Azeem Azhar’s Weekly Wondermissive: Future, Tech & Society
We have a number of exciting State of the Exponential briefings coming up on climate modeling, technology regulation, cryptocurrencies, and AI.
Next Thursday at 4pm London time, we’ll discuss climate modelling with Prof. Myles Allen. He will take us through what climate models are, whether they matter, and what models we need to deal with climate breakdown more effectively. If you’d like to join, register here. If you’re not a Premium member yet, sign up here to access all members-only briefing calls.
❤️ If you enjoy Exponential View, please take a moment to hit the like button on top of this email or comment below; then share it on Twitter or LinkedIn.
Dept of the near future
🤖 Bill De Blasio, mayor of New York and a presidential candidate, argues in favour of a new federal agency to approve any plans firms may have for automation:
conditioned on protecting workers; if their jobs are eliminated through automation, the company would be required to offer their workers new jobs with equal pay, or a severance package in line with their tenure at the company.
The chaser is ‘robot tax’ on large companies:
that eliminate jobs through increased automation and fail to provide adequate replacement jobs [...] go right into a new generation of labor-intensive, high-employment infrastructure projects and new jobs in areas such as health care and green energy.
De Blasio’s plan also specifically addresses the need to reskill workers.
But to focus only on those workers who lose their jobs through automation is too narrow. It may lead to perverse behaviours. Job losses in the retail sector, for example, are not coming as a direct result of automation by retailers. This paradigm shift in the economy is broader than robots and their clunky precursors. Employment in the US is at historically high levels, with 158m people in work. Demographics and tougher views on immigration will mean labour will remain scarce, even resulting in older workers staying in the workforce. There are 7.2m vacancies in the US and 6m people out of work, even as hiring slows down in the face of the trade war. Dominant firm power and weaker labour laws may explain why wages are not rising particularly fast (around 4 percent per annum, lower than before the global financial crisis and much lower than the 1984-2008 average).
This shift is more than automation. Fundamentally, the economy is changing.
A better approach is one which recognsies this. Instituting more worker protections and policies which create resilience, support upskilling, and help people prepare for new modes of work, and the lifestyles that entail seems to make sense. Psychological and cultural barriers need to be overcome, catalysed by articulate political leaders. And the economy needs to be positioned to rapidly create lots of new jobs to support those lifestyles of the not-so-distant future.
Should governments intervene in these sorts of ways? I’d be curious to hear your views in the comments.
(🎧 Listen to my discussion with another presidential candidate, Andrew Yang. Andrew is a proponent of universal basic income.)
🎭 The next hot job is pretending to be a robot, says EV reader, Christopher Mims. Many drones and ground robots are not sufficiently autonomous and are managed by teams of humans. This may be a hot job, but it will only involve small numbers of people. In many cases, single human operators will manage several machines remotely.
⚠️ Information gerrymandering is possible on social networks. This can ‘skew perceptions of how others in the community will vote—which can alter the outcomes of elections.’ Microtargeted political advertising could offer a ‘surreptitious and potent tool for information gerrymandering.’
😉 Hilarious, yet informative. If you are aiming to get your head around cryptocurrencies, this rap battle between Alexander Hamilton, who founded the Fed, and Satoshi, who created Bitcoin, is useful.
Dept of AI
Building models of causal explanation will be an important building block to a more powerful artificial intelligence. Tim Maudlin’s accessible review of Judea Pearl’s ‘The Book of Why’ concisely explains one approach.
Of course, the alternative to causal modelling is the ‘chuck data and computation’ approach. This has the fancy name ‘empirical computation’ and is making strides in the cross-over between AI and biology. (Worth reading to contrast with the approach above.)
🇨🇳 China has already surpassed the US in published papers on AI, according to the Semantic Scholar project. Additionally, researchers found that if current trends continue, China will overtake the US in the most-cited 50 percent of AI papers this year, the most-cited 10 percent next year, and the most-cited 1 percent by 2025.
