What does it mean to govern? As I learned from a talk by Jean-Pierre Landau, all public policy has three components:
As I blogged recently, once you think about it, public policy is transparently the same as AI (in machines) and NI (in animals). If you are someone who is familiar with my academic writing on AI, then I can say to you: allocation = attention, maybe even consciousness; distribution = planning & action; stabilisation = avoiding dithering. If you aren't familiar with the nuts and bolts of systems AI, don't worry about it; this post is really about politics.
What makes governing nevertheless very different from (contemporary machine) AI & (individual) NI is the autonomy of the governed body's constituents. This is why action selection in government is called "distribution". Action selection by governments is almost entirely about taking the taxed productivity of those governed and applying it to problems. What Prof. Landau said was that traditionally, a big part of stabilisation was making sure distribution occurred under the radar, basically out of view. But stabilisation through cloaking isn't working well right now, which is Landau's explanation for civil unrest such as the rise of populist movements. He thought the lack of willingness to turn a blind eye was because the proletariat were still angry about the 2008 bank bailouts. But I suggested a very different model in the Q & A: that what's really new is the information age.
Maybe you just can't hide distribution any more. That doesn't mean government is over. It means that we need to be able to communicate honest, complex models to voters.
I have a thesis that may surprise you: We aren't living in a post-fact society. Because of inequality and political polarisation, beliefs are being used to signal in-group status even more than usual. But whether we disseminate true beliefs is a choice, and increasingly disseminating true beliefs is becoming the only stable strategy. Truth is the future.
I think this thesis is really big deal, so I've tried to keep this post relatively short. In fact, if you like, you can stop reading in just two more paragraphs. At the bottom of the post I'm just going to rattle off some of the evidence behind my assertions above. But I want to finish the main part of this post with this:
This thesis extends that of my 2015 book chapter, AI and Pro-Social Behaviour. I say in that chapter that theoretical biology predicts that AI and IT more generally make us operate more as groups than as individuals, just because AI & IT facilitate communication. At the end of the chapter, I express concern that this could lead to autocracy if a loss of privacy makes idiosyncrasy too dangerous in the face of powerful organisations. But this blog post is describing another possible consequence: That it is now in the interest of liberal democracies to identify and abandon the heuristics of governance that are actually based on deception, including self-deception by politicians.
I say "identify", because politicians (and the rest of us) may never have thought of many common-sense strategies for efficiently coordinating collective action as deceptive. Politicians aren't evil for having used heuristics that may boil down to this. We've required governance for millennia – long before we understood complex systems. We had to guess how to do it. We'll probably always need to use heuristics, but now we can use better ones. And now that we have better models and better data, we need to focus even more on effectively communicating the reasons behind our policies, even when they are sometimes based on guesses. As well as focussing on choosing policies that really do maximise benefit to all.
Evidence and other footnotes:
But isn't truth over? No, it just looks like truth is over because so many beliefs are now being used to indicate in-group identity instead of to communicate information about the broader non-social world. It isn't random, it isn't meaningless, it's just annoyingly not grounded in real-world semantics. This is part of polarization, and that's part of the reason that my work on political polarization is dominating a lot of my time right now.
I updated this post in 6 August 2016 to clarify it, and to link it to my more recent explanation of heuristics. I updated it again 28 September to explain how this (fails to) relate to truthiness.
- allocation: picking the problem to address,
- distribution: solving the problem, and
- stabilisation: making sure the solution sticks long enough that resources used for allocation and distribution weren't wasted.
As I blogged recently, once you think about it, public policy is transparently the same as AI (in machines) and NI (in animals). If you are someone who is familiar with my academic writing on AI, then I can say to you: allocation = attention, maybe even consciousness; distribution = planning & action; stabilisation = avoiding dithering. If you aren't familiar with the nuts and bolts of systems AI, don't worry about it; this post is really about politics.What makes governing nevertheless very different from (contemporary machine) AI & (individual) NI is the autonomy of the governed body's constituents. This is why action selection in government is called "distribution". Action selection by governments is almost entirely about taking the taxed productivity of those governed and applying it to problems. What Prof. Landau said was that traditionally, a big part of stabilisation was making sure distribution occurred under the radar, basically out of view. But stabilisation through cloaking isn't working well right now, which is Landau's explanation for civil unrest such as the rise of populist movements. He thought the lack of willingness to turn a blind eye was because the proletariat were still angry about the 2008 bank bailouts. But I suggested a very different model in the Q & A: that what's really new is the information age.
Maybe you just can't hide distribution any more. That doesn't mean government is over. It means that we need to be able to communicate honest, complex models to voters.
I have a thesis that may surprise you: We aren't living in a post-fact society. Because of inequality and political polarisation, beliefs are being used to signal in-group status even more than usual. But whether we disseminate true beliefs is a choice, and increasingly disseminating true beliefs is becoming the only stable strategy. Truth is the future.
