Concerns with the Wellbeing of Future Generations Bill

By Larks, John_Myers @ 2022-03-09T18:12 (+126)

Summary

This is a critique of the current draft of the UK Wellbeing of Future Generations Bill, a recent attempt to influence policy, supported by some EAs. The Bill imposes a range of new duties on public bodies relating to future generations, ‘sustainable development’ and ‘wellbeing’. We support the underlying goals of the authors of the Bill, but believe that if the Bill were passed as currently drafted, it could significantly damage short and long term welfare. Our concerns only relate to this Bill as currently drafted; we believe in general that it is important for EAs to work more on improving policy in both the long and short terms, and John is involved with a new initiative to help with that. Later in this post we set out thoughts on other approaches which could be effective for achieving long term policy goals.

We do not see compelling evidence that simply passing a law that says there is a requirement for long-term thinking, goals or metrics will necessarily lead to better or more long-term thinking in a way that overall benefits future generations. Vague language, subversion of goals, ignoring or simply failing to improve metrics, contrary incentives of policymakers and the law of unintended consequences are all large risks, in light of the history of well-meaning legislative initiatives that have caused damage to long term welfare. 

We think there is substantial risk that the Bill could damage the welfare of future generations. Simply passing a law that the government must do ‘good things’ does not in fact ensure that it will do good things. If the goal of the Bill is circularly defined as imposing the duties and other legal measures it proposes, then it will do that. But we have no confidence that, as drafted, it will benefit future generations.

Introduction

The Bill establishes a set of national wellbeing goals, formulated by the Secretary of State with public consultation. It places various duties on public bodies and creates a new requirement of future generations impact assessments. It also requires a futures and forecasting report with a 25 year time horizon. It empowers the head of the National Audit Office to examine whether public bodies have followed their wellbeing duties. It also suggests a new Joint Committee on Future Generations in Parliament to review future legislation, and a new Future Generations Commission. There are numerous other provisions. It is based on the text of, and lessons learned from, the Well-being of Future Generations (Wales) Act 2015.

We enthusiastically endorse the goal of making governments’ actions and perspectives more long term in outlook. Government action is often characterized by a distressingly short term perspective, as shown by the multiple suboptimal responses to the COVID-19 pandemic, and indeed in the multiple failures in pandemic preparation before that. 

We also welcome the increasing interest of the EA movement in improving policy and governance in a non-partisan way, as we think these areas are neglected, potentially high impact and more tractable than is generally understood.

However, we are concerned that the proposed Wellbeing of Future Generations Bill may not achieve its stated outcomes. Given the current state of knowledge, we think there is a high risk of unintended and counterproductive consequences from actions based on superficial analysis. We think that the analysis should be taken to a deeper level before action is decided upon.

We know a number of people with experience in government and policy who have expressed the view that perspectives in the EA movement can sometimes – with the very best of intentions – be naïve in terms of the expected practical outcome of particular proposed policy actions, and we have sympathy for this view.

We think that many of the goals of various movements are substantially aligned, and that better discussion between them could improve their impact. We hope that this piece can help to encourage that discussion.

We also acknowledge uncertainty about the extent to which EAs were involved in or able to influence the drafting of this Bill; our understanding is the Bill changed significantly from the initial version, and we have also seen different accounts of how much EAs were involved with that initial version. It can be valuable to try to change a potentially damaging Bill in a less damaging direction, and that may have been the goal of some EAs involved. We intend no criticism whatsoever of any individual or group actions. Our concerns are purely with the text currently proposed as a potential law.

Concerns about the Bill

In our view, the proposed Bill unfortunately falls into the classic policy trap of having broad and well-meaning goals without any plausible theory of change as to how those goals will be advanced by the Bill.

Existential risk neglect

It is worth noting that the Bill's operative clauses do not directly mention existential risks at all. Implicitly these are included as ‘high-impact, low-probability risks’, but given their importance and historical neglect by policymakers we think that they should be explicitly mandated. (The term is mentioned in the explanatory notes but these have little legal weight.) CSER noted in 2018 that under the Welsh predecessor Act “it does not seem that global catastrophic and existential risks are being considered”.

