I wanted to write a summary comment to close this out and clarify a bit more what I am trying to (not) get at I (still hope to be able to address all detailed comments, probably on the weekend, as I am doing this in personal capacity):
1. With re-examining work on systemic attributes I don't mean “systems change, not climate change" style work, but rather something small-c conservative -- protecting/strengthening the basic liberal norms and institutions such as rule of law, checks and balances, etc. at home and the rule-based international post WW2-order and a basic commitment/ norm to a positive sum view of the world globally.
2. My basic contention is that -- when many of those institutions are under much more threat and are much more fluid than before -- working on them is relatively more important, both because greater volatility and more downside risk but also because more surgical interventions are affected by this.
Somewhat crudely, all work that flows through influencing Congress to spend more money on priority X, requires a continued respect for Congress’s “power of the purse" (no impoundment). Similarly, the promisingness of much GCR work also seems heavily affected by macro-level variables on the international scale.
3. It would be good to examine this more thoroughly and see whether there are things we can do that are highly effective on the margin and doing so would require a serious analytical and research effort, not relying on cached priors on system level v surgical interventions debates of days past.
To be clear, I am fairly agnostic to whether this would lead to an actual reprioritizing or whether the conclusion would be that engaging on system-level factors is not promising. I do not know.
Insofar as I am criticizing, I am criticizing the lack of serious engagement with these questions as a community, a de facto conclusion on this question -- do > 95% work surgical work -- that rests on little serious analysis and a lack of grappling with a changing situation that, at the very least, should affect the balance of considerations.
4. In terms of taking action, I would be surprised if the conclusion from this would be -- if more action is warranted -- to simply increase the effort of existing EA(-adjacent) effortson those topics such as around advocating for electoral reforms. It is obviously important to advocate for changes to electoral systems and other institutional incentive structures, in particular if those have properties that would address some of the existing problems.
However, it seems clear to me that this cannot be everything EAs would consider doing on this. By crude analogy, much of these discussions feel like spirited discussions about which colors to paint the walls in the kitchen while there is an unattended fire in the living room. In the same way that our primary strategies on engaging on AI risk are not 30-year strategies to change how technology is governed, seriously engaging on preserving desirable system level attributes / institutions cannot only be about very long-run plays in a time where prediction markets predict a 3/4 chance of a constitutional crisis in the US over the next couple of years and the international situation is similarly fluid.
5. I also do have “this is not neglected” and “this is intractable” in my head as the primary reasons why we should not do this. However, I (and I think many others), have also become a lot more skeptical of using these considerations lazily and heuristically to discredit looking into entire fields of action that are important.
It is certainly true that the average intervention on vaguely improving institutions in a way that is salient with the public already will have a low impact. But it would not shock me at all if a serious research effort found many interventions that are surprisingly neglected and quite plausibly tractable.
I think the analytically vibrant community we’d ideally like to be would dive deeper into those issues at this point in time.
Quick points: 1. I've come to believe that work in foundational political change is fairly neglected, in-comparison to its value. 2. As Scott Alexander wrote, political donations are surprisingly small for their impact. This seems especially true for someone as radical as Trump. 3. Related, the upper-class has been doing fantastically these last 10-30 years or so, and now has a very large amount of basically-spare capital. 4. I very much expect that there could be arrangements that are positive-EV to groups of these wealthy individuals, to help us have better political institutions.
So a corresponding $10T+ question is, "How to we set up structures whereby spare capital (which clearly exists) gets funneled into mutually-beneficial efforts to improve governments (or other similar institutions)"
A very simple example would be something like, "GiveWell for Political Reform." (I know small versions of this have been tried. Also, I know it would be very tough to find ways to get people with spare capital to part with said capital.)
I wrote one specific futuristic proposal here. I expect that better epistemics/thinking abilities will help a lot here. I'm personally working on epistemic improvements, in large part to help with things like this.
