Dept of near future

📸  “With millions of cameras and billions of lines of code, China is building a high-tech authoritarian future,” argues Paul Mozur. Fascinating detail of the scale of China’s watch list (20-30m people), how it is built by private firms and how it might be used by the state to manage increasing social disruption. (See also: profile of Megvii, the second-largest face recognition company in China. Of course, consumer face recognition isn’t quite where it should be. The iPhone struggles to recognise many morning faces, including that of your humble curator. And also, Facebook’s Densepose turns 2-d photos of people intro 3-d models, which in turn could form a part of novel ‘surveillance architectures’ allowing authorities to monitor large numbers of people at once.)

⚖️ McKinsey, facing an employee rebellion, has dropped a contract to help America’s ICE department. Google has similarly faced employee concerns over military contracts, and Microsoft has publicly stated they won’t allow their AI tech to be used in certain applications. The interesting bit? Google & McKinsey were operating within local laws, but not within the squeamish boundary of their employees. On a similar track, office broker, WeWork, will no longer allow employees to expense meat nor will it provide meat at corporate events in order to encourage employees  “to reduce their personal environmental impact”. This idea of companies as ethical and policy stewards is an ongoing theme which we will return to next week. See also, Whitney Dailey’s brief argument on why corporate purpose can help the bottom line.

🕳️ The trouble with the US AI policy is that the current government doesn’t have the expertise and corporates have too much power.

💭 Indy Johar on the UK’s great restructuring: a thought-provoking essay on the complex interplay of Brexit, technology disruption, climate change & the hollowing out of middle management & how we should deal with it.

Dept of quantum computing

In the past few weeks, I’ve ended up learning a little about quantum computing. I’ve enjoyed discussions with leaders from Microsoft’s quantum team, IBM Q, Ctrl-Q, IonQ and the Cambridge Quantum Computing centre.  And I attended an introductory class on quantum computing hosted by Kathryn Parsons at Decoded. I even saw a model of the distillation refrigerator of an IBM quantum processor. The processor needs to be cooled to close to absolute zero to operate. (For more amazing pics of how liquid-helium-punk these rigs are, check this out.)

Quantum computing promises to be a totally novel form of computation, bringing a mastery to the physics unveiled by De Broglie, Schrodinger, Heisenberg, Bohr and their chums less than 100 years ago. I won’t dig too much into what quantum computing is today, but if you need a primer watch Julie Love’s accessible 20-minute presentation.

The critical thing to understand about quantum computing is that it will deliver a tremendous amount of affordable computing power, compute that we need to solve key problems in the world. Assuming we can get enough stable “logical” qubits--the fundamental building blocks of quantum calculations--to build a working computer, the technology could turn vital calculations that are currently intractable into soluble ones.

One example of a quantum speedup is in search problems. These grow exponentially complex in classical computers, but sub-exponentially using Grover’s algorithms on a quantum computer. For a 1 million item path, a quantum computer takes 500 times fewer steps than a traditional computer. To give another practical example, consider trying to understand how FeMoco, the chemical cofactor of bacterial nitrogenase, helps make those helpful prokaryotes so darn energy-shrewd at nitrogen fixation. The Haber-Bosch process consumes about 3% of the world energy, learning from the bacterial approach to nitrogen fixation would have a huge green dividend. There are other wins at the new frontier of novel biochemistries, including carbon capture, identifying new materials and exponential improvements in drug discovery that could be unleashed by quantum computing. Other applications include machine learning and logistics.

The physics has long said we can build these machines. It was in the early 1980s that, most famously, Feynman & Deutsch, initiated the field. Some experts argue that in the past couple of years we’ve made significant progress in tackling the key issues of building quantum computers, and their components, qubits. The biggest technical issues relate to error-correction, coherence and scaling the number of qubits to usable. But I hear time and again from quantum experts that these problems are now increasingly in the realm of engineering, rather than science. (Watch Oxford’s Simon Benjamin discuss this point.)

Today we actually have working quantum computers which are accessible via web services. The best known is IBM’s Quantum Experience, a 20-qubit quantum computer. The more technically au courant might like this brief write-up of using a Rigetti quantum computer to calculate the binding energy of a deuteron.