Some aspects of the Bill do seem useful; for example the published risk assessment in section 16(1)(c)-(e):

The Secretary of State must … publish a futures and forecasting report that contains—
…
(c) an assessment of risks, including high-impact, low-probability risks, environmental risks, global risks and risks that may emerge or grow in the future, for at least the next 25 years; 
(d) each department’s plans to prevent, manage and prepare for the potential consequences of the identified future risks; 
(e) details of the assessment process used to assess the future trends and 30 risks

could be useful. However the Cabinet Office already publishes risk assessments (which proved totally inadequate to deal with COVID-19 risk, something that should have been squarely within their two-year time horizon): there is an inherent problem when such reports are published by entities that are subject to short-term political pressures. We think the CLTR suggestions for three lines of defence against x-risk, including a Chief Risk Officer and Office for Risk Management, are a good point from which to start to develop fully workable proposals. Their approach is similar to the successful model used by financial institutions and other major organizations, where this C-suite role can serve as an influential voice for risk control.

Vague language

In general the Bill seems in many places quite vague and imprecisely worded. This is concerning, because it makes it more likely that the Bill will have unintended consequences by blocking things that were not intended to be blocked, or requiring things that were not intended to be required. Such lack of clarity also imposes risks, costs, and a significant chilling effect upon investment and other societally beneficial activity. The public bodies upon whom the Bill imposes unclear duties must take action of their own, and make decisions about third parties. If the Bill is unclear, that will make it harder for the public bodies to take action, and make it harder for third parties to plan their own socially beneficial actions because they will not know exactly how rules they must comply with will be interpreted.

For example, consider the ‘sustainable development’ duty the Bill introduces on public bodies:

In this Act, any reference to a public body doing something “in accordance with the future generations principle” is a reference to the body acting in a manner which seeks to ensure that the needs of the present are met without compromising the ability of future generations to meet their own needs. 

Protecting the needs of future generations definitely sounds good, but it’s not entirely clear what this actually means in practice. Does the test allow netting of costs against benefits? Does it permit the weighing of short term damage, e.g. the loss of a small area of natural habitat, against all of the healthcare, dietary and other benefits of more investment in infrastructure? The words ‘without compromising’ seem to imply an absolute test. That would be profoundly damaging to welfare: if something imposes only very small costs on the future, but gives large benefits now, we should be as happy to do it as we would be to suffer a small cost now to give future generations a large benefit. On one interpretation, the test may not even net across different areas of future generations’ abilities, so it would not allow us to slightly impair their ability on one axis while substantially improving their abilities on other axes, even if that would overall make them substantially better off. That at least should be made unambiguously clear.

It is not enough that we can work out what the words should mean, or what the most reasonable interpretation of them is. We should write laws to try to guard against future actors’ attempts to creatively interpret them, both to ensure our goals are achieved and to avoid wasting future resources on unnecessary legal/political battles. The current phrasing seems to leave a lot of room for such interpretation; for example, how many of our desires and priorities count as ‘needs’, how literally we interpret ‘without compromis[e]’ and how certain the harm needs to be to reduce ‘ability’. And even if such rules were precisely defined, and the most competent and well-informed body in the world were opining upon them, there would be profound epistemic uncertainty. The risk of incorrect decisions would be very high.

Another significant part of the Bill subjects public bodies to reporting requirements regarding their preventative spending. The Bill defines preventative spending as:

(a) Prevention: expenditure which has the purpose of creating societal conditions which mitigate risks to the future generations principle;

(b) Present and future spending: expenditure on a recognised issue which currently breaches the future generations principle to prevent the issue deteriorating and so contribute to the alleviation of the problem for current and future generations;

(c) Acute spending: expenditure with the purpose of managing the impact of an issue which breaches the future generations principle, but which is unlikely to prevent the issue deteriorating.

This is an extraordinarily broad definition of ‘preventative spending’. On one reading it could effectively cover any government spending. It is possible to come up with some kind of tenuous pretext by which almost any given action has an effect on the needs of the present and mitigates risks to future generations.

Apart from the unworkable breadth, this neglects Goodhart's law: when a measure becomes a target, it ceases to be a good measure. A department may spend billions on preventative spending without having anything to show for it. Wherever possible we should be measuring outputs, not inputs. 

Another problem is that the Bill makes the judiciary the ultimate arbiter of the duties proposed. The judiciary is profoundly ill-equipped to assess quantitatively the sort of long-term tradeoffs and risks that are implicit in the types of decisions covered by the Bill. Under the 2015 Welsh Future Generations Act, the courts have refused to intervene. Although the new Bill has stronger language to allow court intervention, judges are not economists, statisticians, scientists, philosophers or professional forecasters, and tend to evaluate processes (whether the correct paperwork was filled out) or ‘reasonableness’ (whether it seems at all plausible that the decision was acceptable).