I found that linked post from Scott Alexander quite interesting, but it seems like the numbers are no longer up to date. The paper he cites is from 2003. I think the political landscape has changed quite a bit since then. If I had to guess, I'd say political donations have become larger. It would be interesting to see more recent figures and if these are still small for their impact.
I agree with this, yet im yet to hear of concrete ideas which could have significant impact at the kind of high level you are taking about. It would be great to see some ideas fleshed out here on the forum.
I've always been a huge advocate for the cost effectiveness of tractable systematic change on the margins, like lead paint policy and my wife's work to ban certain types of alcohol in Uganda.
But at the really high level you talk about, the "fractured democracy" level I struggle to see where we could have clear impact. It might be one of the least "neglected" areas around, which of course doesn't mean there can't be niche highly cost effective areas we could move the needle
Like you suggest there's also the counterfactual of what if we had been putting hundreds of millions into systemic changes that were then just deleted by the current world order situation. There are good arguments both for and against working on specific programs vs. Systemic change right now. At least the nets still save lives cost effectively as global conflict becomes more likely and global aid is slashed? Given the reduction in aid money donations in global health, donations become a bit more impactful now too.
It's not as if people haven't been thinking and even investing along these lines at times. Open Phil have thrust a bunch at "global economic stability", but this isn't an area I understand well and feel I can judge well.
There have been a few interesting posts on the forum such as this one recently, but I'm yet to hear convincing ideas - which is not to say those ideas aren't there.
I provide evidence for the following claims on the adverse effects of looser macroeconomic policy:
Very strong claim: The first-order effects of too loose policy on the '21 margin were unequivocally bad and will leave most worse off, even the poor, because we way overshot on inflation. Policy in ‘21 added few jobs and real wages are down even for the poor (who suffer the most when their real wages go down even by a little). The second-order effects are that Dems will lose House, Senate, and Presidency. Populists (like Trump) will be elected, whose incompetence will make things even worse. The effects of looser policy in richer countries spill over into poorer countries, where inflation is higher, more persistent (hyperinflation), which leads to less growth, wellbeing, populism and instability.
Medium strength claim: Looser policy on the '21 margin was good for the poorest 10% Americans (~30m) as it created more jobs and higher wages, but bad for the middle class (~150m) as their real wages went down. The first-order effects for wellbeing are still positive on average, as the poorest benefit much more than the middle class lost (a simple rule of thumb is that $1 is worth 1/X times as much if you are X times richer and poorest Americans are more than 5x poorer [utility is logarithmic in consumption]). However, because 150m have lost in terms real income, which can predict midterms and naïve extrapolation of current income decline predict Dems losing 50+ seats, the resulting populist incompetence will have, at the very least, high downside risks (like Trumpists starting trade wars). This will harm everyone (in expectation), even the poorest, in the longer-run.
Weak claim: Looser policy, even on the ‘21 margin, was good for all Americans as it created jobs, increased real wages for most Americans (at least over the medium run) and led to GDP growth in the US and even for the world economy. However, even if real wages increased and inflation is transitory and mild over the medium-run, because voters are too sensitive to high short-term inflation, and inflation coincides with the midterms, and think it’s more persistent than the market, Dems will lose at least the House. Renewed gridlock and polarization will have bad effects and we could have avoided this by moving more slowly to a higher inflation target and being more expansionary under the curve in the longer-term.
I also highlight some general issues with macroeconomic advocacy like reducing central bank independence, problems with lobbying foreign central banks, and having to be very careful and do a lot of analysis before making grants that might have a lot of leverage and are not robustly good (like improving health).
Wow amazing 2022 article by the way. I never saw it at the time, it was before I was on the forum!
After reading the thread I might fall marginally more on your side of the argument there, especially as inflation probably did contribute to the trump victory. But it is also yet another demonstration of how hard philanthropy gets, the higher level you get in politics or economics, with so much disagreement and uncertainty. There's just so much disagreement from experts on almost every major issue, so it's very hard to know on which side to push the money.
