Comment Permalink

This is great and under-emphasized. I think it was @weeatquince who told me that the primary determinant of what gets implemented by governments is what has successfully been tried before, and while I haven't seen much empirical data on this it strikes me as plausible.

One counter-point comes from Michael Rose's book Zukünftige Generationen in der heutigen Demokratie, which finds that low institutional path-dependence (approximated by the rate of recent constitutional changes) had no effect on the institutionalization of powerful proxies for future generations in a (pretty small) fuzzy-set analysis.

On the other hand, former Welsh minister Jane Davidson says that Wales was able to implement their Well-being of Future Generations Act due to the innovativeness of the Welsh government in her new book #FutureGen.

In addition to seeing more EAs get into innovative governments to run policy experiments, it would be great to see further research on policy diffusion and on the importance and proper characterization of governmental innovativeness in the sense you outline here.

ac5y1

Thanks for this Tyler!

The references are great, I wasn't aware of them. Re the first, how exactly do you think low institutional path-dependence and institutional innovativeness interact? They seem like related but distinct concepts to me.

I agree that it would be great to see more research on those questions, though I wonder if a thorough review of the policy diffusion literature might be sufficient. I definitely would like a clearer characterization of governmental innovativeness; I felt kind of hand-wavey in this post.

See in context

Longtermist reasons to work for innovative governments

by ac

Oct 13 20202 min read 8

74

PolicyAI safetyImproving institutional decision-makingAI governanceLongtermist institutional reform

Frontpage

Epistemic status: speculative

Longtermists have good reasons to be interested in institutional innovation. Broad longtermist institutions could focus on improving the political representation of future generations; more targeted ones could focus on particular cause areas such as AI governance.

The community’s approach to institutional change so far seems to be two-pronged:

Research new institutions. Examples of such work include Will MacAskill on age-weighted voting and Gillian Hadfield and Jack Clark on regulatory markets for AI safety.
Get into influential positions in the most influential governments, so that we are better placed to take important decisions, including regarding new institutions. For instance, to my knowledge, CSET was established partly to help build the political capital of thoughtful longtermists in D.C.

Here’s another approach which might be promising:

Get into influential positions in the most innovative governments (i.e., those most willing and able to create and change institutions), so that we are better placed to test institutional innovations.

Many of the innovations which are interesting by longtermist lights can only be implemented by governments (e.g. most of the ideas in this post, regulatory regimes); and, presumably, governments vary in their willingness to test new institutions. There’s also no clear reason why the most influential governments should be the most innovative. Field-trials would help establish which ideas work well empirically, and running them in innovative governments would allow for more ideas, and stranger ones, to be tested. According to Robin Hanson, “the key resource needed for institution adaptation efforts is actually real organizations willing to risk disruption and distraction to work on adapting promising institution ideas.”

Successful implementations could also propagate new institutions to more influential governments. A cursory look at the literature on the diffusion of policies and institutions suggests that institutions set up in one country often quickly spread to others. In the context of environmental institutions, for instance, “diffusion mechanisms contributed to a significant extent to the international spread of environmental ministries and agencies, particularly in the 1970s.” One reason for this might be the reduced risk for those following the innovator: it’s already been shown that the institution can work well.

If the above is correct, perhaps some fraction of longtermists aiming for a government career should look towards the most innovative governments. Some questions that would help clarify the value and feasibility of this approach include:

Which governments are the most innovative?

Could EA’s work for these governments?

Is the diffusion of institutional innovations a robust phenomenon?
Which countries have historically had the most success diffusing their innovations?

Thanks to Maxime Riche, Jeremy Perret, Adam Shimi, and Laura Green for feedback.

74 Reactions

Mentioned in

84Why scale is overrated: The case for increasing EA policy efforts in smaller countries

58Foresight for Governments: Singapore’s Long-termist Policy Tools & Lessons

36EA Updates for November

Comments8

Sorted by

New & upvoted

Click to highlight new comments since: Today at 11:05 PM

jared_m5y21

Thank you for this post. I agree with the attractiveness of the opportunity! Two top-of-mind potential avenues for someone to explore this would be:

1) Nation state-level political innovation: one might become an e-resident (and an active one) of Estonia, and study upfront how its public sector innovations are or aren't working. The country is small and public leaders are accessible. So even if it's difficult for a foreigner to work for the Estonian government, I suspect it isn't that difficult to become a moderately influential voice on their model as a digital nomad working in Tallinn and reporting on Estonia's innovation under their new residency program. Other states that embrace well-educated foreigners in high-leverage public sector advisory roles under the right conditions include the Gulf States: Qatar, UAE, etc.

2) Local political innovation: City Halls in wealthy cities are great sources of innovation. For example, in NYC, Mayor Bloomberg's former aides went on to advise Mexico City and many other cities on innovations piloted in New York. Hopefully de Blasio's aides will distill and share the advantages and drawbacks to their Meatless Monday pilot in New York's public schools. Toronto City Hall aides would have a front-row seat to which innovations were most promising and which were most problematic in Alphabet's shuttered plan to remake part of the city. Anne Hidalgo's climate and COVID policies will certainly lead to lessons for other big-city mayors, for a European example.

