🔮 Softbank’s billions; co-living; platform economies; safe AI; Russian meddling; atoms, mosquitoes, urbanisation++ #185

Dept of podcasts

The Exponential View podcast is back. Weekly, long-form conversations with world experts on exploring the new political economy of the information age.

The first episode is a MUST LISTEN discussion with Dr Kai-Fu Lee, on entrepreneurship in China, which nation is winning the AI race, and how to get AI to work for humanity.

Find it on iTunes, Spotify, Overcast or Soundcloud 🎧

Dept of the near future

💲💲💲 Masayoshi Son, Softbank’s boss, announces he plans to raise a new $100bn fund every couple of years, investing $50bn a year.

🏁 China’s quantum future. China’s plans to become a quantum computing superpower may involve tens of billions of dollars of investments over the coming years. Their “national strategy for 'innovation-driven” development' may help them surpass the West." This strategy includes state-directed university research, influencing the private tech giants and a “civil-military fusion”. (This week, the US did step up and gathered tech leaders for a discussion on a national quantum strategy, including a $218m funding pool from the Department of Energy. The EU has a $1bn quantum initiative. Earlier this year, the UK announced its own £20m thrust.)

🚪 Co-living firms, like Common & Ollie, are a booming trend. Millennials live in small rooms, with shared amenities, like kitchens & mini golf while paying service fees WeWork-style. A 98-square ft apartment with free shampoo, room cleaning and shared services can run to $1,800 per month. Fascinating model. (Also, millennials are pushing down the US divorce rate by delaying marriage.)

🔋 We’ll need a new battery to meet the demands of decarbonisation, argues my friend Jason Pontin. “If batteries are to play their part, they must cost less than one-fifth the likely minimum cost of lithium-ion batteries”, new possibilities include sulfur-flow, artificial photosynthesis and molten salt.  (Interestingly, The 129MWh Li-Ion battery installed by Tesla in South Australia appears to have an appealing payback period of only three years. )

👔 JP Morgan Chase has a superb report on the platform economy in the US. Around 4.5% of families generated income via an online platform in the previous year, representing a 165% increase in five years. What is fascinating, is that only one in ten could be described as being wholly employed within that platform economy; most seemed to dip in and out, during the times of financial stress (when platform income can represent more than half of all income). The increase in the supply of drivers has seen earnings from transportation decline by about half in five years. 
One wonders whether they are taking the place of a safety net in the US. And conversely, if platform work is going to grow - which it will - how should we think of the kind of safety net that class of work requires. (See also: EV reader & Uber employee, Guy Levin, teamed up with the Oxford Martin School to evaluate whether British Uber drivers were happy. With all the caveats that this is a company-supported study, it was enlightening that Uber drivers were only working, on average, 30 hours a week, and “valued flexibility over a salary or fixed hours.” Carl Benedikt Frey, one of the academic co-authors, makes this useful observation: “If income levels are the prime determinant of people’s well-being, one would expect Uber drivers to be much less satisfied with their lives relative to the general London population. Yet we find the opposite to be true. Uber drivers report higher levels of life satisfaction and worthwhileness than other London workers.”)

😱 A fascinating paper by Adam Thal: status comparison on social media causes affluent Americans (particularly men) to grow more supportive of conservative economic policies. Another unintended consequence of those products, and the irony that the hyper-liberal Facebook workforce has designed a product which generates conservative political responses! (This paper had not been peer-reviewed when I read it. h/t @ruchowdh.)

Dept of venture capital

Masa Son’s plans for Softbank’s series of Vision Funds is rather staggering, planning to deploy $100bn of capital every couple of years. The entire US venture capital industry is around $85bn, the global industry clocks in at around double that.

My observations: Softbank uses its financial might to be kingmaker. Take this $400m investment in OpenDoor, a fascinating AI-powered real-estate play. Or the $300m put in Wag, a dog-walking marketplace.

Because of network effects, particularly data network effects, many information age companies, such as marketplaces or those reliant on AI, will become winner-take-all. So even if these pre-monopolies are in small sectors (like dog walking), they can dominate those sectors with sufficient capital to buy the liquidity.

Softbank flexes its financial muscle. It's reasonably widely known in the startup industry that their negotiation tactic is pretty close to “take our cash or we’ll give it to your competitor”.

The second observation: if the global economy is at an inflexion point.  That is a transition from an industrial era economy, marked by Fordist-companies, tangible stuff & things pulled out of the ground, to an information age one, defined by market-platforms which co-ordinate various players and use AI & automation, then $50bn a year doesn’t seem so large. The global economy runs to $80 trillion a year. Masa Son is making large bets that this transition is happening (it is) and there are legacies to be built. (On the other side are those who struggle to get it. I was fascinated to read this current analysis in Harvard Business Review on how large company boards struggle with innovation and “keeping up with new technology".)

The immediate impact of Softbank’s overflowing coffers will likely be increasing valuations of venture deals. We've started to feel the impact. 2018 late stage valuations are up 20% since last year, and the number of mega rounds, of $100m or more, is surging across the world.

