Matt's Thoughts In Between - Issue #59

Forwarded this email? Subscribe here. Enjoy this email? Forward it to a friend.

AI risk without Terminator

How much should we worry about AI catastrophes? Nick Bostrom’s (excellent) book Superintelligence (see previous coverage) sparked a debate that led to many prominent people arguing that AI safety should be a top priority for people who worry about existential risks to humanity. But for every proponent, there’s a counter-claim that superintelligence fears are just apocalyptic scaremongering.

It’s therefore well worth reading Paul Christiano’s new-ish essay “More Realistic Tales of Doom”, in which he outlines two fairly mundane scenarios that could see huge dilution of the things (most of us) value today (Some smart commentary here)

The first he describes as “going out with a whimper”. The idea is that machine learning (ML) will make it easier and easier to achieve “easy to measure” goals at the expense of our true “hard to measure” goals and “we ultimately lose any ability to influence society’s trajectory”. The second - “going out with a bang” is darker: gradual systemic reliance on ML makes society more and more fragile, until there is a sudden catastrophe. It’s not cheery reading, but it’s sober, measured and one of the most thought provoking things I’ve read recently.

What if history isn't over, but growth is?

As I've mentioned before, I'm currently hosting a fortnightly reading group in London for Peter Thiel's Stanford class Sovereignty and the Limits of Technology and Globalisation (See previous Thiel coverage). It's been a great experience so far, but last week's session, "Optimism to Stagnation" seemed especially relevant for TiB.

You can see the full reading list here, but if you're pressed for time, I recommend this 1997 essay by John Horgan on "Why I think Science is Ending" and Robert Gordon's 2012 paper, "Is US Economic Growth Over?". These are pessimistic - and, for me, ultimately unpersuasive - arguments that we may have come to the end of a historically brief golden age in which it was reasonable to expect the future to be better than the past.

Each fortnight at the reading group we like to ask the meta-question of why Thiel assigned these readings in particular for this course (Related: he just gave a big interview to a German newspaper). My view is that Trump and Brexit can be read as cautionary tales of the vulnerability of our politics to just a decade of slow economic growth. If that's true, what if Gordon is right and growth is over? Perhaps the apparently stable political institutions of western liberal democracy are just an idiosyncratic artefact of anomalous continuous economic growth. That's not a cheery thought either.

In praise of lobbyists (sort of)

Lobbyists are often blamed for the public’s diminishing trust in the political process. Not unrelatedly, “there’s too much money in (US) politics” is an almost completely uncontroversial opinion. That’s understandable: there is plenty of impressive anti-lobbying evidence, like this extraordinary thread in which data scientist and journalist Rob O’Dell demonstrates that US corporations literally wrote thousands of bills that became law verbatim.

And yet… the more you look at it, the harder it seems to explain why there is so little money in US politics. If lobbyists really have the power ascribed to them - to move tens of billions of dollars in public policy - the return on investment that corporates get from lobbying is astronomical - 100x or more. So why don’t corporations invest more? Shouldn’t lobbying spend rise until the rate of return drops to that of a corporation’s next best investment?

One answer is that lobbying is misunderstood. It’s better explained as a legislative subsidy - a way of “assist[ing] natural allies in achieving their own, coincident objectives” - than vote buying. I’m not sure if I wholly believe that, but the more nefarious you think lobbyists are, the more it should puzzle you why there aren’t more of them…

Quick Links

Putin vs the world. The fascinating story of how even GPS bends for the Russian leader.
How to be a CEO. Superb thread on the competencies of business leaders from a startup perspective.
The other side of the table. Excellent mass Twitter Q&A on what Leave arguments Remain voters find most persuasive (and vice versa)
Trade Union suppression, Vine edition. What happened when social media creators tried collective bargaining? (Spoiler: lose/lose)
When decentralisation kills. Great chart on the impact of local devolution on carbon emissions in China.

Your feedback

Thanks for reading. If you enjoy Thoughts in Between, I'd love it if you'd forward this to a friend who might enjoy it too. And feel free to hit Reply if you have any thoughts or feedback.

Until next week,

Matt