Hurts infrastructure investment

One of the biggest drivers of human progress has been infrastructure investment: we all benefit from the huge investments previous generations made in roads, railways, electricity transmission and so on. But over time it has become more and more expensive to build the infrastructure we will need for the future because of what Holden calls vetocracy. Technological innovation has led to ever-increasing political demand by residents for more land use regulations to protect them against economic externalities such as loss of daylight. Measures such as Environmental Impact Assessments, CEQA, and EU protection of some species (that are not even endangered in the UK) have all led to an increasing vetocracy which appears to be profoundly harmful to overall welfare. 

Unfortunately, far from improving this situation, there is a high risk that the Bill would exacerbate that problem. By imposing extensive and ill-defined duties on public bodies it will very likely make all these problems worse, reducing future welfare directly, in exchange for an uncertain benefit. There is an extensive history in infrastructure investment of such vaguely defined rules being abused to block infrastructure that would be beneficial to future generations.  The Welsh Future Generations Bill led to the canceling of a road project whose benefits were projected to exceed costs by 2.2x, and then a halting of all road investment.

The dominant constraint on infrastructure investment in the UK is not the discount rate paid by private actors, or the discount rate used to assess government projects, but political limitations, especially reluctance of politicians to think beyond the next election, and resistance to new development that may cause negative spillover effects for neighbours. This Bill would not solve these political limitations, and may inadvertently provide additional ammunition for those who are seeking to veto projects.

Red tape

Even in areas where the Bill does not outright prevent long-term investment, the extensive procedural requirements imposed will increase the costs and uncertainty of any such investment and therefore decrease the overall amount of such investment.

Regulatory capture

If the Bill successfully creates a powerful Commission and an influential ‘national conversation’, these will be subject to the same political pressures as any other political institution. Public conversations about welfare tend to be captured and dominated by the highly privileged, so we think they will turn into a focus on matters of concern to the better-off members of high income deciles, rather than on traditional EA concerns like poverty, housing and pollution, all of which disproportionately harm those on lower incomes. Equally it seems likely that this capture will end up disregarding catastrophic risks and animal welfare to the extent that these are not already priorities to the most politically influential or the animals in question are not photogenic. 

The Bill does not seem to have any mechanisms to prevent such capture, or even to prevent the normal drift of institutions. It is likely that EA influence over the Bill is highest during the drafting: as such we should try as much as possible to ensure it produces positive outcomes even if it will later have a very different group of people at the helm. Our concerns about capture by short to medium term middle class preoccupations are stronger because the Bill seeks a Commission with such extensive regulatory powers and corresponding duties on Ministers and public bodies. We think there is a high likelihood that these new institutions will damage net welfare.

Consider for example the ‘national conversation’ on wellbeing that the Bill seeks to create:

The Secretary of State must provide for a public consultation (to be called a “national conversation”) on a series of wellbeing goals for the United Kingdom, which includes persons the Secretary of State deems to be appropriate. 

 The national conversation must engage communities across the population, with particular focus on young people and children from different social backgrounds. 

As far as we can see there is essentially nothing in the Bill to guide this consultation towards EA goals. Given this, it seems likely that people’s other specific and short-term goals and objectives will have considerable influence over the conversation. People do care about future generations, so this concern will probably influence the conversation. But it’s not clear why we should expect this influence to be significantly stronger or more influential than in other pre-existing debates. If our primary goal is to reduce existential risks to humanity, it would be better to design an institution specifically focused on those goals.

Insofar as the proposed institution makes statements or recommendations adverse to the interests of those in power, they will work to subvert it. So it is critical to fix the underlying incentives to prevent such subversion.

Moreover the existence of most welfare problems and many catastrophic risk problems is already well known (and it seems unlikely that substantial new categories of catastrophic risk will emerge purely from a discussion among members of the general public rather than from more in-depth research). There is extensive information about wellbeing and people's choices embedded in market prices, and it would be possible to measure those preferences by income decile to permit more nuanced welfare assessments.

