funds?
Lightspeed Grants and the S-Process paid $20k honorariums to 5 evaluators. In addition, running the round probably cost around 8-ish months of Lightcone staff time, with a substantial chunk of that being my own time, which is generally at a premium as the CEO (I would value it organizationally at ~$700k/yr on the margin, with increasing marginal costs, though to be clear, my actual salary is currently $0), and then it also had some large diffuse effects on organizational attention.
This makes me think it would be unsustainable for us to pick up running Lightspeed Grants rounds without something like ~$500k/yr of funding for it. We distributed around ~$10MM in the round we ran.
I’m hesitant to ask you about this so feel free to pass.
Can you say more about how it is that your current salary is $0?
I think most people would be surprised you are not currently receiving a salary. I also assume that as a not-for-profit founder even when you have had a salary it is lower than most or all of your team.
Yeah, I have personally appreciated the short turn around when needed. And seen plenty of situations where people need funds quickly.
I expect there are a lot of these tradeoffs. I think these should be solved by different services not by trying to solve all of the different types of funding together. LTFF ended up here for historical reasons, but now seems to be struggling to serve all of these markets while also crowding out any new funders in the space.
If I set up a new fund it would have rounds and I would just accept that this fund would not be able to support those kinds of applicants that need a fast turn around or longer investigations. A lot of the problems of the applicant experience with LTFF is uncertainty. Having one application form that try’s to solve all of these problems means that the expectations can’t be specific enough to be meaningful.
To be fair, since announcing AISS shutting down, I am having a lot more conversations with people with negative experiences. So now I am comparing the counterfactual good to the bad. It is a big blind spot. People don’t make public posts about this. It’s no one’s job to collect all of the counter cases.
“Choose not to apply” is big part of the problem.
A person gets funded by LTFF but because they had a bad experience as an applicant they choose not to apply and put themselves through that experience again. They also choose not to apply or raise funds anywhere else. So we lost the value of that project continuing. Maybe they went on to do something pretty neutral or maybe they went on to work on capabilities.
People talk privately about this but not publicly. Which leads to others convincing themselves to not apply at all. Some potential donors have more people with bad experiences than good in their network and so don’t want to donate to LTFF. Yes, I’ve spoken to a few people in these cases and despite my own experience I encouraged people to apply anyway.
A core part of my decision to shut down AISS was that I didn’t want to have to be an applicant again. We had gotten funding from LTFF and SFF before but I had reached a point where if we had gotten the funding we were waiting for I would not be continuing after that funding had run out.
This is just some of the downstream effects of bad applicant experience for things actually funded by LTFF.
There are some issues with rejected applications too. Mostly related to the how the applicant and others update based on the rejection.
Project A get rejected, apparently because it might not be the best version of that kind of thing and could crowd out the market for a better version to come along. Someone sees the gap, starts planning Project B that could be better. They chose not to apply because in exploring the idea for their project they find out Project A wanted to do this but couldn’t get funding. Project B wrongly assumes that funders are not interested in funding this kind of project and so give up.
I think if we as the question “should the LTFF shutdown?” The answer is always “no, it clearly does good that no one else is trying to do”
But if you ask “if the LTFF never existed and you were designing a fund from scratch would it look exactly like the LTFF?” I think this is a clear “no” as well.
I don’t think LTFF is able to fix its current problems or be drastically modified to what you might have designed if it never existed. And of course it crowds out anyone else from solving this.
Lack of transparency around the full distribution of decision timeframes, and other aspects of the process, means it’s difficult to make an informed choice the first time around. And one bad experience can be enough to burn people long term, particularly if theyre making the reasonable assumption that LTFF is representative of the average funder.
Some of my thoughts on funding.
It’s giving season and I want to finally get around to publishing some of my thoughts and experiences around funding. I haven’t written anything yet because I feel like I am mostly just revisiting painful experiences and will end up writing some angry rant. I have ideas for how things could be better so hopefully this can lead to positive change not just more complaining. All my experiences are in AI Safety.
On Timing: Certainty is more important than speed. The total decision time is less important than the overdue time. Expecting a decision in 30 days and getting it in 35 days is worse than if I expect the decision in 90 days and I get it in 85 days.
Grantmakers providing statistics about timing expectations makes things worse. If the mean or median response time is N days it is now N+5 days is it appropriate for me to send a follow-up email to check on the status? Technically it’s not late yet. It could come tomorrow or in N more days. Imagine if the Uber app showed you the global mean wait time for the last 12 months and there was no map to track your driver’s arrival.
“It doesn’t have to reduce the waiting time it just has to reduce the uncertainty”—Rory Sutherland
My conversations about people’s expectations and experiences with people in Berkeley are at times very different to those outside of Berkeley.
After I posted my announcement about shutting down AISS and my comment on the LTFF update several people reached out to me about their experiences. Some people I already knew well, some I had met and others I didn’t know before. Some of them had received funding a couple of times but their negative experiences led them to not reapply and walk away from their work or the ecosystem entirely. At least one mentioned having a draft post about their experience that they did not feel comfortable publishing.