Here is the twist. If quantum computing promises so much including in the not-so-distant-future, quantum computing giving humanity its biggest dividend ever - surviving or mitigating the impact of climate change - then shouldn’t we be doubling down on it?

This would lead to expectations of huge levels of investment. A multi-trillion dollar payout in the foreseeable future is worth something today, even discounting for risk.

What we actually see is a very nascent field, with investment levels slowly rising, but the field remains tiny. We looked at three vectors: what had governments announced, what did we find from public announcements of investments in quantum computing companies, how much had large companies invested in the quantum computing domain.

What we found was surprising.

Given the large potential of quantum computing, the actual investment levels are low (with one exception, see at the end of this). We reckon, from a rough LinkedIn count, that fewer than 2,000 people are involved in companies working in any part of the quantum stack (and that includes all the marketing and PR types attached to these groups in large companies). IBM Q, which runs a developer ecosystem via the IBM Quantum Experience API, should have the deepest team. My scan on LinkedIn (far from perfect) shows only 300-or-so names attached to quantum computing in all of IBM. The startups are of similar scale, Rigetti, numbering less than 150.

Some estimates go beyond this. There are 7,000 researchers working on quantum computing around the world, a more healthy but still small number, according to the European Commission.

The venture dollars flowing intro quantum computing is small with D-Wave ($175m), Rigetti ($70m), Cambridge Quantum Computing ($50m) and IonQ ($20m) leading the pack. The European Commission further estimates that total global annual investment in quantum research is some €1.5bn per annum. (Deloitte has further estimates: suggesting that there is about $2.2bn investments globally by governments in quantum computing.)

So quantum computing is Schrodinger’s opportunity, simultaneously here and not here at the same time.

On the one hand, quantum computing is getting all the accoutrements of a technology close to maturity (press briefings, Gartner reports, analysis from investment banks and management consultancies) and the large tech firms are trumpeting working systems within 5 years.

On the other, investment levels are tiny by the standards of what large companies can put to work, or what VCs invest (Wag, a dog-walking app, recently raised $300m.) This suggests that these smart investors are discounting the potential upside very heavily, i.e. there are many hurdles, which these investors cannot easily enumerate, to overcome or the time frame to realise is very long.

Which is it? Close to maturity or facing a long journey?

Here are the three things holding us back:

1. The science has been hard. This hasn’t been metaphorically quantum physics, it has really been quantum physics. However, as many of the players in the field argue, lots of the problems are now engineering rather than science problems. And there are lots more engineers than theoretical physicists out there.

2. Venture investors need returns within a time frame and with less certainty. The fact that there are only three venture-backed startups building the hardware (D-wave, with their application-specific system, Rigetti & IonQ) suggests that most venture investors are more cagey about whether this really is the realm of engineering vs. research, and just how much capital it will take to make these things work.

3. Governments haven’t stepped up to the plate as the “investor of first resort” in basic technologies, in recent years. (I have a podcast coming up on exactly this topic with venture capitalist & economist, Bill Janeway, in a couple of weeks.)

In fact, it seems to be today second & third reasons that are the main attenuators. There is still too much uncertainty about when the technical barriers will be overcome and we can actually have useful quantum computers accessible. And many governments are chary about really backing core research, despite the ample long-term evidence of their positive impact in fundamental domains.

In this excellent BCG report, which I recommend reading, the consultants’ estimates for the size of the quantum computing market by 2035 vary by a factor of 28, from $57bn (in their high case) or $2bn (in their base case)!!

The projections suggest:

A substantial market for quantum computing, but the timing could vary widely depending on when the critical technical milestones are reached that unlock actual computing capacity.

In other words, the mid-term potential for this market is rather binary, dependent on when we actually get it all to work.

Of course, there is an exception to all this. And that exception is China where a National Laboratory for Quantum Information Sciences will attract between $10-16bn of investment when it opens in two years. This is pretty significant, likely enough to put China in the lead in developing this technology. China’s Alibaba also has a public-cloud implementation of quantum computing, an 11-qubit machine, at this point smaller than IBM’s cloud-accessible device.

