Thoughts in Between
TiB 169: AI authors; better ways to fund startups; China's "Digital Silk Road"; 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.
Should authors fear AI?
What it might it be like to be an artist in a world where machines can create art as well as humans can? Novelist and neuroscientist Erik Hoel (yes, the one I wrote about last week) has a new piece on this topic. To show how close we are already, he has AI writer GPT-3 (see TiB 119 and TiB 124) “rewrite” a passage from his novel; the results are impressive (Incidentally Hoel’s jumping off point for the essay is Harold Bloom’s The Anxiety of Influence, one of the most interesting books on the creative process)
One way of thinking about how the world will adapt is to look at a field where machines long ago surpassed humans: chess. There’s both a pessimistic and an optimistic lesson here. The pessimistic take is that machines don’t stop getting better once they pass human performance. There was a brief window in which there was excitement that the combination of human and machine (so-called “centaurs”) was better at chess than either. This didn’t last long, as we discussed in TiB 136. Today, humans add nothing to the partnership, thanks to DeepMind’s AlphaZero.
The optimistic take is that this hasn’t destroyed human chess - far from it: its popularity is at an all-time high. We value some things just because a human is doing them (and we like to document the extremely rare edge cases in which a human can outwit the machine). I suspect a world of abundant AI authorship actually raises the status of the best human authors, and creates new markets in curation, personalisation and artisanal “machine-free” stories. And, perhaps, demand for a new form of cultural history: tracing the human text in the training set that “taught” the machine to speak…
Can we design better ways to fund startups?
Last year in TiB 129 I wrote about designing an “Optimal Kickstarter” for public goods, based on this post by Matt Clancy. I pointed to Alex Tabarrok’s idea of Dominant Assurance Contracts (DACs) as one of intellectual antecedents of this concept. The core idea of a DAC is that pledgers to projects that don’t meet their funding target not only don’t pay, but actually get paid a “refund bonus”. The goal is to avoid the failure mode in which good* projects fail to get funded by shifting the Nash equilibrium from “do not fund” to “fund”. More in this talk by Tabarrok.
Maybe that sounds bizarre, but Tabarrok has an update: DACs work in practice. A new experiment (paper here; write up on Marginal Revolution here) suggests that they increase the funding rate of good projects by over 50%. That makes DACs an interesting mechanism for funding public goods, but - credit to my friend Marc - it’s interesting to ponder whether it might be a good way to fund startups too.
One problem in less mature startup ecosystems is that you have a lot of investors who don’t want to commit unless there’s a lead investor (this is less of a problem in Silicon Valley, where “high resolution fundraising” is more common). This means that it takes much longer to raise capital and, again, a lot of good companies don’t get funded. In many non-US ecosystems, government is the largest investor in VC funds. I wonder what would happen if some of this capital was redirected to refund bonuses. Not only might it be cheaper, but I suspect it might end up with a more diverse set of founders getting funded. It would, at minimum, be an interesting experiment.
*where “good” means that the total social value (i.e. the sum of the benefit each individual receives) is greater than the cost
Podcast: Meia Nouwens on Taiwan, China and defence
I've written quite frequently about China and Taiwan (see, e.g., TiB 136, 158, 161), but I'm at best an interested amateur on these subjects, so I was excited to interview a true expert in this week's TiB podcast - Meia Nouwens of the International Institute of Strategic Studies. Meia studies Chinese defence, security and technology and in this conversation she serves up a feast of knowledge on some of the world's most important geopolitical topics.
We focus on three big themes, any one of which could have been a full conversation: China's "Digital Silk Road" - i.e. Chinese public and private investments in global digital infrastructure; the political of semiconductors and the future of Taiwan; and the use of artificial intelligence in defence.
Throughout Meia touches on a number of favourite TiB themes, including:
- How the CCP influences Chiense technology companies
- Why ASML, Europe's most valuable but unglamorous technology company, has become so geopolitically important
- Tacit knowledge in semiconductor manufacturing
- Whether the Chinese military is ready to mount an invasion of Taiwan
- ... and much more
I learned a great deal from Meia. If you're interested in how China's changing role in the world, I think you'll get a lot out of our conversation.
- The Digital Silk Road, mapped. A good companion to this week’s podcast - an interactive map of the global investments of China’s tech companies.
- Amo, amas, aMATLAB? Fascinating study (albeit with a small sample size) suggests that linguistic ability is the best predictor of speed of learning to code.
- The lead paint of our era. Horrifying paper on the impact of particulate matter polution on cognitive function. And one on its relation to psychopathy.
- Whither Downton Abbey? Interesting paper on when and why wealthy Americans became the brides of choice for British aristocrats.
- Basement Jacks? Where in Europe are adults most likely to live with their parents?
The bit at the end
Thanks for reading Thoughts in Between. Please forward this to someone who might like it - or share the URL on Twitter or LinkedIn.
You can reach me by hitting reply; I'm always happy to hear from readers.
Until next week,
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.