Thanks for framing this opportunity, and I look forward to hearing which approaches people pursue to run opportunities like this down!

tylermjohn5y9

ac5y1

Thanks for this Tyler!

weeatquince5y8

Hi Alexis, thank you for the post. I roughly agree with the case made here.

I thought I would share some of my thoughts on the "diffusion of institutional innovations":

* I worked in government for a while. When there is incentive to make genuine policy improvements and a motivation to do so this matters. One of the key things that would be asked of a major new policy would be, what do other countries do? (Of course a lot of policy making is not political so the motivation to actually make good policy may be lacking).

* Global shocks also force governments to learn. There stuff done in the UK after Fukishima to make sure our nuclear is safe. I expect after the Beirut explosions countries are learning about fertiliser storage.

* On the other hand I have also worked outside government trying to get new policies adopted, such as polices other countries already do, and it is hard, so this does not happen easily.

* I would tentatively speculate that it is easier for innovations to diffuse when the evidence for the usefulness of the policy is concrete. This might be a factor against some of the longtermist institution reforms that Tyler and I have written about. For example “policing style x helped cut crime significantly”is more likely to diffuse than “longtermism policy y looks like it might lead to a better future in 100 years”. That said I could imagine diffusing happing also where there are large public movement and very minimal costs, for example tokenistic polices like “declare a climate emergency”. This could work in favour of longtermist ideas as making a policy now to have an effect in many years time, if the cost now is low enough, might match this pattern.

I also think that senior government positions even in smaller countries can have a longterm impact on the world in other ways:

* Technological innovation. A new technological development in one country can spread globally. * Politics. Countries can have a big impact on each other. A simple example, the EU is made up of many member states who influence each other.

* Spending. Especially for rich countries like in Scandinavia they can impact others with spending, eg climate financing.

* Preparation for disasters. Firstly building global resilience -- Eg Norway has the seed bank -- innovations like that don’t need to spread to make the world more resilient to shocks, they just need to exist. Secondly countries copy each other a lot is in disaster response -- Eg look at how uniform the response to COVID has been -- having good disaster plans can help everyone else when a disaster actually hits.

I think it matters not forget the direct impact on citizens of that country. Even a small country will have $10-$100m annual budgets. Having a small effect on that can have a truly large scale positive direct impact

jia5y6

Thanks for writing this post! It's cool to see people thinking about less direct, but potentially more neglected and tractable paths to affecting influential governments.

Do you have thoughts on the difference between intentional and unintentional diffusion?

Intentional. Country A actively tries to export its policies. Perhaps it is trying to establish itself as a leader on a specific issue, or shape global standards in a way that benefits local companies. The literature on "niche diplomacy" might be somewhat relevant here.
Unintentional. A is "doing its own thing", and isn't going out of its way to export its ideas. Possible implications:
- The path to impact from A to influential country B is fuzzier.
- It might be harder for someone working in A to tell who is learning from their policies. So we'd need people working in B to tell people in A that they're having an impact.

Also, getting people to pursue this path might be challenging because of things like status effects and people preferring to live in EA hubs.

ac5y0

Hi Jia, I don't know much about policy diffusion, unfortunately. From the brief reading I have done, both seem to occur. Learning from other governments appears to be important; according to some researchers, exchange via membership in international organisations is a central causal factor. "Intentional" governments could make active engagement in these a higher priority.

MichaelA🔸5y3

Thanks for this post - this seems like an important and under-discussed idea!

It's not obvious to me that status effects will be that important. Influential positions in any government are pretty well-regarded, and it could be easier to climb the policy ladder in innovative governments (e.g. if they are smaller).

I think that that should indeed help increase the status that people can expect from pursuing this path, and thereby reduce the extent to which worries about status will stop people going into this path. But I'd guess that what Jia had in mind might've been primarily about status within EA/among EAs.

In general, EAs might be biased towards roles in explicitly EA organisations partly because that gives them more status among other EAs, because it's easier for other EAs to understand whether and why those roles are important. If you're doing a really important role in some important government/foundation/company, other EAs may just not know whether or why that role or org is important. (See e.g. here and here for some similar arguments.)

And this issue may be more intense the less well known or influential-seeming the government/foundation/company is. So this might make EAs biased towards working at EA orgs rather than the US government, and being towards the latter rather than a smaller and more innovation government.

This definitely isn't a reason why we shouldn't highlight the value of this path, though - if anything, it would push in favour of saying more loudly and more often how useful this path would be! (Conditional on this path indeed being a good idea, which I'd currently guess it is.) That way, we can increase how well EAs understand the value of this path, and thus allow people pursuing this path to get a more appropriate level of status (relative to e.g. people working at EA orgs).

(I'd probably prefer it if people in general cared less about status. But I think a lot of people do care about status, myself included, and it seems hard to make oneself care substantially less. So it might sometimes be best to just try to make status more aligned with impact.)

tamgent5y2

Governments may also compete on being innovative, adding to your diffusion point.