The hard question is not so much about what aspects of their lives people would like to improve, but what efficiency gains and trade-offs we could make through policy to achieve some of those improvements. Single issue polling is notoriously unreliable because it cannot deliver opinions on trade-offs, and it is generally impossible for anything other than a highly focused group doing intensive amounts of work to assess with any degree of accuracy the likely outcomes of a range of potential policy actions in a highly complicated world of unintended consequences. Even then, they frequently get it wrong.

Makes procurement worse

The UK currently has an extremely process-orientated procurement system, requiring a great deal of documentation in order to protect against possible judicial review, even if this documentation does nothing to improve the quality of the actual decision. Smaller firms are disfavoured relative to larger ones with expertise in navigating the system, and innovative approaches are shunned in favour of established techniques that firms are confident that officials know to be legally safe. This was a significant problem during the initial covid crisis, where a rapidly changing environment required rapid decision making under uncertainty. Existing rules made it difficult to procure PPE, resulting in lengthy (and often ultimately pointless) legal challenges, and even delayed funding for an EA organisation, Our World In Data, that was providing vital information for tracking the pandemic. This problem has been recognised by many, and is the subject of a Green Paper on possible reforms. 

Unfortunately, this Bill seems like it could make the situation worse. Insofar as it subjects procurement to more duties and barriers such as impact assessments without any mechanisms to reduce such barriers that we can see, it could actually increase x-risk by making it harder for future governments to act swiftly and proportionately in response to a new danger.

Missed opportunity for bipartisanship

Most of the driving concerns behind Longtermism are at least potentially bipartisan. Both major parties care about issues with long-term impact like climate change, and many politicians are personally concerned with how posterity will view them. Leaders of many parties, including Boris Johnson, signed onto the Future Generations Pledge that was pushed by the Bill’s sponsor’s organisation. So it seems possible to craft a Future Generations Bill that could attract cross-bench support. 

Additionally, there are many advantages to doing so. By incorporating insights and concerns from across the political spectrum, rather than just one group, the Bill can avoid errors that might arise if it results from discussion within a small echo chamber, and ensure the benefits are spread more broadly across society. A bill with broad support is more likely to survive over time, avoiding both explicit repeal and more subtle undermining. This especially holds true for a bill that aims to kickstart a ‘national conversation’; presumably this conversation cannot be truly national if a large portion of the electorate regards it as a mistake.

Given this, it is disappointing that the Bill seems to be relatively partisan; worse, it does this by being unfavourable to the currently ruling party - the party which has been in power for most of the last century. This Billl has some conservative supporters, but has failed to appeal to influential right wing voices, and crucially is opposed by HM Government. Some supporters seem to think most of the chance for this or a similar bill being passed rests on a future Labour government, but this may not happen for many years. And if the Bill is politicized in that way, it may be repealed by a subsequent Conservative government.

Risk of making EA look bad

We know many people in government and policy spheres who are highly sympathetic to many EA goals, even if they have not hitherto had exposure to the EA movement. Many of them have decades of experience and reading research on the efficacy of various policy mechanisms. Endorsing a bill of this nature without sufficient discussion within the EA community to ensure the best possible bill is likely to damage the reputation of the EA community in the eyes of those people.

This can happen partly because of the content of the Bill - by promoting bad and ineffective policies, it could make the EA movement, or at least the policy-focused parts of it, appear to have a poor causal understanding of policy, economics and governance. It could also happen because of the way the Bill has been promoted: to the extent that supporters have exaggerated the size and fervour of support, it risks making us look naive about the practical realities of UK politics.

Risk of an echo chamber

The Bill does not seem to have been broadly discussed in the EA movement. There is a risk it was designed by a small ‘echo chamber’. For example, while work on the Bill has occasionally been mentioned on the forum, to our knowledge there has never been a specific post discussing what it should contain and how it should be written. There doesn’t seem to be a strong argument for secrecy here: given the Bill is public and, if it is to be passed, must be publicly debated in Parliament, it seems reasonable for EAs to discuss it first. Given the range of expertise in the EA movement, including fellow travelers with experience at the highest levels of government, it could be useful to discuss in more depth to work out optimal ways to proceed.

Towards workable theories of change for long-termism

As a theory of change, the proposals in the Bill seem fraught with difficulty to us and to many people with whom we have discussed them who are familiar with the way that government frequently works.

We think there are other, potentially more promising approaches that should also be considered, and a wide range of expertise that could be brought in to help. 