There was definitely a point for me where I had already given up but just not realised it. I had already run out of funding, I was still waiting for a funding decision but even if I had gotten the funding at that stage there was no way that I would still be working on the project when it next needed funding.
I lost a lot of sleep waiting for a decision from someone on the other side of the planet. Aware that they are likely overwhelmed themselves. I stressed about when it was appropriate so send another follow-up email. What difference would it really make? I don’t want to be rude and pushy, particularly to the person who decides if I will be keeping my job
Funding renewals should be treated differently from initial funding. I know someone who got funding for a year to work on their project. They then reapplied for another year and got rejected. Imagine applying for a full time job. The hiring manager asks you one follow-up question via email but never interviews you for the job. Almost a year goes by and you have been working hard but still never even met your manager. You are up for a review, and a different manager reviews your written report. This new manager again never meets with you but decides to fire you with no severance payout. This may not be what is technically happening, but the emotional experience is far better expressed with “I got fired by someone I have never met” than “my tuning application was not accepted”
Region beta paradox. Fully fund or reject. Don’t under fund to see how it goes and then be disappointed it didn’t go well.
I was supposed to give some ideas that might lead to some positive change…
Grantmakers should have a hard due date. Prioritise communication over reducing total decision time. Prioritise reducing late-time over reducing total decision time.
If I were to be an applicant again I would set my own hard decision date. If I had not heard back from the grantmaker by that date then the next day I would send an email to the grantmaker to withdraw my application. Eliminating the uncertainty and inability to take actions.
If I were to run an AI Safety Fund
Fixed not rolling application rounds. Fund raising rounds tied to application rounds. So on day 1 you open applications and donations for that round. On day 90 you close applications and donation. You now have a fixed budget and a fixed pool of applications. No need to reject incase we get a better application next week. No need to delay an acceptance while hoping to get more funding next week. Fixed due date for decisions. This date should be embarrassingly achievable. At the end of the round we update all applicants and donors of the $$$ in applications and only $ funded projects. Somewhere between these two numbers is the funding gap. Rejected applicants now know they were rejected because of funding shortage and become donors or fundraisers themselves.
Applicants only need to be ranked relative to each other not against a funding bar.
I do agree the EA Funds made a mistake not returning to fixed grant rounds after the Mega Money era was over. It’s so much easier to organize, coordinate, and compare.
I think it’s actually better for applicant to have a deadline too. Plenty of people procrastinate on applying, some to the point of eventually not bothering at all. Also, if you are rejected but encouraged to reapply it’s clearer when it’s ok to apply. SFF has two rounds a year and I’ve applied, been rejected and then applied again the next round. I’ve probably applied more times to SFF than LTFF and this mostly comes down to there being an application deadline.
Also probably helps when getting rejected and comparing to other things you see get funded. If you see something a year ago that is just like what you were doing or seems much less value got funded it’s hard to understand that the funding bar might have been different then. It’s a much cleaner comparison when you are all in one group.
I think even for myself getting funded by SFF one round and not a later round. I think “oh, look at all these new projects that didn’t exist before that are so clearly awesome. Fair enough they don’t have funding for me any more”
A non-trivial fraction of our most valuable grants require very short turn-around times, and more broadly, there is a huge amount of variance in how much time it takes to evaluate different kinds of applications. This makes a round model hard, since you both end up getting back much later than necessary to applications that were easy to evaluate, and have to reject applications that could be good but are difficult to evaluate.
I’m interested to know why we haven’t seen Lightspeed Grants again?
Some of my thoughts on Lightspeed Grants from what I remember: I don’t think it’s ever a good idea to name something after the key feature everyone else in the market is failing at. It leads to particularly high expectations and is really hard to get away from. (Eg OpenAI) The S-process seemed like a strange thing to include for something intended to be fast. As far as I know the S-process has never been done quickly. It also seems to be dependent on every parameter being in place in order to be run so can easily be held up by something small.
At the time of applications it had a clear date when decisions were expected, this is much better than everyone else’s vague expectations. It ended up taking much longer than expected for decisions but still pretty quick overall. This was managed really well though. Certainly much better than LTFFs handing of a large volume of slow decisions last year. I judge the timing of decisions more on how late they are than the total time, but this is a proxy for the real problem which is uncertainty. Lightspeed got the basics right when it comes to comms.
I think it was really good over all and I’d expect that the major issues not accounted for in the first round can be managed. I would really like to see this run again or a variation of it.
You seem to be misunderstanding both Lightspeed Grants and the S-Process. The S-Process and Lightspeed Grants both feature speculation/venture grants which enable a large group of people to make fast unilateral grants. They are by far the fastest grant-decision mechanism that I know out there, and it’s been going strong for multilple years now. If you need funding quickly, an SFF speculation grant is by far the best bet, I think.
