Previously: Announcing Balsa Research, Why Balsa Research is Worthwhile
This is an FAQ for Balsa Research, based on the conversations we’ve had with lots of people over the project’s first few weeks. It was last updated on 10/17/22.
Q: This seems like it will be very hard?
A: Yes. It will be very hard.
Q: Don’t most policy projects fail?
A: Yes, they do.
Q: What will you do differently to succeed at this hard problem where most others fail?
Short A: Our focus will be on cutting the enemy, not growing, fundraising and hitting vanity metrics. We will be a full stack operation that backward chains from the outcome and coordinates all the steps to make it happen, including being able to support allies to encourage people to champion our causes and support good things, and we will work to make life easy for lawmakers and give then what they need and want in ways that are rare.
Long A: Good question. We plan to do a bunch of things differently, including but not limited to:
Backward chaining from success. Most projects don’t do this.
Focus on champions. As part of backward chaining, most lose sight of this. As an example, Scott Wiener was this for YIMBY.
Correct incentives. Most projects focus on enabling fundraising and growing, we won’t. We believe our motivations are unusually robust and that we can preserve that.
General interest motive. Having a general interest ‘special interest’ group helps solve the diffuse benefits problem for many useful actions. Someone needs to be the champion.
Academic studies and quantification of things that matter to legislators and key interest groups. Many things that would be very helpful simply aren’t being quantified right now.
Drafting legislation. If you don’t write the bill, don’t expect someone else to do it.
Full stack. Make everyone’s life easier and offer them things they need, be strategic.
Unique intellectual approach. Almost no one takes a similar intellectual approach to that seen on this blog, and this has already created unique useful resources that many people rely upon. If that approach can be scaled sufficiently, the sky's the limit.
Do something, not be someone. Credit matters for momentum effects but ultimately what matters is cutting the enemy, not who everyone remembers as doing the cutting.
The right moment. This is a unique moment both in terms of the power of LLMs/AIs, and also in terms of potential support for a policy initiative centered around growth and life, also known as The Abundance Agenda. As Sarah puts it in that post: This is the basic thesis of what Derek Thompson calls the "abundance agenda”, Ezra Klein calls “supply-side progressivism”, Noah Smith calls “new industrialism”, Katherine Boyle of Andreessen Horowitz calls “American dynamism”, and the Institute for Progress simply calls “progress.” The degree to which, after consideration, people embrace the underlying goals, and also see the need to push for them, is remarkable. So is the level of resources available on the sidelines that could be unlocked, but has mostly been rightfully disillusioned with past attempts.
More details are available throughout the first two posts, especially the second one, and this will be expanded upon over time.
Q: Won’t you probably fail anyway?
A: Yes, the effort probably fails. Most startups and ambitious projects fail. It is still worth trying. The willingness to accept this rather than create a story to claim a fake ‘win’ is part of why it might work.
Q: How big do you intend the policy arm (the 501c3) to get and how fast?
A: We intend to take it relatively slow. The policy branch will look to add a few full-time employees when we find the right people, and we’re not afraid to wait until we find the right ones, with best guess being a team of 4-7 full time people at the end of 2023.
Q: How big do you intend the tech arm to get and how fast?
A: We will treat this like a start-up, grow carefully, and not rush to hire dozens of people.
Q: How do you prioritize issues?
Short A: Best balance of biggest win versus difficulty of winning, with a focus on issues where the case seems indisputable and there are few if any real losers, and it would build momentum.
Longer A: Issue selection balances many criteria, with the most important being (1) how big a win is available and (2) the estimated difficulty of the win, where win includes both the direct physical impact of the change and the momentum effects it generates. We also want to (3) aggressively avoid overly partisan or culture war issues, especially when they represent value disagreements, and (4) want to start by focusing on areas where the win seems ‘clean’ and overwhelmingly clear, where the failure to have already addressed the problem is a visible sign of our civilizational inadequacy. We want win/win situations where there are few if any losers, and what losers there are could be reasonably compensated. Ideally the people opposing the change are making a mistake on their own terms. Also (5) we want to focus on where we would be a good messenger and can reasonably quickly and affordably ‘make the case.’ If we have strong expertise making itself available on an issue, it will get higher priority. Finally, one must admit there is some amount of (6) what catches our attention and makes us care. Some of that is that it will inspire others too, some of that is that motivation matters.
