[Image by 955169 from Pixabay]
Dear friend,
Welcome to the latest edition of This Week in Disruptive Tech, a newsletter that explores the intersection between tech, business and society. This week we look at AlphaFold, Bitcoins, and Chinese lunar mission. Plus some interesting numbers, tidbits and perspectives that say something about what happens when tech meets the real world.
If you have suggestions, feedback, tips—or anything that you would like me to share in the newsletter—mail me at ramnath@foundingfuel.com or DM me on Twitter at @rmnth. If you haven't subscribed already, you can subscribe here.
Happy reading!
NS Ramnath
What DeepMind’s AlphaFold tells about AI
What's the news? DeepMind, a part of Alphabet, has created an AI system that can accurately predict the three dimensional structure of protein using its DNA sequence. Traditionally, it took months and tens of thousands of dollars to map protein structure. It has been one of the toughest problems in science. DeepMind's AlphaFold can do it in a matter of days. It has huge implications for understanding diseases and discovering drugs.
Why should you care? It shows what AI can do. One of the best quotes from Demis Hassabis, DeepMind founder and CEO, goes like this: “Step 1: Solve intelligence. Step 2: Use it to solve everything else.” When he made the statement, DeepMind was best known for doing very well in a game called Go.
“How did a company best known for playing games just crack one of science's toughest puzzles?”
That's the title of a fascinating piece in Fortune. My biggest takeaway from the story is not ‘that’s the power of AI’, but what it takes to unleash that power. Turns out it’s good old management principles.
1. First of all, don't underestimate the value of games.
“Games are a particularly good arena for a kind of AI training called reinforcement learning. This is where software learns from experience, essentially by trial and error, to get better at a task. In a computer game, software can experiment endlessly, playing over and over again, improving gradually until it reaches superhuman skill, without causing any real-world harm. Games also have ready-made and unambiguous ways to tell if a particular action or set of actions is effective: points and wins. Those metrics provide a very clear way to benchmark performance—something that doesn’t exist for many real-world problems, where the most effective move may be far more ambiguous and the entire concept of “winning” may not apply.”
2. What works for one problem, might not work for another.
“Hassabis’s and the DeepMind team’s first instinct was that protein folding could be solved in exactly the same way as Go—with deep reinforcement learning. But this proved problematic: For one thing, there were even more possible fold configurations than there are moves in Go. More importantly, DeepMind had mastered Go in large part by getting its AI system, AlphaGo, to play games against itself. “There isn’t quite the right analogy for that because protein folding is not a two-player game,” Hassabis says. “You’re sort of playing against Nature.””
3. Real world shifts gears, and so should you.
“When Hassabis’s team was working on Go, Starcraft 2 and Fold projects they noticed that all used to go well for some months, and then, depressingly, nothing will happen. “The company’s management strategy for overcoming this, he says, is to alternate between two different ways of working. The first, which Hassabis calls “strike mode,” involves pushing the team as hard as possible to wring every ounce of performance out of an existing system. Then, when the gains from the all-out effort seem to be exhausted, he shifts gears into what he calls “creative mode.” During this period, Hassabis no longer presses the team on performance—in fact, he tolerates and even expects some temporary declines—in order to give the researchers and engineers the space to tinker with new ideas and try novel approaches. “You want to encourage as many crazy ideas as possible, brainstorming,” he says. This often leads to another leap forward in performance, allowing the team to switch back into strike mode.”
At one point in the NYTimes video “A Rare Look Inside Pixar Studios”, Lee Unkrich, a director, takes the journalist through some of the hard work the studio does to get the movies right, and says, somewhat wistfully: “A lot of people think maybe we push a button on the computer and a movie pops out.” The lesson from Pixar and DeepMind is that it doesn’t.
Dig Deeper
- AlphaGo vs. Lee Se-dol: Why a win for AI is not a lose for humanity - Founding Fuel
- How did a company best known for playing games just crack one of science's toughest puzzles? - Fortune
- Inside DeepMind's epic mission to solve science's trickiest problem - Wired
Data bites
$4.8 billion The value of stock held by Ant Group in Paytm. Reuters reported that Ant Group wants to sell its stake in Paytm in part because of the escalating conflict between China and India. Both companies have denied the news. But that didn't stop the talks about the sale, given the rate at which India has been banning Chinese apps. (Reuters)
-70 degree Celsius The temperature at which the Pfizer vaccine, which has now been approved by the UK, has to be stored. The only way to keep it that cold during transport is by placing it under dry ice, which is even colder at -78 degree celsius. As countries prepare to vaccinate people, the demand for dry ice has shot up. (The Atlantic)
34 terawatt-hours of energy The annual electricity consumption of gaming devices (“equivalent of 5 million cars.”) It's just one of the ways the gaming industry, which earned $120.1 billion in revenue last year, impacts the environment. Playing For The Planet Alliance, a group of gaming companies, has pledged a 30 million tonne reduction of CO2 emissions by 2030, besides introducing green elements into the game. (The Verge)
China and the new race to the Moon
What’s the news? On Tuesday, China landed a lunar probe on Mons Rumker, a previously unexplored part of the Moon. The spacecraft, called Chang’e 5, named after Chinese goddess of the moon, will collect 2 kg of soil and rock samples, guided by mission control on the ground. It’s expected to return to Earth around mid-December.