Taking a low percentage "hit based" approach on human welfare issues is one thing, but when it's super unclear even if you make that hit whether its positive or negative EV is where I start to think why not just take a punt on something deeply uncertain but never negative EV like shrimp instead.
I've now tried to clarify what I mean in my post, Nick.
I agree with you that concrete suggestions are lacking, my claim is that this is -- at least partially -- due to too little effort on this angle and that this seems worth re-examining in a change of rapid and profound system-level changes.
Sorry also to clarify I wasn't saying you need to provide suggestions, more that there have been a few posts along these lines without any concrete suggestions yet and id love to see some.
Love the post a lot by the way, thoughtful and balanced. Nice one!
Why do you think improving democracy is intractable? None of the highest priority world problems are tractable.
Risks from artificial intelligence - nobody knows if an AI safety solution is even possible yet that warrants hundreds of millions in funding
Catastrophic pandemics - Preventing/mitigating pandemics is a trillions dollar endeavor. Incredibly costly. How is this tractable?
Nuclear war - Exactly how is this cause area tractable?
Factory farming - Good luck on this cause, especially without the force of an enlightened government to demand change.
Comparing to the toughest problems, how is improving democracy intractable? Of course, tractability needs to be balanced with importance and neglectedness.
Developing strong evidence that some specific reform (ie maybe sortition) could be a real improvement could be done in the millions of dollars range. That could be cheaper than training your LLM. That's definitely cheaper than fusion power.
What is the value to humanity of learning what kind of governments are best? Even in the short term perspective, the value of an improved government could be trillions of dollars of tax dollars saved. In the long term perspective, every top priority world problem would immensely benefit from enlightened governance.
Sortition as a specific reform might be slightly harder to implement on some political campaign, yet imagine hypothetically sortition yields 10% greater ROI in taxpayer benefits whereas ranked choice or approval voting might yield closer to 0%. Of course we don't know the numbers, and that's a huge problem. Ranked choice might be more tractable, yet it also might be mostly useless.
Yet we don't know, because nobody is doing any testing, there's no empirics and I bet, there's no funding.
The campaign for ranked choice voting-style reforms in 7 states cost nearly $100 million, and failed in 6 out of 7 states (it was narrowly protected in Alaska) in 2024. The linked article is a decent description.
Ranked Choice Voting is a good way to reduce polarization in politics, elect more popular (and less extreme) candidates, and increase competition. It would also reduce the power of Trump over the Republican Party, which could lead to more Congressional pushback.
Despite the disappointing 2024 results, I believe there significant opportunity in 2026: the midterms have a more politically engaged turnout, and given the current situation, voters might be more open to reform.
There are alternatives to Ranked Choice Voting (like approval voting) and I'm no expert on them. It does seem like, if this campaign were to run for 2026 it would need to start soon.
There's also a decent chance that it would be perceived as very hostile by the current administration, and retaliation could do significant damage to the community or the specific funders behind a campaign
I very narrowly leaned yes on this (rather than a clear yes), because the Top Four Ranked Choice as proposed, would have only allowed political moderates to even make it to the ranked choice election.
I still generally support RCV and there's definitely other reasons why these initiatives failed (like being anti-endorsed by incumbent/establishment politics) but wanted to point out that specific implementations of RCV aren't always in favor of increasing democracy and this may be part of why many initiatives failed.
I think it's similar to the FPP dynamics you're referencing. Essentially, it's most advantageous to vote for candidates with broader appeal in the hopes they actually make it to the general election.
Ranked Choice Voting is a good way to reduce polarization in politics, elect more popular (and less extreme) candidates, and increase competition. It would also reduce the power of Trump over the Republican Party, which could lead to more Congressional pushback.
Since the form of RCV used in the US eliminates candidates bases on plurality tallies of first-choice votes in each round, it arguably does none of the above and just perpetuates the problems of our current system. It does not fix vote-splitting or the spoiler effect, despite the claims of its proponents.