Q: Why haven’t you prioritized a small number of issues yet and deprecated everything else?
Short A: Too early. It’s too early to know the right targets, having preparations in place for a crisis/opportunity is valuable, and we believe it will be valuable to offer candidates and lawmakers (and the public) a full-stack set of policy positions.
Longer A: Too early. Three reasons. First, there has been a dearth of breadth-first searches done on the potential space of actions, including brainstorming interventions that have not been considered. The more salient an issue, the more partisan, the more forces likely aligned against you, and the harder to get things done behind the scenes, and the more likely there is a sideways/unconsidered solution out there. Second, because there is inherent value in completing the Official Policy Binder that for all topics explains the key considerations, provides sources and offers proposed policy positions, and having it as an available resource that offers a jumping off point and position on every question, and expanding and improving that over time. The goal is to make that a go-to resource. Finally, since background work such as commissioning papers that quantify and writing potential laws are often neglected and then become urgent needs in a crisis/opportunity, having such resources in reserve has a lot of value. When pressure rises and something must be done, you want to be the one offering a Something, which they therefore must do, either temporarily for the emergency or permanently. Life comes at you fast.
Q: Some of your posts framed your agenda in active terms. How much of your agenda is convincing the government to do new things versus convincing them to stop doing things they are already doing?
A: The bulk of the low hanging fruit is in convincing the government to stop doing harmful things it is already doing. That does not mean that such actions are trivial, or that they are shovel ready. One must still provide an alternative course of action. This new course must transition us from the current equilibrium to a new one, putting a new policy regime in place even if the final destination is mostly to not interfere. Without good answers about how that works, one won’t get buy-in. Without knowing who is in charge of getting the harmful thing to stop and how to get them to actually stop, your plan won’t work. When it involves convincing states or local governments to stop, it is even trickier. In other places, like tax policy, the line between doing things and not doing things can be highly unclear. There are also places where Balsa will advocate for genuinely new ideas or for doing additional things, which are anticipated to mostly not be the low-hanging fruit and thus not prioritized short term.
Q: What will be your top issue priorities for the coming year?
A: Jones Act repeal, NEPA reform and otherwise enabling new energy infrastructure currently seem like the most appealing initial targets, with Jones Act repeal (including the Dredge Act) as the medium-difficulty target and NEPA as the aspiration. Both are areas where there are clear cases to be made that aren’t being made and things to try that aren’t being tried, both are straightforward to address at the Federal level, and both are immensely valuable. That does not mean there aren’t better opportunities out there, and we encourage anyone to make the case for that. We will continue to gather information both on those and other fronts and hope to have a full (preliminary) prioritization list done by the end of the year.
Q: How do you overcome the problem of concentrated benefits versus diffuse costs when trying to do something like Jones Act repeal?
Short A: Fight fire with fire, you can be (and recruit) the champion on the other side, and the core actual concentrated benefits are often vanishingly small enough to make everyone whole.
Longer A: Fight fire with fire by being the other side of the coin that would enjoy a concentrated benefit, a special interest in opposition, that builds the Explicit Coalition of general interests to counter the Implicit Coalition of special interests that all work together to protect rent seeking and concentrated benefits in general. Also by buying off or convincing those who think they have concentrated benefits that they are wrong and should stand down, as even they get more benefit from ending the diffuse costs. For example, while it is reasonable for shipbuilders in particular to oppose Jones Act repeal for selfish reasons, unions in general should strongly support repeal and the shipbuilders can be made whole through navy contracts.
Q: Why are you focusing on federal policy/elections, when it may be easier to win and make change happen on the state level?
Short A: The stakes are higher, even relative to the additional difficulty. Comparative advantage.
Longer A: Our comparative advantage lies at the Federal rather than the state level, and the stakes are higher even relative to the level of difficulty. There is lots of value available at the state level and it is our hope that others will pick up that ball and run with it, but this is one place where we have a clear focus. A few intuition pumps behind this decision are:
We come from multiple states, my audience is mostly nationwide if not international, and those who want to help come from across the country. Focusing on any one state would be a poor fit for the coalition. Focusing on many states is a mess.
The state level is easier in some ways, yet it is harder in others. States are idiosyncratic and particular, and local special interests are more powerful there. If you want to attack concentrated benefits, often it helps to operate in a larger rather than a smaller sphere. Expanding the pool of decision makers strengthens the diffuse and general interests.