Why should you care? It’s yet another example of the reinvigorated race to the moon, and harkens back to the 60s and 70s, when the US and USSR competed for space supremacy. In fact, the last time a country picked a lunar sample (170 gram) was the USSR in 1976 and that came a few years after US astronauts got back 382 kg of rocks and soils from the Earth's satellite.
The samples themselves have value. As The Economist memorably put it last year, the moon is “a museum of the solar system’s past.” But, China, which wants to compete with the US for world dominance, as the USSR once did, has bigger plans for the Moon, including setting up a crewed base there.
The focus right now is on the 2 kg sample. China is not doing it all by itself. Many space missions happen with international cooperation these days, and Chang’e 5 is no exception. China is expected to share the samples with other countries—presumably not with India or the US.
The sample also has some deep emotional value. When I was still in school, I once visited the house of a distant relative. He showed me a tray full of smooth, round stones, and asked me to run my hands over it. I did. He said, “now, your hands have touched every part of our great nation,” and went on to explain that his work took him to different parts of the country, and wherever he went, he collected a small stone. He felt one with the entire country whenever he touched them, and wanted to share that experience with others.
Dig Deeper:
- China lands probe on surface of the moon and collects lunar soil (CNN)
Tweet of the week
The second order impact of 10X
What's the news? Bitcoin surged 170% this year, almost touching $20,000—a record high. The previous high was in December 2017, when it touched $19,511, only to see a 70% drop over the next year, 2018. Now, there's a big debate on what will happen to its price in 2021.
Why should you care? In an essay in Bloomberg, Niall Fergusson, historian and author of The Ascent of Money, had this tongue-in-cheek passage.
“This year’s Bitcoin rally has caught many smart people by surprise. Last week’s high was just below the peak of the last rally ($19,892 according to the exchange Coinbase) in December 2017. When Bitcoin subsequently sold off, the New York University economist Nouriel Roubini didn’t hold back. Bitcoin, he told CNBC in February 2018, had been the ‘biggest bubble in human history.’ Its price would now ‘crash to zero.’ Eight months later, Roubini returned to the fray in congressional testimony, denouncing Bitcoin as the ‘mother of all scams.’ In tweets, he referred to it as ‘Shitcoin.’
“Fast forward to November 2020, and Roubini has been forced to change his tune. Bitcoin, he conceded in an interview with Yahoo Finance, was ‘maybe a partial store of value, because … it cannot be so easily debased because there is at least an algorithm that decides how much the supply of bitcoin raises over time.’ If I were as fond of hyperbole as he is, I would call this the biggest conversion since St. Paul."
Even smart people tend to underestimate the impact of new technologies because they miss the second order impacts of disruptive technologies.
In a 2018 essay, angel investor and entrepreneur Balaji Srinivasan explained how this plays out using the example of iPhone. He wrote:
“A new technology is typically not mildly superior to an existing technology in every respect but is instead 10X better on one key axis. That 10X improvement draws customers and provides the capital and rationale for fixing the other defects. The early iPhone camera is a good case in point—while far worse than a dedicated digital camera in most respects, it had one 10X advantage going for it: its ubiquity as a bundled piece of a network-connected smartphone. That led to a rapid rise in use and a concomitant rapid investment in the feature set of network-connected, phone-based cameras.”
Dig Deeper
- Bitcoin Is Winning the Covid-19 Monetary Revolution - Bloomberg
- And What Has the Blockchain Ever Done for Us? - Balajis.com
Quick bites
- US President Donald Trump is annoyed with big tech. Washington Post reports that he has threatened to veto the annual defense bill unless Congress repeals Section 230 which has been shielding Facebook, Twitter, YouTube and sites from being responsible for what users post on their sites. (Washington Post)
- Singapore has given regulatory approval to lab-grown meat, which will allow Eat Just, a clean meat startup, to sell chicken nuggets. Unlike the offerings from Beyond Meat and Impossible Foods, which are plant based, this is grown from animal cells in the lab and raises philosophical questions at least for a subset of vegetarians. If you eat it, are you still a vegetarian? (BBC)
- In Foreign Affairs, Francis Fukuyama, Barak Richman, and Ashish Goel argue that a new set of middleware can solve the problem that regulation, breakup, data portability, and privacy law can't. “A competitive layer of new companies with transparent algorithms would step in and take over the editorial gateway functions currently filled by dominant technology platforms whose algorithms are opaque.” (Foriegn Affairs)