One plank of China’s AI strategy is the home-grown semiconductor industry. As much as 14 percent of China’s chip demand is now met locally. One industry boss reckons that proportion could almost treble within a decade. The maturing of 5G, AI and IoT applications may actually mean that China is ‘standing on the same starting line for once and not playing catchup’. After all, some of the most interesting chips that we’ve seen come out of Western markets recently are not from stalwarts like ARM and Intel but from youthful firms with new architectures, like Mythic (founded in 2012), Cerebras and Graphcore (both founded in 2016).
The Allen Institute for Artificial Intelligence has successfully created an NLP system capable of passing an 8th-grade science test. Aristo, which is built on top of Google’s earlier language model, Bert, is not really comparable to human intelligence. It could be useful for a range of applications including record-keeping and search engines.
🌍 The algorithmic colonization of Africa: Adeba Birhane argues that tech communities in Africa need to engage more critically with the implications and embedded values in the technologies; the culture around those technologies, which they are importing from the West, is just as important. Good critical read.
♂️♀️ Pew has put out a tool which lets you ‘see’ gender as a computer does. The data behind the tool helps to better understand how machine vision works.
How AI is changing warfare: The Economist surveys the possibilities of an arms race in military AI.
Dept of creative destruction
The transition to a new style of economy, with new economic arrangements, continues—and is increasingly clearly reflected in the types of firms that are the world’s largest.
This week, the recidivist oil major ExxonMobil, dropped out of the top 10 largest firms in the S&P 500 Index for the first time since the inception of that index in 1957. British retail stalwart, Marks and Spencer, fell out of the FTSE100 index. (M&S was the byword for affordable mass-produced quality for decades.) Only twenty-six of the FTSE100 from 1984 still make it on the list.
We should expect the make-up of industries to change if the thesis of this newsletter is correct. PwC’s recent Global Top 100 report shows what has happened over the past ten years:
- Big companies are massive. The largest firm of 2009, ExxonMobil, had a market cap of $337bn. Today’s leader, Microsoft weighs in at more than $900bn. (But note that the growth in the market cap of the top 10 firms has relatively underperformed the overall S&P500 index, which would suggest marginally lower levels of concentration.)
- The balance sheet of these largest ten firms is dominated by intangible assets and retained profits. Microsoft’s reported some $44bn of tangible assets on its balance sheet. ExxonMobil, a much less profitbale firm, reported nearly $250bn of long-term physical assets on its balance sheet at the end of 2018. (See our briefing with Stian Westlake on the intangible economy.)
- The US share of the Global Top 100 has grown from 45 percent to 63 percent by market cap, at the cost primarily to Europe. The biggest risers driving this growth were Apple, Amazon, Microsoft and Alphabet. These firms added $3trn in market cap, accounting for the vast bulk of the gains in the US share of the Top 100.
The scale of these platform companies is hard to comprehend. And it does sometimes feel like they are collectively waving a huge middle finger at us.
Take Google. The headline on this article gets straight to the point: ‘Google is a bald-faced IoT liar and its Nest pants are on fire.’ Google is shutting its Nest service, meaning that a bunch of integrated IoT products will lose functionality. One of the draws of Web-based protocols was the ability to coordinate across applications and devices. Common protocols meant collaboration. It is what made the Internet. The behemoths are now choosing to enclose what were pseudo-public spaces to the detriment of open platforms. It breaks the original Nest promise in favour of Google’s business model. But we’ve already paid for our Nests. I hope robust open platform alternatives make their way to the market pdq.
Facebook can’t stop expanding. It has launched a dating service in the US. The app will make use of the strong data footprint Facebook has on users. The app will be run as a separate app, with separate user experience. (See this for more details.)
Brian Norgard, formerly the chief product officer at Tinder, points out the importance of being able to signal through the brand. (Match Group owns Tinder, Match.com and Hinge.) A recently single friend of mine concurs telling me that the experience (and dates) he found on Hinge, Tinder and Bumble were all quite different.