Only two years ago, I was pretty sceptical about where quantum computing was in its cycle. And as Jerry Neumann points out, quantum computing was only “five years away” back in 2000, so quantum could be one of those technologies, like controllable fusion, that is always just around the corner. But something feels a bit different this time… I’d just love to see higher levels of investments in this exciting field.

---I’m intrigued about speaking to founders tackling the problems in the quantum computing domain, including associated toolsets or applications. Early stage, pre-A-round, preferably, but not exclusively, based in the UK. I’m also happy to look at more mature teams. If this fits you or someone you know, hit reply.---

EV reader Rodolfo Rosini has his own take on why there have been so few investments in quantum computing.

Dept of AI

Some interesting quick fire nuggets on AI & automation this week:

• Automation in Asian markets may lead to the rise of slavery as "displaced workers without the skills to adapt or the cushion of social security will have to compete for a diminishing supply of low-paid, low-skilled work in what will likely be an increasingly exploitative environment.”

• Blockbuster AI research is getting more expensive, says Ryan Carey. AlphaGo Zero probably cost $10m and current trends suggest these experiments are getting an order of magnitude more expensive every year or so.

• Using AI to manage natural disasters.

• Love this example of using satellite imagery and deep learning to accurately identify solar panels across Australia.

• Another lovely use: Flipkart using machine learning to make Indian addresses more tractable.

• Scriptbook is an AI which predicts the success of a screenplay.

• On new military AI research centres from Russia, the US and China. See also, how Tsinghua University plans “Civil-Military fusion” in AI research. It intends to “closely integrate the national strategy of military-civilian integration and the AI superpower strategy.” One interesting example is the creation of a 'Computational Legal Studies' Master of Engineering program in the School of Law, which is also an attempt to integrate the school’s AI and liberal arts so as to try a brand-new specialty direction for the subject.”

Short morsels to appear smart at dinner parties

What do populist parties in Europe have in common? Less than you may think, but the one value all the right-wing populist parties share: a dislike of immigrants. (Academic paper.)

♀️ Women invent the future is an anthology of sci-fi stories written by women, published by our friends at DotEveryone.

Hang out with positive people: “certain health behaviors appear to be contagious can influence obesity, anxiety and overall happiness.”

🐦 Twitter is finally tackling fake accounts. 70m suspended already.

Magic Leap, the company building a mixed-reality headset, seems to be gaining more sceptics than acolytes right now.

Pokemon Go hits $1.8bn in revenue in two years. (Impressive, but pales in comparison to Fortnite.)

Running up to the financial crisis, investment banks loved pushing high-coupon structured notes, bets on baskets of bank and mortgage stocks not falling below a certain floor, to investors. They were horrible products and essentially called the top of the market, leaving yield-hungry investors nursing losses to their principal. These notes are back again, now the basket is on the FANGs. Is this a sign of structuring desks calling the top for our digi-leviathans? And won’t bankers ever learn?

👔Wearing a tie may restrict blood flow to the brain.

Russian info-ops sought to discredit local US news outlets for years. (Also, “the warning lights are blinking red”, says a top US intelligence official on Russian cyber attacks.)

Half of young Brits want an electric car.

Baidu launches a blockchain-based stock photo service.

🤯 Arise, disodium helide! Yes, helium can form compounds. The bonding is just weird.

End note

My week was pretty chemical. It started with discovering the helium can form compounds, something that flew in the face of everything we’re taught in chemistry. I was late to learning about this marvellous compound, but it is too marvellous not to share. Then I had a splurge of quantum computing and its applications in chemistry. This prompted me to write a short essay on the topic.

And to wrap it up, I heard the a brilliant joke, which I just have to share with you.

Question: Why should you never trust atoms?

The answer is below our sponsor message...

Have an amazing week of discoveries and growing.
Hasta la vista,

😎 Azeem

⚗️ P.S. Disodium helide deserves a wider audience. You can do this by sharing this week’s issue with your friends. Forward | Tweet | Share on Linkedin