Government could be much better

Making government action more long term in perspective can be seen as a subset of the broader set of priorities of improving government action in general. Government action is often highly suboptimal, in the sense that it could be improved in a way that would make everyone, including future generations, better off.

Real rates of risk free return in the economy are currently negative, incentivising market actors to take extremely long term views. That is of course constrained by individuals’ longevity expectations and potential discounting of other lives relative to their own, but market incentives are nonetheless more long-term than they have been for most of human history. That makes private investments in long term infrastructure where much of the expected revenue is decades away - for example, railways,  renewable/nuclear power generation, or commercial space programs – much more attractive than they would be if real interest rates were higher, because they can be funded at lower costs and because they are more attractive compared to competing uses of that capital that would yield higher short-term benefits but lower long term ones.

Despite this societies are generally failing to act accordingly, even considering a timeframe within the expected lifespan of the average current voter. Large economic externalities are currently not internalized. Polluters still often do not in practice pay. Those who increase catastrophic risks, including AI risk, often do not pay the risk-adjusted costs of their actions. If those externalities were completely internalized by regulation or otherwise, the market would provide an extremely strong incentive for entrepreneurs, workers and other market participants to take a quite long term view. At present, the market has solved half of the collective action problem in addressing long-term perspectives.

But governments have failed to regulate to ensure that such externalities are internalized. Worse than that, we often see governments take or omit to take actions in a manner that substantially fails to reflect that negative real interest rate. 

And that is not a problem unique to long-term perspectives. Governments fail to optimize for welfare in the short term across a vast array of current issues. Even in the countries at the current technological frontier, that is often true to an underappreciated and often depressing degree, including issues like urban land use, road and public transport, education, scientific research, healthcare and finance. All of these can be described as coordination problems.

These problems can be so bad that we are aware of confidential examples of at least one area where, even given a clear set of instructions to implement a policy with extremely broad non-partisan support and potentially enormous welfare benefits, the relevant government department has been so dysfunctional that it is incapable of carrying out those steps - even though doing so would be directly beneficial to the politicians ultimately responsible. Governments often suffer not only from pursuing the wrong policies, but from internal obstacles which prevent them from implementing their intended policies. 

There is extensive research on the pathologies of governmental action; a key insight is that government action is often oriented towards volatile and manipulable whims of the population on average and towards special interests of particular actors on the margin. As a result, it is often less useful analytically to focus solely on the interests of a ‘government’ as a whole. Governments and wider society consist of interest groups and individuals. 

Given that governments often fail to optimize even in the short to medium term, it seems unlikely they are currently capable of optimizing for the long term without substantial reform. We agree there may be cognitive or resource limits that prevent existing departments from doing a good job on long term questions which funding a dedicated department or dedicated analytical help may be able to help with. 

Sector-specific approaches may be needed

Rob Wiblin has commented, while clarifying that this was only his first impression: 

I would guess that a 'longtermist' review of most policy areas isn't going to add all that much. The longtermist take is barely different from the shorttermist one, or we're acting in the area not to maximise wellbeing in the fully long run, so it's the wrong frame to have. So I think I would share Sam's intuition that it'd be better to get the longtermist perspective in the few places it really does make a difference — science and tech, war and peace, some sorts of infrastructure, maybe demographics or something like that. (Oh and climate change & similar.) Then you might have enough good analysts to see what it could really imply. I worry longtermism as a mass idea would turn into something like heritage listing or super conservatism about what already exists which is the reverse of my view. Anyway, inasmuch as there's a few specific technologies that we think are super important to advance at any given moment (e.g. broad spectrum vaccines, ability to align AI with our goals, fake meat, etc) might it be more straightforward just to aim to get them properly funded? [emphasis added]

We agree. We also agree with some supporters of the Bill that maximizing short-term economic growth, as currently measured using metrics such as GDP per capita, is not always optimal for long-term welfare. That is trivially true in matters such as nuclear or other arms proliferation. That is partly a question of definition. If we select a welfare function to value long-term risks at what we consider to be the correct levels, we can optimize for short term growth incorporating that welfare function in a way that appropriately values long-term risks. We think the perceived disagreements between the ‘neoliberal’ and ‘progress studies’ movements on the one hand, and the EA movement on the other, may partly stem from such questions of definition. With more careful definitions of ‘progress’ and ‘growth’, we think those disagreements would be considerably reduced.

