Thoughts in Between

by Matt Clifford

TiB 201: Democracy-saving tech; The Power Law; a quant history of women; and more...

Welcome new readers! Thoughts in Between is a newsletter (mainly) about how technology is changing politics, culture and society and how we might do it better.

It goes out every Tuesday to thousands of entrepreneurs, investors, policy makers, politicians and others. It’s free.

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

Technology to save democracy?

It's fashionable to believe that the internet is destroying liberal democracy. The best antidote is the work of Audrey Tang, a former programming prodigy and now Taiwan's Digital Minister. Tang is interested in building digital tools to help resolve difficult political questions via deliberative democracy. There's a very gentle video introduction to her work here and her ministry's web page is here. Rob Wiblin at 80,000 Hours has a new podcast episode with Tang this week. She's been on several podcasts before, but I think this is probably the best, largely because Rob isn't afraid to let the conversation run for a couple of hours, so there's plenty of time to dive into detail.

Tang's most famous project is Polis, which she describes as "pro-social social media". The basic idea is to allow large groups of stakeholders to express views on a controversial question and to evaluate the views expressed by others. An algorithm clusters these views by similarity and participants are incentivised to contribute statements that attract broad consensus across clusters. Over time, this creates a set of norms that define a solution space for resolving the question. In Tang's words:

*As the weeks go by, you literally see the clusters inching closer to one another, because they found some broad principles ... that actually convinced them that “If you implement those values, then I can actually live with it” ... Now it’s our job as policymakers to craft a policy that can actually satisfy all the 10 criteria that’s already broadly agreed upon by people who initially felt very differently

Polis has now been used to help solve policy questions in the real world many times, from the debate in Taiwan on whether UberX should be allowed to operate there, to work by the think tank Demos on the UK's post-pandemic recovery, as well as by many more organisations. It's not without its challenges - see this MIT Tech Review piece on why it hasn't become more influential in Taiwan - but it's refreshing to see a positive vision for what technology can bring to the actual practice of political and social decision making. If you're not familiar with it, it's certainly worth your time.

A quantitative history of women

Two favourite occasional TiB topics are macro-history (see TiB 83 and 105 ) and the way that machine learning is changing the humanities (see TiB 81 and 122). Last week I stumbled upon perhaps the best paper I've seen in this genre: "Herstory: The rise of self-made women" by Arash Nekoei and Fabian Sinn (There's also a short and accessible write up in VOX EU, which includes the best graphs). The authors call it "a quantitative history of women over the last 5,000 years". They used machine reading techniques to digest Wikipedia and traditional encyclopedia entries across nearly 300 languages and to create an index of 7 million prominent historical figures since 3000 BC, coded by geography, gender, family background and more.

The overall story is perhaps predictable - women make up just 10 percent of the entries in the dataset - but what's more surprising is that this number has remained relatively stable over time. In fact, women's share of prominent people born in the 20th century is lower than the share for people born 5000 years ago! What has changed, of course, is the proportion of women who were "self-made" (which here means simply whether they were born to or married another individual in the dataset): this has risen from sub-5 percent for most of history to close to 20 percent today.

The dataset allows the Nekoei and Sinn to ask where and how the "rise of the self-made woman" started. The authors hypothesise that for most of history the return to high levels of skill was low for women, owing to social norms. But literacy and the market for writing that developed from around 1600 prompted social and institutional innovation that allowed small but increasing numbers of women to build careers in their own right. For example, they find evidence that countries that saw increases in the number of female writers and poets in the 17th century were more likely to have higher proportions of "self-made" women in the 20th century. That's far from the whole story, of course, but it's a great example of new kinds of history that would not be possible to write without recent advances in data science.

What is venture capital good for?

This week's TiB podcast episode is a conversation with journalist and author Sebastian Mallaby, whose new book The Power Law is an exploration of the history and value of venture capital (VC). I like to think I know a lot about VC, given my day job, but I learned a huge amount from the book and highly recommend it. Sebastian begins his story with some of the post-war early experiments in proto-VC and takes it right up to the present day with the rise of Softbank and Tiger Global and the globalisation of startup culture.

One of the main things I took from the book, and one of the topics we discuss in the podcast, is just how long it took the VC industry to iterate towards the legal structures and industry norms that make it succeed. The early days of VC are full of stories - like ARD or even the funding of Fairchild Semiconductor - where the technology and commercialisation succeeded but the structures used stymied the success of both founders and funders. (Should this make us more optimistic about crypto at the margin? Much of what looks ridiculous today is just real-time experimentation with new structures. The Power Law tells us that even the smartest practictioners can take years to alight on the right answers... but once they do it seems obvious)

We also discuss:


Quick links

  1. By gradient descent? Machine learning salaries are falling, apparently.
  2. "Surprising things believed by extraordinary scientists". An interesting list by Michael Nielsen.
  3. Real-time inequality. Nicely executed site that allows you to visualise US income and wealth inequality by multiple factors over time.
  4. Is this… shape rotation? What Wordle teaches us about information theory. Very good video.
  5. No category for content marketing? What do VCs do with their time? Interesting study and chart.

What do you think?

Thanks for reading Thoughts in Between. If you like it, please do share it with a friend by forwarding or share the link on social media.

Thanks too to everyone who replied last week with links to books and videos comparing the relative prowess of various animals. I'm sorry I couldn't reply to you all!

Until next week,

Matt Clifford

PS: Lots of newsletters get stuck in Gmail’s Promotions tab. If you find it in there, please help train the algorithm by dragging it to Primary. It makes a big difference.