As a concrete example, the most encouraging state level win recently has been the YIMBY/housing legislation coming out of California. It is not a coincidence, in Zvi’s opinion, that this is happening in America’s largest state, where the state is almost its own federal level (California would be one of the world’s top 10 economies on its own, etc) to take on the locals, where everyone enjoys the benefits of housing built elsewhere. If California was split into several regional states things would be a lot harder. There are constitutional barriers that complicate any Federal effort to get more housing built where people want to live, but everyone getting to enjoy the benefits of everyone else building all at once gives us hope that turning YIMBY into a Federal cause could make things much easier.
State power in particular is far more tied up in local relationships and dynamics, and local interests, and individuals we do not know and have no connections into. That is not the game we are good at playing.
Many key issues can only be properly addressed at the Federal level, such as the FDA, NEPA or the Jones Act. The Federal level has constitutional checks on how much it can impact the states but the states have essentially no chance against the Feds.
Federal action impacts Presidential politics far more, and impacts the culture and zeitgeist more. These are key things we want to impact. State policies often won’t cross over as much as one might want or hope.
This is what seems exciting, we know it will be a new approach, and we see a path to success.
None of this means we won’t help in elections on the state or local level, or help with policy on those levels, if we see opportunity and/or are asked nicely, especially if the tools we already have built and evidence and work we have already assembled and done directly applies. There is far from zero temptation to get involved in NYC, SF or California, or to work somewhat with those doing so.
Q: How will you avoid losing epistemic rigor/becoming just another partisan advocacy group?
A: We are keenly aware of this danger. It is Balsa’s most likely failure mode. We are happy to make major sacrifices, in terms of ability to raise money and grow and notch short term optical wins, in order to avoid or at least postpone this. If it did happen, Zvi would hope he would notice, consider the project no longer interesting or worthwhile, and walk away. Careful and slow hiring with this danger in mind is a crucial aspect here, as is raising money with a keen eye towards those who would move incentives in this direction. Effort will continue to be made to build up a brand that is not compatible with partisan advocacy. If we became another partisan advocacy group, that would destroy what makes us special. Periodic reminders of the unique destructive madness of both parties should also be helpful.
Still, despite all this, the danger remains. Over a sufficiently long time horizon, especially over decades and as founder effects fade, things like this become increasingly inevitable. It is the fate of all organizations, in the end. When that happens, one must learn to start over again.
Q: What technology are you planning to develop, and how does it work?
A: We don’t plan to make our tech publicly visible, since we don’t want it to be used by those we do not want to help. However, in short, our first target is to use recent research in ML and data science to quickly identify and give advance warning of out-of-touch or counter-productive messaging. Eg. in 2020, the pro-choice group NARAL began supporting “defund the police”. NARAL’s core message was popular, but it eventually became clear that “defund” was politically toxic and caused widespread blowback. Our goal is to forecast such outcomes, not just before the next election cycle, but before an idea has left the drawing board.
Q: Can I contribute to Balsa in a particular policy area I’m interested in (for example, FDA reform)?
A: Yes! Or at least, we’d love to have you give it a shot. A lot of people with policy interest or expertise have already reached out to us via our Google form, and we encourage more people to do the same. In case it was not obvious, the only way Balsa is going to cover its bases is by getting this kind of help. Right now, what is most scarce is force multiplication and coordination, and figuring out the best ways for outsiders to usefully investigate matters and share what they know, and we are eager to run a bunch of experiments on this on those who are up for it.
Q: I’m a programmer or a data scientist, how can I help?
A: We expect to have full-time job postings for technical roles later on as we build the organization. However, don’t feel forced to restrict yourself to roles that focus on data science if your interests lie elsewhere. Having the data science skills and cultural code will be useful to everything we do, so if there’s an area you’d like to write about, investigate or otherwise explore, that’s great. If you’d like to focus on data science, then there will be plenty of straight data science puzzles to solve and data to analyze (and seek out!) as well, so if that’s where your head is at, let us know and we’ll reach out when we have a potential good fit.
Q: It seems like your goals are very broad and ambitious, how will you accomplish that with a smaller team? Do you plan to narrow down your focus?
