The new economy of platform companies explored. How Kodak invented, and buried, digital photography. Unveiling Buzzfeed’s numbers. Exploring the ethics of algorithms and more progress from the solar frontier.

Share this (please!): Twitter | LinkedIn | Facebook new

Forwarded by a friend? Subscribe

Dept of the near future

🌍 Networks and the nature of the firm. Excellent essay by Tim O'Reilly on the new economics of platform companies. Must read

👴 Debunking the great man myth of innovation. Musk, Jobs and their ilk succeed with the help of many others, including the state

💉 Designing life one base pair at a time. Nice feature on gene synthesis. (Also - $120m for Editas a Crispr-CAS startup.)

👱 Humans will be an important source of competitive advantage even as automation improves. Machines spread rapidly in an economy and their competitive advantage trends to zero as we reach a new normal.

🔥 Buzzfeed’s exponential growth: Revenues up 16x in 2 years; editorial investment up 14x. Rare to get a look inside exponential scaling. Enjoy it

🔮 A Kodak engineer developed digital photography. However the company refused to capitalise on it directly, that much we know. In this great profile, the ambivalence of Kodak’s execs, focusing on their own retirement not the company’s future is nicely exposed:
“When you’re talking to a bunch of corporate guys about 18 to 20 years in the future, when none of those guys will still be in the company, they don’t get too excited about it”. Excellent read

📚 How the physical book might survive.

Dept of renewable energy

Fossil fuels seems to be going the way of the dinosaur. KKR, the “barbarians at the gate”, were wrong footed on a natural gas buyout resulting in $4.1bn write off.
Oil appears to have hit a new long-term normal - prices permanently low. Result: effective dissolution of OPEC with many geopolitical and military ramifications.
I wonder how much of the rapid improvement in solar is being fed intro declining fossil prices? 💡 Ramez Naam has a must read brilliant blog post arguing that solar in on an exponential cost decline and might hit levels of 2c per kWh in 35 years. (US average consumer pricing for electricity is 12c / kWh today.)

All automotive innovation is around self-driving and autonomous-rich cars, all of which have electric, not fuel, powertrains. If in five years time, the top models from the major marques & the cars from Tesla, Google and Apple are all electrical, the impact will be felt by octane suppliers.

Most drivers who try electric vehicles will not return to gas. This is my experience too, having driven a BMW i3 a few times. Driving electric just feels so much more sensible. Apply the pedal to get power, take it off to slow down, minimal use of the brake; and sublime access to instant power. Do try electric if you haven’t.

Also - the Apple Car leaks ICMYI. And is this the “Frank Fearon” referred to as an Apple employee in the news story?

Dept of ethics & algorithms

⚖ Just because something comes from an algorithm, doesn’t mean it isn’t biased Excellent, short read
Can algorithms be moral? A discussion with Google’s in house philosopher.
My take is peri-Gramsci: the product managers, CEOs & data scientists may impose a new type of cultural hegemony. Their implicit world view underpins what classifications and predictions an algorithm can make, what representation of the world is fed in to train those algorithms and what counts as ‘right’ or 'wrong’ in the outputs of those algorithms.
In fact one of the topic classification projects we ran at PeerIndex was actually called Project Gramsci. Why? We knew that however good we got at topic classification we were still only presenting an approximation of our own skewed world-view. (See also Google & gorillas which was obviously a horrible accident but also a painful expression of the issue.)
Also think about the implications in law enforcement & precrime algorithms, for example, these recent revelations with the TSA no-fly list.
(We’ll discuss some of these themes at the AI Ethics dinner on September 21st. Let me know if you want to attend. You don’t need to be involved in AI to make a good participant, happy to have linguists, ethicists, designers, anthropologists or unicyclists. A good backgrounder is this collection of issues raised by Jonathan Zittrain in his inaugural Harvard lecture.)

Little snacks for beach-time reading

💩→🍸 Bill Gates gives a lovely update on the OmniProcessor, a machine that turns human poop into drinkable water.

🐣 How Baidu used deep learning to learn Mandarin. (Accessible blog post.)

How Google enhanced voice-transcription with RNNs.

🎩 An Oxford University course on machine learning is now online. The beginning (at least) looks accessible.

Car companies are dealing with the issues of software vulnerabilities and security updates that software firms have lived with for decades. Details on how Volkswagen spent two years trying to hide a vulnerability in software systems in its cars.

Nice intro to DIY algorithmic trading. I did this a bit back in 08 with simple automated relative value trades, Moore’s Law has marched on.

👰 White, thin and pretty - the best way to overfund a kickstarter campaign.

Positive data on the benefits of the gig economy for workers.

Finish schools phase out handwriting classes. One irony being the the MNIST handwriting dataset is a staple for artificial intelligence researchers starting out. (The Finns have also phased out teaching in subjects)

How Facebook, Twitter and Apple are eliminating online equality. Providing ‘celebrities’ better tools and easier discovery runs counter to many of the democratising advantages of the Web. In my view, Twitter’s suggested user list has been an affordance that has hurt the overall development of the network. Rich-get-richer dynamics often do that.

In issue 19 we investigated the idea that the brain is substantially more complex than previous models suggested - pushing out estimates for the singularity. This 📹 beautiful video takes us into layers of the complexity.

We filmed the entire nervous system of a fruit-fly larva in real time. Amazing.

Modifying crops by spraying them with RNA rather than cumbersome gene-editing. Good intro

☔️ A massive El Nino is on its way. Nickname: Godzilla

Tropical diseases are making their way back into America’s poorest regions. Another side-effect of rising inequality.

🍀 Lovely illustration converting different perceptions of probability to actual probabilities.

IBM’s Watson will whoop you in fantasy football.

🏭 Air pollution in China kills 1.6m people per year.

⭔ Gorgeous. New tessellating pentagon discovered.

End note

Normal service will resume next week. But I am open to feedback on the newsletter. Just hit reply.