Moreover we also agree on the critical importance of accelerating the pace of research into improving societal coordination techniques, to improve short and long term policy and governance and reduce long term risks. However we do not think this Bill represents such an improvement. In fact, we think it overall represents a retrograde step, employing a number of approaches that have damaged welfare in previous attempts.

Conceptual goals that are intellectually appealing to EAs – for example, increasing longtermist views and actions by government – may not map well or at all to a single, concrete and tractable goal, no matter how potentially beneficial that goal may be when expressed abstractly. The problem isn’t that this Bill doesn’t give us everything we might want; it’s that we don’t think it even moves in the right direction at all.

We think that there are institutions that can be created or changed to make government more long term, but there may be different things we need to change for different aspects of long-termism. 

For example, the appropriate institutions for AI risk may be very different to those for pandemic risk. Even the EA movement itself has not reached a consensus view on the optimal way to address AI risk. In contrast, there seem to be a number of high-return ways to mitigate future pandemic risk that governments are failing to take with appropriate speed. We therefore suggest that it is useful, when attempting to define goals, to ensure that those goals are carefully defined and that they map to particular tractable areas and theories of change in the real world, given real world constraints, complexities and feedback loops.

Governments have many pathologies, which vary by country. The average UK Member of Parliament and Cabinet minister has insufficient skills in statistics, economics and other social sciences to optimize for any given welfare function. We think that improving those skills is a tractable and high impact cause area. But solving for the relevant political constraints is extremely difficult and, we think, profoundly neglected.

More broadly, we know of no competent and well-resourced institution that is well incentivized and currently focused on designing better systems of government, in either the United States or the United Kingdom. We hope to start a dialogue that will work towards beginning that process.

Ultimately, we think the base rate is that many efforts to improve policy in the last 50 years have ended in failure or worse. The law of unintended consequences has often applied. Of course, policy improvements are still urgently needed and very good when done carefully. But any approach to improve governance and policy, whether in the short or long term, should start with appropriate epistemic humility. It should only be embarked upon after a careful consideration and discussion of plausible failure modes. We are not aware of a comprehensive analysis having been conducted in respect of the proposed Bill. But we welcome the noble intentions, and we think that there are many other potential approaches with better chances of success.

Potential areas for research

We think that there are many tractable sub-areas where improvements that would improve long-term perspectives within government could be made within ten years. John aims to write further on this topic. Such areas include:

• Reducing x-risk: Introducing more lines of defence, including a chief risk office and officer reporting to Parliament.
  • We think this may be a tractable area of reform. The COVID-19 pandemic is still salient and the situation in Ukraine highlights another area of risk. A useful first step might be to iteratively seek the views of policy experts, policymakers and politicians on successive revisions of that proposal, then start a small organisation devoted to building the largest possible cross-party coalition of influential supporters behind that change. Building a public list of such supporters would help to show social validation and make it easier to recruit more.
• Pollution: ensuring that the long-term costs of pollution are properly reflected in what the polluter must pay.
• Improving the ability of governments to optimize (given the incentives provided by current negative real interest rates) to invest and to enable third parties to invest with a long-term perspective, in subject areas such as x-risk, longevity and infrastructure. That might be done at different levels of abstraction:
  • Change an individual decision (e.g. allow more housing on a single site where it will increase welfare)
  • Change individual policies (e.g. allow more/better housing across the country in locations where it will increase welfare)
  • Change policy frameworks (e.g. improve the planning system through street votes; or allowing direct competition within/by the public sector)
  • Change the means by which policy is set (e.g. rethink the planning system: move to more decentralised processes to allow central government to be more focused)
  • Improve the skills of the policymakers, which can be divided into:
    • Improving the skills of elected politicians (e.g. improve the resources for administrative support staff for each elected politician)
    • Improving the skills of officials (e.g. improve the skills in the planning department)
  • Improve the incentives, tools and powers of the policymakers (make career progression in the civil service depend on systemic improvement in the policy area) 
  • Improve the appointments and replacement process for policymakers (improve how civil servants and/or politicians are recruited, selected and retained)
  • Improve the way of setting that process (have a body setting policy on improving the skills of politicians and improving the hiring/promotion/incentives/firing process for civil servants)
  • Increase contestability of who makes decisions in a way such that competition can drive better optimized decisions.
  • And so on, to higher levels of abstraction
• Investing in meta-research on how to address coordination and collective action problems in policy and governance. The EA movement has minimal knowledge of the state of the art in institutional economics, policy analysis and related fields. That suggests to us that more resources invested in this area might result in considerable useful progress. This appears to us to be a tractable, neglected and high impact area of research. It has potential synergies with the work on sector regulation because that work can provide test beds for different theories of change. Particular aspects of that may include:
  • Achieving a broad consensus behind goal definition, which has its own challenges given different priorities for different species within the EA movement;
  • Optimal policy design to achieve goals, if and when they can be precisely defined. Modern theory in institutional economics illustrates that it is possible to achieve a winning electoral coalition for a specific goal chosen from a surprisingly large set, if that goal can be coupled with arbitrarily chosen orthogonal priorities to help to build that coalition.
  • Optimal movement building to build a winning coalition behind those policies
  • Creating institutions to reverse Olsonian and other reasons for decline and drift in quality of policy and governance.
  • Optimal fundraising to provide resources for movement building.