Ranked-choice systems that actually count all voter preferences ("Condorcet-compliant") would actually improve these things, as would cardinal systems like Approval Voting, STAR Voting, Balanced Approval, etc. that allow voters to evaluate each candidate independently.
Glad to see you raising this. I raised a related question here (has a slightly more US-centric angle to it). In that post I do suggest some interventions, but there's not a lot of careful research behind it.
What I am trying to say is that this being one of the longest treatments on this to exist feels like a failure / blind spot of the community.
We're in the midst of very severe systemic changes, domestic and international, and -- ideally -- there'd be lots of thorough analysis on the forum and elsewhere.
approval voting should be far and away the top priority IMO, given that social utility efficiency computer simulations show it would massively improve human welfare. and it only has a one-time cost per implementation, and then it's "free" forever.
another option is instant runoff voting, a form of ranked voting that's commonly just called "ranked choice voting". but it's less accurate, more expensive, and appears to be losing political viability. as someone else here noted:
> There are alternatives to Ranked Choice Voting (like approval voting) and I'm no expert on them. It does seem like, if this campaign were to run for 2026 it would need to start soon.
in addition to that, i would say the next most high-priority issue, based on the EA metrics of neglected+tractable+impactful, would be election by jury.
My rough sense is that one reason for EA's historical lack of focus on systemic change is that it's really hard to convert money to systemic change (difficult to measure effectiveness, hard to coördinate on optimal approach, etc.). On the other hand, I do think that this leads to an undervaluing of careers that work in systemic change (and important considerations that cross cause areas, since they're also hard to donate to). This might not be true if you have AI timelines too short for systemic changes to come into being.
Not super confident about this, though. Feel free to try to change my mind.
I like your point about careers and systemic change, and that it is harder to convert money directly into results. I also agree with @jackva though that measurement problems aren't a likely reason for the lack of investment, Open Phil are investing in lots of speculative and almost impossible to measure things, I don't think that's the issue.
Thanks! I don't think that hard-to-measure explanation is quite right -- lots of other similarly speculative / hard-to-measure interventions that EAs have been traditionally very excited about.
I think it has more to do with priors of low neglectedness and low tractability and a certain aversion to act in ways that could be seen as political.
That said, my goal here is not to re-litigate the whole "surgical v systemic change" debate, but rather to say that current changes seem to suggest that systemic work should be relatively more important and it's something that seems (vastly) under-discussed and not systemically explored.
I agree that "systemic change" (broadly speaking) seems highly intractable for EA and that it does not seem to be that neglected outside of EA.
However, one possible advantage to this lack of neglectedness is that there are probably plenty of people already working on, or interested in, some forms of systemic change who are already sympathetic to EA ideas but don't know how to translate that into action. A lot of those people may also be mid-career, like me.
EA might be able to play a useful coordination role between such practitioners and publicise ideas that could improve their effectiveness with relatively little resources. There are plenty of common "best practices" like red-teaming, Delphi method, or reference class forecasting that can help decision-making but are still not widely used in much of government (at least in my experience, though I'm sure it will depend on the country involved). Of course, there is already a lot of material out there already on "how to be more effective" but much of it is either (1) low-quality; (2) focused on business and personal success; and/or (3) rather expensive (for courses aimed at business leaders).
Even something like mildly improving the epistemics (and perhaps morals) of people who are already in positions to shape important policies could be quite impactful, though very hard to quantify. I personally know many people in policy who genuinely want to do good but would not be willing to switch to one of EA's cause areas. It seems like EA's message to such people is basically just "earn to give", which feels like a waste. That's not to denigrate earning to give at all - it just feels like there is more potential there that EA is not tapping into.
My comments are more about EA turning to focus on system-level interventions in general, rather than to address the problems arising in this particular time of flux. But, you know, even if the best time to plant a tree was 20 years ago, it could still be worth planting today.