👴 Facebook is for older people. British pensioners now outnumber teens on Facebook by 1.5 million. (There are other parts of Zuckerberg’s empire, like Instagram, still frequented by youth.)
WeWork, the cultish landlord, raised its last round of finance at a valuation of $47bn. It may go public at $20bn. Giles Turner at Bloomberg has more juicy details of WeWork’s current challenges.
🛒 Excellent profile of Shopify and how it is supporting a new class of independent retailers. Interesting fact: Shopify is now worth more than eBay.
The drone bubble is bursting.
🔥 Schadenfreude: The 167 biggest tech flameouts in history.
🌪️ Climate breakdown: 408.59ppm | 3,909 days
Each week, we’re going to remind you of the CO2 levels in the atmosphere and the number of days until reaching the 450ppm threshold.
The latest measurement (as of September 5): 408.59ppm; 12 months ago: 405.66ppm; 25 years ago: 360ppm; 250 years ago, est: 250ppm. Share this reminder with your community by forwarding this email or tweeting this.
The Bank of Montreal is cutting down its reinsurance business, partially due to climate change risks. The Bank has said that climate change ‘poses physical risks from disruptive weather events and transition risks from adapting to a lower-carbon global economy.’
🌞 Without subsidies, solar is now cheaper than coal in 22 percent of Chinese cities.
🚗 Electric cars have lower lifetime emissions than petrol or diesel cars.
Short morsels to appear smart at dinner parties
💬 Human speech has a universal transmission rate of 39 bits per second. (Full paper with interesting graphs here.)
The European Space Agency narrowly avoided the first traffic incident in space.
Looks like uBiome may have misled investors in its $83 million investment round last year.
🥇 Runner, Ellie Pell, took home two trophies for winning a 50k race because the organisers didn’t expect a woman to win.
🍿 The movie version of this true story about how an Iranian engineer, working for Dutch intelligence, played a crucial role in the famous Stuxnet cyber attack practically writes itself.
Barcelona has a vision to take back the streets from cars.
💲💲💲 A bitcoin holder transferred $1bn of the cryptocurrency in a single transaction. As crypto-legend Nick Szabo points out, this degree of ‘confidence in Bitcoin is splendid but tempts fate.’
British readers will recognise the political conniptions over the past week. For non-British readers, these shenanigans this week are like nothing I’ve seen (or studied) before. But they are at least getting citizens to engage in politics. More than a million tuned in to watch BBC Parliament, a channel dedicated to showing debates in the House of Commons.
I’m sure our once stable polity will provide for some good watching next week, too!
What you’re up to—Notes from EV readers
Congrats to Nigel Verdon for raising $10 million in Series A for his open banking and compliance platform, Railsbank.
EV’s Elise Thomas was on a podcast talking about Libra and national security; her team also put out an analysis of the Chinese state-linked information operation targeting the HK protests on Twitter
Katie Irani and Shehnaz Suterwalla are inviting you to join them at the free, thought-provoking design school inside the Victoria and Albert Museum in London starting next week.
Edward Saperia and his team have produced the 2019 Election Tech Handbook: a crowdsourced resource for technologists building things for the general election and politics.
James Pember ponders the future of music streaming and the holy wars between algorithmic and human curation.
Review of Brett Erlichman’s Re-Engineering Humanity.
Daniel Stanley is organising a summit in London on Sept 23rd to tackle how the advertising industry contributes to issues like hate speech, invasive tech and the climate crisis. 40 percent off tickets at this link.
Pearse Keane just published a paper evaluating a deep learning software to develop medical image diagnostic classifiers by health-care professionals with no coding skills.
Tom Campbell and his team at FutureGrasp explore why state AI initiatives should be of strong corporate interest.
Congrats to Ed Challis whose firm RE:Infer won the best startup award at FinSum in Tokyo!
Marsi Parikh writes about why non-linear systems thinking is essential for the wicked problems that lay ahead of us.
Congrats to a number of founders and investors whose firms are celebrated amongst Britain’s hottest 25 places to work.
Share your news and projects with email@example.com