Pandemics, AI and other global catastrophic risks can all be seen as coordination problems. Almost no-one wishes to suffer such catastrophes if they can avoid it. These coordination problems are generally examples of deadweight losses.

Homo sapiens evolved in small groups and with a short expected lifespan. It is not surprising that we have not improved our societal coordination techniques fast enough to keep pace with the increasingly rapid pace of technological change over the last few millennia. However, we strongly believe that can be changed. It is possible to innovate in societal techniques to improve coordination problems. The resources devoted to that question have hitherto been tiny compared to the enormous potential welfare benefits. But they should be carefully directed to achieve a good return on the resources invested. We would welcome views from others, including on potential areas for research and policy priorities for longtermists.

How EA is the Bill?

In the course of writing this article we came across a range of opinions about how EA-affiliated the Bill was. At one extreme, many non-involved people we spoke to seem to be under the impression that this was an ‘EA Bill’ - counterfactually caused by EAs, and warranting a significant epistemic update on the credibility of the movement (if policy was your expertise). On the other hand, many more involved EAs have downplayed the direct contributions and indicated they shared many of our concerns. It seems possible that EAs drafted, or strongly influenced, the initial version, and since then have had more limited impact but have publicly supported / avoided publicly criticizing the Bill, in a way that creates an unfortunate ambiguity. 

Establishing moral credit is not our main concern here, but hopefully we can help demonstrate to onlookers that the EA movement is not quite so enamored with making more quangos as it might have appeared.

Conclusion

While policy change to reduce the neglect of future generations seems very desirable to us, we do not think this Bill is a good way to go about it. The guiding philosophy behind the Bill seems to be that if we mandate the government to do something we will end up with that thing, rather than a mere symbolic representation of that thing. There is little institutional design to keep the mandated bureaucracy on-target to achieve longtermist goals; in contrast it does seem likely to contribute to the vetocracy that makes effective action so difficult in modern democracies. Additionally, the Bill risks reputational damage for the EA movement, due to the perhaps inaccurate impression some onlookers have that this is an ‘EA bill’.

Acknowledgements

We are most grateful to Sam Dumitriu, AlasdairGives, Jonathon Kitson, Seb Krier, Pedro Serodio, Haydn Belfield, Sam Hilton, Nathaniel Bechhofer, Angus Mercer and Sophie Dannreuther for their kind comments on earlier drafts of this. All errors are our own.

Nathan Young @ 2022-03-10T11:30 (+11)

I made a "vote on comments" app for the Future Generations Bill.

Can you write comments that most people agree with? It only takes a minute.

https://pol.is/5cafmyhtrx

This is the app that Audrey Tang talked about on the 80,000 Hours podcast

Charles He @ 2022-03-11T21:55 (+10)

Wait, what? This is awesome. 

In the comment above, Nathan Young has linked his Polis app.

To explain mechanically what seems to be going on in that page:

1. The website asks a bunch of questions and gets responses from people.
2. The answer to these questions creates a set of preferences for each person (their positive or negative reactions to each question).
3. It takes each person's preferences, and compares this to other people, using some measure of "distance" (e.g. counting fraction of agreement or disagreement).
4. It then uses the distances to arrange people's preferences onto a 2d grid (embeds the preferences onto a low dimensional space)
5. It then groups the preferences on the grid ("clusters"), like, by drawing circles around them.

This sounds complicated if your not a nerd, but it's not, really! See footnotes^[1][2][3]!