I think this thesis is really big deal, so I've tried to keep this post relatively short. In fact, if you like, you can stop reading in just two more paragraphs. At the bottom of the post I'm just going to rattle off some of the evidence behind my assertions above. But I want to finish the main part of this post with this:
This thesis extends that of my 2015 book chapter, AI and Pro-Social Behaviour. I say in that chapter that theoretical biology predicts that AI and IT more generally make us operate more as groups than as individuals, just because AI & IT facilitate communication. At the end of the chapter, I express concern that this could lead to autocracy if a loss of privacy makes idiosyncrasy too dangerous in the face of powerful organisations. But this blog post is describing another possible consequence: That it is now in the interest of liberal democracies to identify and abandon the heuristics of governance that are actually based on deception, including self-deception by politicians.
I say "identify", because politicians (and the rest of us) may never have thought of many common-sense strategies for efficiently coordinating collective action as deceptive. Politicians aren't evil for having used heuristics that may boil down to this. We've required governance for millennia – long before we understood complex systems. We had to guess how to do it. We'll probably always need to use heuristics, but now we can use better ones. And now that we have better models and better data, we need to focus even more on effectively communicating the reasons behind our policies, even when they are sometimes based on guesses. As well as focussing on choosing policies that really do maximise benefit to all.
Evidence and other footnotes:
- Anyone who follows Will Lowe and both of me on twitter can probably guess that the above realisation is as much to do with Brexit as about the Trump melt-down of the US Republican party. What is particularly elegant about Brexit is the way it illustrates how stabilised distribution had been handled deceptively by the UK via the EU. The UK got almost as money back from the EU as it sent them, but that money was spent on poor, powerless regions like Cornwall, Wales, and Staffordshire, and on scientific and academic funding. These are clearly things that the UK should be spending its money on directly, but it was easier politically to ring-fence them against being political pawns by diverting the money through Brussels; in fact, it was probably well worth the 8% administrative overhead the EU charges (note: that is shockingly low for a government bureaucracy).
Sending money to poor regions is one of the things any government has to do, because one of the things that defines a governing region is a common currency, and that currency's policy will be set to advantage the more productive regions. This has to be compensated for by direct capital redistribution. The USA does this, the UK does do it directly as well as through the EU. Ironically, one of the EU's biggest problems is that it went to currency union before political union, so it sometimes has trouble handling local corruption when it tries to do this. Landau pointed out how insane it was that people were complaining about far smaller transfers of capital to Greece than had recently happened to Spain and were currently happening to Poland. But the Polish transfers are somehow out of the spotlight, and the Greek ones are somehow in. This was the example he used to illustrate the importance of stabilisation. - I've also recently (earlier this year) seen Tyler Cowen talk about how we are in the great stagnation. He claims that innovation has kind of stalled after the early 20th century, what with the massive changes of the radio, telephone, indoor toilet, automobile, etc. He is very smart and very convincing. After all, he's not the only one who argues that life and cognition are about damping unpredictability – that's the original Gaia hypothesis, and also correlates with some brilliant work by Jessica Flack on understanding primate coalitions.
Nevertheless, I'm increasingly certain Cowen is wrong. I feel that we are getting ever more empowered. Now that the early 20thC took care of gross bodily functions like moving faster, hygienic defecation, and temperature control, what's changing fastest now is cognition and computation. And cognition and computation are the main defining attributes of our species, our adaptive advantage. Cowen is part right, a lot of our power is expressed in apparent stasis so we can focus on stuff we like to do like raising children and roses, but when once in a while you need to really do something like govern or teach undergraduates, you notice how fast the world is changing.
Living in Princeton has been nostalgic for me since it's rural like my grandparents' home, and also I've been playing softball like I used to in Chicago. So maybe that's why I've been hyper aware today of just how different navigating London is from my own first trips abroad in the 1980s. American girls by me were able to assure themselves they were on the right train going a direction that would lead to a connection without looking at a book or getting out of their seat or talking to anyone. All that information is readily available in clearly readable signs, some of which constantly update, others are just redesigned maps on the wall. I was able to find an amazingly good hotel, and from it the best café within 2 miles (probably), and from that the most amazing short-cut back to my hotel (certainly) because of services like booking.com and Google Maps. I would even in 2006 have had to live in a neighbourhood for weeks, maybe months, to have been able to do that well out of two hours available for breakfast/lunch, and I probably would have missed my train to Bath. Thirty minutes with Internet services on my smartphone was an adequate surrogate for many person-days of physical exploration. That's amazing. What if we can do the same thing with political decisions? - If you are wondering about these parentheticals / caveats above
- "public policy is transparently the same as AI (in machines) and NI (in animals)."
- "What makes governing very different from (contemporary machine) AI & (individual) NI though is the autonomy of the governed body's constituents."
- Benjamin Kuipers has a white paper claiming that corporations and governments are already a type of AI.
- Lots of other animals also exploit collective intelligence / collective action
Comments