First and foremost, I agree with the point. I think looking at this especially from a lens of transformative AI might be interesting. (Coincidentally this is something I'm currently doing using ABMs with LLMs)
Dropping some links below, I've been working on this with a couple of people in Sweden for the last 2 years, we're building an open source platform for better democratic decision making using prediction markets:
One challenge here is that many systematic changes take time and so some desirable changes might take long enough that we'd only be able to implement them past the point where it would be useful.
That seems true for most things EAs fund apart from direct service delivery interventions such as distributing malaria nets.
I.e. it is a valid consideration but it is not a justification to work on surgical instead of systemic interventions in areas where all interventions are operating uncertainly over multi-year indirect theories of change (the majority of what EAs do outside GiveWell-style GHD work).
[Edits on March 10th for clarity, two sub-sections added]
Watching what is happening in the world -- with lots of renegotiation of institutional norms within Western democracies and a parallel fracturing of the post-WW2 institutional order -- I do think we, as a community, should more seriously question our priors on the relative value of surgical/targeted and broad system-level interventions.
Speaking somewhat roughly, with EA as a movement coming of age in an era where democratic institutions and the rule-based international order were not fundamentally questioned, it seems easy to underestimate how much the world is currently changing and how much riskier a world of stronger institutional and democratic backsliding and weakened international norms might be.
Of course, working on these issues might be intractable and possibly there's nothing highly effective for EAs to do on the margin given much attention to these issues from society at large. So, I am not here to confidently state we should be working on these issues more.
But I do think in a situation of more downside risk with regards to broad system-level changes and significantly more fluidity, it seems at least worth rigorously asking whether we should shift more attention to work that is less surgical (working on specific risks) and more systemic (working on institutional quality, indirect risk factors, etc.).
While there have been many posts along those lines over the past months and there are of course some EA organizations working on these issues, it stil appears like a niche focus in the community and none of the major EA and EA-adjacent orgs (including the one I work for, though I am writing this in a personal capacity) seem to have taken it up as a serious focus and I worry it might be due to baked-in assumptions about the relative value of such work that are outdated in a time where the importance of systemic work has changed in the face of greater threat and fluidity.
When the world seems to change in rather fundamental ways, we should seriously re-examine whether intuitions and priors generated under different conditions still hold and this requires, I would contend, a more serious analytical effort.
Some stylized examples to clarify the dynamic I am positing
There are two key claims I am trying to make:
(1) Systemic interventions are becoming more important relative to surgical interventions because (a) the system-level is much more in flux than it used to be and (b) many typical interventions might depend on system-level characteristics that are becoming uncertain or cannot be taken for granted anymore.
To give five stylized examples, they are meant to be illustrative not precise or necessarily the most pressing, they are selected for public salience and understandability:
In case this appears as though the system-level characteristics are myriad and unconnected, I think that’s not so – most of them can be traced to a couple of key system level qualities such as, domestically, protecting rule of law and checks and balances and, internationally, key principles of the post-WW2 order.
Given they are now much more uncertain, this undermines the effectiveness of many typical interventions (b), but it also, because they are in flux, possibly raises the importance of engaging on them (a) in absolute terms.
(2) It is quite possible that prior-based judgments on this lead us astray, especially if informed by intuitions before recent changes. This is why I think we should explore more systematically and rigorously whether the balance between surgical and systemic interventions should shift or whether changes in relative importance are not sufficient to make up for differences in tractability and neglectedness.
What next steps could be
To clarify, I am not a likely person to be able to carry this forward – I mostly wanted to raise this as an issue to evaluate for people who have or allocate research / analytical capacity with some flexibility.