 

There's very important details/subtleties/extensions here: 

• This relies on "high quality" questions that "interrogate and cover the space of opinions" "correctly".
• Even with perfect questions, small, subtle changes in the embedding algorithm or clustering will get very results. The envelope for different results could be further, dramatically larger if you can tweak the questions or their presentation.

In short, what you see isn't necessarily "Truth" or even truth-seeking, and the groupings may not be "real".

1. ^{^}

   For the underlying methods, this isn't hard to achieve. To calibrate, I think this is an example of unsupervised learning, and could come out of a CS200 class project. If you can code basic Python or start in "data science", you could do this! Gates are open, come in! 

2. ^{^}

   Here are some notebooks:

   • https://tonio73.github.io/data-science/cnn/CnnVsDense-Part2-Visualization.html
   • https://ipython-books.github.io/88-detecting-hidden-structures-in-a-dataset-with-clustering/
   • https://jakevdp.github.io/PythonDataScienceHandbook/05.11-k-means.html
   • https://towardsdatascience.com/a-gentle-introduction-to-hdbscan-and-density-based-clustering-5fd79329c1e8?gi=9c53e5cef389
3. ^{^}

   The second and third notebook is a good introduction to clustering and uses k-means. But some find k-means a little too basic. Polis seems to be using density methods like HDBSCAN.

   The embedding might be UMAP or TSNE. My guess is that the clustering is probably trivial, but there might be some subtle issues related to the embedding.

Charles He @ 2022-03-11T22:00 (+5)

So why do this? 

Here's the actual results. 

So for one, you can see groups of people and opinions. So you can now see "different camps" and "count opinions". 

This could be valuable and hard to achieve. In much social media, voting doesn't take advantage of granular, individual preferences, so it's hard to unravel deep opinions between groups, say on LW or the EA forum.

 

But counting camps is just the beginning, and probably isn't the most valuable thing from this approach. 

For one, under certain conditions, this granularity allows you to understand deep differences in opinions. This can improve communication and understanding in ways not available right now. 

To see this, look at what actually happened on the page. 

There's two groups. For Group A:

• The broad overview is that group A wants to see broader involvement of longtermism and ideas, in a public process.
• However, Group A does not support the Bill, though I guess they have sentiment for something in this direction.
• There's also oddball questions that group A reacted to. You should be skeptical about the signal to noise of information from these questions, but the implication might be that group A has positive views on the role/size/value of government.

 

For the other group, Group B:

• Group B has different preferences and do not want to involve government, not in a broad way that has the government highly engaged or powerful in longtermism or vice versa. One explanation *might* be that it distrusts government entry or control on the issues (presumably related to dysfunction, misuse, capture, drift, and dilution).

• However, much of Group B also has "different preferences about x-risk". This suggests other interpretations about their views. There's a bit of mischief here, but this isn't that important.

Nathan Young @ 2022-03-14T00:19 (+4)

Here is the full report: https://pol.is/report/r8ef5zucaxxvtzkiym69b

Also Charles, let's have a call some time! https://calendly.com/nathanpmyoung/omni (anyone who reads this far is also welcome)

Charles He @ 2022-03-11T22:03 (+2)

Again, the groups might not be real. It's unclear what the groups represent. This might be a flaw that is endemic to this approach, or it could be fixed in some way.

There's more: 

• Instead of naively embedding preferences and creating the groups, you can "take actions that encourage grouping along an existing structure or ontology". If done correctly, the consequent content might be more useful and even more truthful.
• I think this system like most approaches, requires high quality interpretation. One implication is that this probably puts pressure on or cuts out a niche for "punditry". This can be good or bad.
• All voting in some sense flattens complex issues into a simple, literally one dimensional axis. Because it builds on voting, Polis can't completely escape this issue, but tries to alleviate them by interrogating the space with more questions. Doing this runs into other issues, e.g. curse of dimensionality. I think this is why question design and UX choices are deceptively important.
• Implementation involves other sensitivities/issues (e.g.  flavors of panopticon a la Bentham, Foucault or something). It will be interesting to talk to some informed people about this.

Nathan Young @ 2022-03-13T22:53 (+4)

I think they are really useful and would recommend the forum had some kind of Polis functionality. 

John_Myers @ 2022-03-12T13:11 (+1)

Thank you Nathan - this is extremely interesting.