I think we generally stayed within our communicated timelines, or only mildly extended them. We did also end up getting more money which caused us to reverse some rejections afterwards, but we did get back to everyone within 2 weeks on whether they would get a speculation grant, and communicated the round decisions at the deadline (or maybe a week or two later, I remember there was a small hiccup).
Ironically one of the big bottlenecks was funding. OpenPhil was one funder who told us they wouldn’t fund us for anything but our LW work (and that also soon after disappeared) and ironically funding coordination work doesn’t seem to pay well. Distributing millions of dollars also didn’t combine very well with being sued by FTX.
I am interested in picking it back up again, but it is also not clear to me how sustainable working on that is.
What kind of numbers are we talking about needing here???
How much did it cost for the last Lightspeed round? How were the operations funded? How much distributed? $5mm?
How much would you need for operations in order to run it again? If you had the operations funding would the grants funding still be a problem? How much would you need for operations just for the grant decisions but not the distribution of funds?
Lightspeed Grants and the S-Process paid $20k honorariums to 5 evaluators. In addition, running the round probably cost around 8-ish months of Lightcone staff time, with a substantial chunk of that being my own time, which is generally at a premium as the CEO (I would value it organizationally at ~$700k/yr on the margin, with increasing marginal costs, though to be clear, my actual salary is currently $0), and then it also had some large diffuse effects on organizational attention.
This makes me think it would be unsustainable for us to pick up running Lightspeed Grants rounds without something like ~$500k/yr of funding for it. We distributed around ~$10MM in the round we ran.
I’m hesitant to ask you about this so feel free to pass. Can you say more about how it is that your current salary is $0?
I think most people would be surprised you are not currently receiving a salary. I also assume that as a not-for-profit founder even when you have had a salary it is lower than most or all of your team.
I donate more to Lightcone than my salary, so it doesn’t really make any sense for me to receive a salary, since that just means I pay more in taxes.
I of course donate to Lightcone because Lightcone doesn’t have enough money.
Yeah, I have personally appreciated the short turn around when needed. And seen plenty of situations where people need funds quickly.
I expect there are a lot of these tradeoffs. I think these should be solved by different services not by trying to solve all of the different types of funding together. LTFF ended up here for historical reasons, but now seems to be struggling to serve all of these markets while also crowding out any new funders in the space.
If I set up a new fund it would have rounds and I would just accept that this fund would not be able to support those kinds of applicants that need a fast turn around or longer investigations. A lot of the problems of the applicant experience with LTFF is uncertainty. Having one application form that try’s to solve all of these problems means that the expectations can’t be specific enough to be meaningful.
I spoke to another person just today that had a horrible experience with EA Funds.
I have mostly believed that despite its flaws, it is better to have the LTFF than nothing. I am starting to think that it is a a net harm.
Even considering that you could presumably choose not to apply? (I guess you think it is bad in a systematically surprising way).
To be fair, since announcing AISS shutting down, I am having a lot more conversations with people with negative experiences. So now I am comparing the counterfactual good to the bad. It is a big blind spot. People don’t make public posts about this. It’s no one’s job to collect all of the counter cases.
“Choose not to apply” is big part of the problem.
A person gets funded by LTFF but because they had a bad experience as an applicant they choose not to apply and put themselves through that experience again. They also choose not to apply or raise funds anywhere else. So we lost the value of that project continuing. Maybe they went on to do something pretty neutral or maybe they went on to work on capabilities.
People talk privately about this but not publicly. Which leads to others convincing themselves to not apply at all. Some potential donors have more people with bad experiences than good in their network and so don’t want to donate to LTFF. Yes, I’ve spoken to a few people in these cases and despite my own experience I encouraged people to apply anyway.
A core part of my decision to shut down AISS was that I didn’t want to have to be an applicant again. We had gotten funding from LTFF and SFF before but I had reached a point where if we had gotten the funding we were waiting for I would not be continuing after that funding had run out.
This is just some of the downstream effects of bad applicant experience for things actually funded by LTFF.
There are some issues with rejected applications too. Mostly related to the how the applicant and others update based on the rejection.
Project A get rejected, apparently because it might not be the best version of that kind of thing and could crowd out the market for a better version to come along. Someone sees the gap, starts planning Project B that could be better. They chose not to apply because in exploring the idea for their project they find out Project A wanted to do this but couldn’t get funding. Project B wrongly assumes that funders are not interested in funding this kind of project and so give up.
I think if we as the question “should the LTFF shutdown?” The answer is always “no, it clearly does good that no one else is trying to do”
But if you ask “if the LTFF never existed and you were designing a fund from scratch would it look exactly like the LTFF?” I think this is a clear “no” as well.
I don’t think LTFF is able to fix its current problems or be drastically modified to what you might have designed if it never existed. And of course it crowds out anyone else from solving this.
Lack of transparency around the full distribution of decision timeframes, and other aspects of the process, means it’s difficult to make an informed choice the first time around. And one bad experience can be enough to burn people long term, particularly if theyre making the reasonable assumption that LTFF is representative of the average funder.