A: While we plan to express a wide variety of views, we do intend to narrow our focus in the sense that we will be selecting key areas in which to go in more depth, commission studies and hire lobbyists - we don’t intend to and couldn’t grow the team to the point where we could do that across the board. We also intend to work with volunteers and academics, and commission outside help. Ultimately the goal is not to have Balsa in particular cause most or all (or even that many) of the desired changes to individually happen. To the extent there is a goal, it is to get some of them, worthwhile in their own sake, and encourage others to see that it is possible and use that momentum.
Q: There’s only two years until the 2024 election, how would you have significant influence in so short of a time period?
A: It is indeed unfortunate that there is not more time before the 2024 election, which restricts the range of plausible actions. In particular, it is likely too late to recruit a new candidate. It is decidedly not too late to take measures to materially improve the experienced economic conditions on election day, if one could convince the Biden administration’s decision makers or pass relevant laws. Economics is forward looking, so things that improve future economic conditions would improve conditions in many ways now, in advance. Other changes, if achieved, would also be anticipated and felt right away - consider how quickly Dobbs impacted the country. However, it is true that this operation will take time to scale up, to do its homework and then to build relationships and talk to the decision makers, and then for its champions to do their work, so a lot of this is taking a longer view.
What it is definitely not too late for, in terms of 2024, is technology and other ways one can directly assist campaigns. Many aspects of general election campaigns do not even begin until after the primary is over, and many more don’t happen until after the midterms. Our studies will help figure out what things that would be good are also popular and can help win campaigns, and how to present them so they are helpful, and this in turn can also help some people decide who to support, including in primaries.
Q: What will you do about AI policy/existential risk concerns?
A: The sad truth is that, while a lot of work has been done trying to figure it out, no one has a good plan for AGI risk from a policy perspective. We are hopeful about export controls of AI hardware and compute governance as being better than nothing, but that is nowhere close to a full plan. We have asked those we thought were most likely to know, and they don’t know. We are very open to pitches on this. If good solutions are found, then the right way for Balsa to contribute is to include such policies within a broader platform, and allow others who have more resources and a more narrow focus on such topics (such as OpenPhil or Protect Our Future) to do the lobbying and other such work. In terms of pandemic preparedness and other non-AI existential risks, we know more about what would be actionable and worthwhile, and we fully intend to include those in our platform in ways we expect will be broadly compatible with Protect Our Future’s proposals. Again, we intend for others to take the lead there, unless we gain connections and opportunities that are unique to us.
Q: Is this an effective altruist project?
A: No. While all the core members know many EAs, and no doubt some who consider themselves EAs will become involved, and some EA donors may choose to fund us, we do not consider ourselves EAs nor do we consider this an EA project. Of course, if we did not think what we were doing was worthwhile and effective, we would not be doing it.
Q: How are you structured legally?
A: There will be multiple legal entities. Balsa Policy Institute will be a 501c(3) and be the focus of our policy work. There will in addition be a public benefit corporation in which we will do our technological work, because once something gets into a 501c(3) you cannot get it out again, and this would substantially impair our freedom of action. We anticipate a third entity in the future to help with campaigning.
Q: Will you be affiliated with a particular party?
A: No. One of our top priorities, as noted above, is to avoid becoming one more partisan advocacy organization.
Q: How will you be funded?
A: The policy arm is a 501c(3) and will be funded by donations. The tech project will be a public benefit corporation and thus be funded by investment rounds.
Q: Do you accept donations?
A: Yes, however we are still in the final stages of setting up the legal entities involved as of 10/17/22. Stay tuned, or contact us for details.
One thing that might help a bit is to clarify what the enemy is that you intend to cut. What, as opposed to who, I expect is the right interrogative there. Really focus on the fact that you don't care who makes the improvements, so long as they get made. That will help the non-partisan mission aspect I suspect, a bit like google's old "Don't be evil" line. Being able to seriously ask "Are we supporting a policy or a person/party?" and demand answers might be very valuable.
Anything more you can share about the tech arm?
Regarding NEPA reform, one angle to consider is that really we should be able to have stronger environmental protections and expedited permitting if we had a less insane process. It's really rather incredible to use the legal system as the primary means of adjudication. Many other countries have that as a ministerial rather than legal process and are still able to build stuff. That political economy angle is important to consider because there's a lot of interests vested in the NEPA status quo and one way to address that is to not be gutting a critical environmental law but rather modernizing it.