Matt's Thoughts In Between - Issue #84

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

The indispensable Jeff Ding, whose work I’ve written about several times before, has a new podcast - a spinout from his ChinAI newsletter. I highly recommend the first episode, in which Ding interviews Remco Zwetsloot of Georgetown’s Centre for Security and Emerging Technology, the author of this new report on strengthening the US AI workforce. 

The conversation surfaces a number of striking facts, particularly on the US’s reliance on international AI talent. According to Zwetsloot, 50% of the US AI workforce is foreign born, rising to 70% of US AI graduate students. This seems particularly important as AI increasingly becomes a national security issue (a point Zwetsloot makes when asked to explain why AI is different to previous technologies), but US immigration gets tighter. There’s also discussion of the flight of AI researchers from academia to the private sector: the report suggests that 20% of AI academics have left for industry!

I want to highlight and celebrate the unusually rigorous format of this podcast. As Ding points out, too many podcasts are the interviewer earnestly agreeing with their subject - whereas Ding (good naturedly) challenges Zwetsloot throughout (There’s even a “favourite footnote" segment…) One of his points of rebuttal draws on Michael Teitelbaum’s argument that warnings of “talent shortages” in STEM are perennial but unfounded, which is worth reflection. I’m looking forward to the next episode.  

One of the animating questions of my life is “what makes people choose to start (technology) companies?“. This matters partly because I believe that there are lots of incorrect answers that drive millions (or more) of public spending (See Josh Lerner’s Boulevard of Broken Dreams for the classic treatment on bad public policy and entrepreneurship).

So I was fascinated to come across this paper (via Tyler Cowen), which suggests that most of the traditional explanations for what drives cross-country variation in entrepreneurship are wrong. The author, Valentina Assenova, looks at a large dataset of entrepreneurial activity and finds that the most powerful driver is not the availability of venture capital or R&D spending, but social norms. In short, more people start businesses in countries where entrepreneurship has a higher social status.

This is plausible - and I’d like it to be true, as I’ve spent most of my career trying to raise the social status of founding tech startups - though there are some reasons to be skeptical. The paper doesn’t distinguish between high-growth, Silicon Valley-style startups and small businesses, which means it finds, for example, that sub-Saharan Africa has some of the highest rates of entrepreneurship in the world. That’s quite possibly true, but probably not what our politicians are trying to replicate.

I’ve been reflecting recently on political developments on both sides of the Atlantic and the realisation that few politicians - or, indeed, members of the public - have any deep attachment to due process and norms, so long as their side “wins". It’s hard to imagine (m)any Leave activists endorsing proroguing Parliament if used by Jeremy Corbyn to achieve the opposite effect. And we know that some Senators’ views on impeachment are remarkably correlated with which party holds the presidency. Lest you think this is limited to one side, how many Remainers would have supported the Speaker’s creative interpretation of convention if circumstances were reversed?

Why are people willing to put at risk norms that have contributed to centuries of stability? I suspect the answer is found in this pseudonymous (and frankly offensive) column published during the 2016 US Presidential election (it turned out to be by Michael Anton, who later joined the Trump White House), which called 2016 the “Flight 93 Election” (“charge the cockpit or you die”) and argued that a Hillary Clinton presidency was an existential threat to America.

This cataclysmic tone may seem more common on the right, but we see similar language on the left, particularly in the environmental movement. I’m not drawing moral equivalence, but the problem is, what can’t be justified if the alternative is the end of civilisation? Technocracy gets a bad press, but - if you care about the rule of law - it seems vastly preferable to competing apocalyptic visions. 

I mentioned last week that I’d love to grow TiB’s readership ~1% per week from now until New Year. I’m delighted to say that we hit this almost exactly last week, so… if you enjoy this, please forward it to a friend or share this on social media (LinkedIn seems to work particularly well for some reason).

As ever, feel free to reply if you have comments or say hi on Twitter.

Until next week,

Matt