Here are some ideas that I think could be valuable along the lines of what is discussed here:
Explorations of existing models of direct and risk-factor work (e.g. Ord’s model from Precipice) to see whether current shifts plausibly seem relevant updates in favor of more systemic work
Systematic investigations by major EA(-adjacent) research orgs on those topics; e.g. work on institutional quality that is informed by recent developments (a lot of EA work on institutions seems to be about marginal improvements, such as different voting rules, under overall very stable institutional conditions; not “meeting the moment” of a time where many institutions and norms are questioned to a degree unparalleled for decades)
Deep dives on 80k podcast on what is happening in the US; and what is happening with regards to geopolitics and how these changes might affect things EAs care about
This post presents the executive summary from Giving What We Can’s impact evaluation for 2025. At the end of this post we share links to more information, including the full report and...
I used AI to fix transcription errors, rerrarange the ideas, and suggest tweaks to the title and some sentences.
Three of the most exciting projects to come out of EA in recent years are, in a vague sense, CEA spinouts:
* Kairos is directly a spinout of CEA and now handles most support for university AI safety groups. Basically everyone I've found who knows them is really excited about what they do
* NEST is an opinionated ideas-fi...
Very grateful for the amount of discussion here.
I wanted to write a summary comment to close this out and clarify a bit more what I am trying to (not) get at I (still hope to be able to address all detailed comments, probably on the weekend, as I am doing this in personal capacity):
1. With re-examining work on systemic attributes I don't mean “systems change, not climate change" style work, but rather something small-c conservative -- protecting/strengthening the basic liberal norms and institutions such as rule of law, checks and balances, etc. at home and the rule-based international post WW2-order and a basic commitment/ norm to a positive sum view of the world globally.
2. My basic contention is that -- when many of those institutions are under much more threat and are much more fluid than before -- working on them is relatively more important, both because greater volatility and more downside risk but also because more surgical interventions are affected by this.
Somewhat crudely, all work that flows through influencing Congress to spend more money on priority X, requires a continued respect for Congress’s “power of the purse" (no impoundment). Similarly, the promisingness of much GCR work also seems heavily affected by macro-level variables on the international scale.
3. It would be good to examine this more thoroughly and see whether there are things we can do that are highly effective on the margin and doing so would require a serious analytical and research effort, not relying on cached priors on system level v surgical interventions debates of days past.
To be clear, I am fairly agnostic to whether this would lead to an actual reprioritizing or whether the conclusion would be that engaging on system-level factors is not promising. I do not know.
Insofar as I am criticizing, I am criticizing the lack of serious engagement with these questions as a community, a de facto conclusion on this question -- do > 95% work surgical work -- that rests on little serious analysis and a lack of grappling with a changing situation that, at the very least, should affect the balance of considerations.
4. In terms of taking action, I would be surprised if the conclusion from this would be -- if more action is warranted -- to simply increase the effort of existing EA(-adjacent) efforts on those topics such as around advocating for electoral reforms. It is obviously important to advocate for changes to electoral systems and other institutional incentive structures, in particular if those have properties that would address some of the existing problems.
However, it seems clear to me that this cannot be everything EAs would consider doing on this. By crude analogy, much of these discussions feel like spirited discussions about which colors to paint the walls in the kitchen while there is an unattended fire in the living room. In the same way that our primary strategies on engaging on AI risk are not 30-year strategies to change how technology is governed, seriously engaging on preserving desirable system level attributes / institutions cannot only be about very long-run plays in a time where prediction markets predict a 3/4 chance of a constitutional crisis in the US over the next couple of years and the international situation is similarly fluid.
5. I also do have “this is not neglected” and “this is intractable” in my head as the primary reasons why we should not do this. However, I (and I think many others), have also become a lot more skeptical of using these considerations lazily and heuristically to discredit looking into entire fields of action that are important.
It is certainly true that the average intervention on vaguely improving institutions in a way that is salient with the public already will have a low impact. But it would not shock me at all if a serious research effort found many interventions that are surprisingly neglected and quite plausibly tractable.
I think the analytically vibrant community we’d ideally like to be would dive deeper into those issues at this point in time.