Now it is true, as Eric Feng points out in this excellent analysis of the VC business that: “return on [venture] investment is roughly the same now as it was 15 years ago despite having about 3x as much money in the system. Venture capital has been able to scale as an industry.”

But with the Softbank funds, we are in uncharted territory, never has then been as large a fund pouring as much money in as young companies. There might not be $50bn of great firms to invest in each year. And does anyone really know how to do it at this scale? I’m curious to see how it will play out.

Dept of artificial intelligence

Lovely summary of a recent talk by the founder of DeepMind, Demis Hassabis. Full of wonderful nuggets, one of my favourites: “Deep learning is an amazing technology and hugely useful in itself but in my opinion, it's definitely not enough to solve AI, by a long shot.”

Specification, robustness and assurance: How DeepMind thinks about AI safety. (Good read; see all the Machine Intelligence Garage practitioner’s guide to AI safety.)

🚒 Deep learning pioneer, Yann LeCun, calls Musk’s alarms to regulate AI ‘nuts’:

Then he [Musk] read Nick Bostrom’s book ["Superintelligence"] and put one or two together and said you know, ‘we’re doomed’. He also likes the idea that he’s going to save humanity, right? So that was kind of a perfect.

AI is a technology which naturally leads to a multipolar, competitive world, argues Michael Horowitz. (He also argues that smaller countries could overtake China and the US, about which I am doubtful.)

Hungry judges are harsher at parole hearings. Right? Many of us will have used the example in our talks on bias in decision making. Well… maybe not, a further study identifies that prisoners representing themselves were often heard at the end of sessions, closer to lunchtime. Controlling for self- vs counsel representation eliminated the previously perceived bias. New anecdote needed!

🌊 Google's new service uses machine learning, flood simulations and rainfall records to predict floods and issue warnings to some of the most at-risk populations in India.

Using text mining and NLP of large corpora of scientific literature to “generate valuable molecular hypotheses and has the potential to accelerate scientific discovery.”

Dept of Facebook

Remorseful billionaire, Brian Acton, sold WhatsApp to Facebook for $21bn, left the firm (and $850m on the table) and urged us to  #deletefacebook. Here is why.  (See also, Kara Swisher’s profile of Instagram’s Kevin Systrom who is also leaving the firm.)

Facebook had a dire week, exec departures & tell-alls aside.

First, we discovered the firm was allowing advertisers to target users using phone numbers provided to them for emergency account recovery.

Second, the company confirmed on Friday that nearly 50m accounts have been compromised, as hackers exploited bugs in "View as" feature to access security tokens. (Some scant details of the hack are here.)

Some users even claimed the Facebook was blocking stories relating to the hack in their newsfeed.

Azeem’s note: And an apology about the amount of Facebook in Exponential View. I almost find it tedious, but it is hard to understate the impact the firm has on the economy, the media, civic discourse and our limbic systems. Its scale makes it crucial to understand the whys and wherefores and look a little deeper than normal. What happens in Facebook matters.

Conversely, what happens at the top of Tesla doesn’t matter that much (with the major car companies having, thankfully,  generally decided electric is the future), so Elon Musk’s recent antics are only of prurient interest, in my view.

Short morsels to appear smart at dinner parties

Absolutely stunning long read: how Russia helped swing the election for Trump. (Also, Russian online trolls pushed the recent “Boycott Nike” campaign.)

Distributed systems are complex, but with the rise of token networks, they will form a key part of our future economic infrastructure. Here is a brilliant introduction to them.

⭐ Amazon is launching a series of stores called 4 Star which only feature products rated four stars or higher.

📰 Apple is making more moves into bundling news. It brings a huge audience, but little revenue. Webzine Slate makes more money from articles that get 50,000 page views on their own site than they do from the 6m they get via Apple News. (I'm not sure how to feel about this. Distribution is extremely valuable, so the assessment of which of Slate and Apple brings more to the table, in this case, is not clear-cut.)

Scott Galloway: The end of Snap & Tesla.

China’s payment infrastructure is years ahead of the West. The country may be entirely cashless by 2020.

📺 Lancet study links restricting screen time for kids with higher cognitive performance.

Scientists figure out how to store information in a single atom.

Scientists used a gene drive to turn female mosquitoes sterile, raising hopes for a success in battling malaria.

Ⓜ️ How urban is the world? 55%? Or 85% Here’s the debate (with nice graphs).

Endnote

As many of you know, it was my birthday this week. Comes around every year and I love it! Thanks for the many happy wishes.

I've enjoyed sharing my birthday with EV readers for the past three years. (Did you know that in 184 weeks of pulling together Exponential View, I reckon I've read between 25m and 60m words?)

If you did want to give me a present for my birthday, please do one or more of the following four things:

  1. Tweet this recommendation for the newsletter.

  2. Retweet this tweet.

  3. Recommend the podcast by retweeting this.

  4. Share this on LinkedIn.

And with that, have a fantastic week,

Cheers,

😎 Azeem

P.S. Scroll down for newest updates from EV readers. Some of you have been doing exceptional things. Please email Marija if you have something exciting to share.