Monday, May 9, 2016

The Continuance of Growth

Timothy B Lee, a technology reporter I'm generally fond of, seems to think we are near the end of technologically-lead productivity growth (outside of IT, medicine, and transport), and that the future is just (i) the rest of the world catching up to where America is now, and (ii) a lot of status competitions and endlessly bidding up the price of positional goods. There's no room in his view for continued improvements in consumer welfare.

This confession is rather shocking, as it shows an incredible myopia about the prospect of near-future technology, and also a lack of imagination about future consumption goods that aren't positional. It really reminds me of the apocryphal quote of Charles Holland Duell, who allegedly said that "Everything that can be invented has been invented" (only Tim is actually saying it!). I know my own position is much closer to what Duell actually said in 1902:
In my opinion, all previous advances in the various lines of invention will appear totally insignificant when compared with those which the present century will witness. I almost wish that I might live my life over again to see the wonders which are at the threshold.
And of course he was right. The century that followed was amazing and saw technological improvement and economic growth that would seem to be science fiction or fantasy to anyone from previous centuries. "Common workers having second homes or vacationing on other continents?? Please!"

In this essay I will argue that Tim is wrong on all counts. There are prospects for dramatic economic growth, both in and outside the sectors he identifies as ripe for improvement; these prospects will be objective improvements over the quality of life of even rich Westerners today; and these improvements will be non-positional, so their price will track marginal costs of production rather than auction pricing.

I. Further Automating the Production of Consumer Goods

Consumer goods are not positional, so their prices track with marginal cost. If you can lower the marginal costs, the prices fall. We have seen this marginal cost improvement most dramatically these last few decades in the production of transistors. A new phone today can replace a multi-million dollar super computer and a whole store's worth of consumer electronics from 1990. But other consumer goods have also gotten cheaper, and will get cheaper yet.

Tim dismisses food as already "too cheap to matter", but most calories consumed today are in the form of a few crops (wheat, corn, rice, soybeans) and their manufactured oil byproducts. This isn't because they're really that good for us or tasty, but because they are amendable to automated farming and thus cheap. One farmer can tend thousands of acres of corn. The same cannot be said of fruits and vegetables, and if you asked Tim I bet he'd agree that Americans don't consume enough of them. And of course they're expensive because they're labor intensive. Fruit and vegetable farming require large numbers of low-skill migrant laborers to work long hours in the sun doing repetitive work. Machine vision and handling is now reaching the level of quality needed to automate the planting, tending, picking, sorting, and packing of these crops, and the price declines that follow will be substantial. Also, lots of labor will be freed up for more productive activity.

Drone aircraft are also playing a role in the further automating of agriculture. Drones can patrol over fields on a near-constant basis, observing them for water issues, blight, ripening status, soil health, etc. This technology is already being deployed.

Healthy proteins are also expensive, and getting more so all the time, as cattle and fish farming push the limits of Earth's biosphere. But we are on the verge of using biotechnologies such as CRISPR to produce as much edible proteins as we need from plant sources, even algae. One study I saw suggested that algae ponds about the size of Rhode Island would produce enough protein to meet the needs of every human currently alive on Earth. If being able to feed everyone on Earth with such a small input of resources isn't growth, I don't know what is.

Tim admits that personal electronics are a growth area, but doesn't seem to think about what it means that factories in China are automating the production of phones and other tiny devices with dexterous robots. Lights-out factories will eventually be able to produce any number of phones, VR headsets, etc. for the cost of materials and electrical power. And that's how you get the equivalent of $50 Android smartphones in every product category. Instead of rich people having one VR headset the family shares, there will be devices of equal complexity in every room.

Transport is also a huge consumer good, both consumed directly (our cars) and indirectly (shipping of products). All forms of transport from the largest cargo planes and ships to personal mobility pods are about to become drones, and we will stop owning our cars directly but rather call them up as needed and share the cost burdens with the other passengers. Automated fleets of long-haul trucks and planes will transform supply chains, and self-driving cars will change our daily life in ways we probably can't imagine now. Of course Tim admitted transport was sector was ripe for disruption, but it also sort of contradicts his point about only catch-up growth existing going forward. Self-driving cars are analogous to having a personal driver take you everywhere, which is not something the top 20% of America have access to. That's a Top 0.1% thing. But soon we will all have it.

The most significant consumer good of all though, by dollars spent, is construction (both residential and commercial). A large part of why housing is expensive in many parts of America today has nothing to do with technology, and is all about the limits on developing new housing supply imposed by local law. But even without those limits, building houses, offices, factories, etc. consumes a great deal or labor and is time-consuming as well. This limits the size of the interior space that any one human can reasonable afford to consume. But if technologies like Broad Group's factory construction or contour crafting takes off, you'll see huge efficiencies in this sector which will translate into growth. Maybe London and New York apartments won't have to be the size of shoeboxes in the near future.

II. Automation of Services

Tim mentioned private jets as one the few things that "the rich" have that everyone else doesn't, but I think this is overly focused on physical goods over services. Most economic activity is services these days, after all, and the rich consume a lot of them. Those are going to become democratized with the near-term explosion of machine learning and automated services.

Tim mentioned medicine as being ripe for improvement, and he's right. But this is more than catch-up growth by making medicine cheaper; healthcare is about to go places that rich Westerners today can only dream of. Look to the convergence of AI (such as IBM Watson), Sensors (such as FitBit and Open Water), and Robotics (such as this robot surgeon) to imagine the possibilities. The ability to have ubiquitous sensors feeding information into a personal AI physician is unprecedented in terms of healthcare management, and it will be available at the cost of software (nearly zero) with necessary machine-precision surgical interventions available for the cost of mass-produced machines (pretty cheap, and consistently higher quality than hand-made stuff).

Another example is private banking and financial advice. They've been unaffordable to the poor for a long time partly because they require a lot of human labor and partly because the banks that sit at the center of the global financial sectors collect monopoly rents and brokerage fees, keeping things expensive (and making financiers very wealthy). Both of those factors are going away. Finance is being decentralized thanks to open financial networks like Bitcoin, and banks sitting at the center of all the webs are about to starve to death as financial transactions make an end run around them. Also, the rise of "robo advisors" is the first step in AI offering tailored, personal investing and cash management advice to everyone for the marginal cost of software (aka, zero).

Similarly, education is being democratized. Artificially intelligent tutors are bringing personalized education to every corner of the world for the cost of an Android laptop and data connection. Although Tim calls this "catch up growth", because aristocracy have always had private tutors, it is non-linear growth nonetheless because there will be much less "cognitive waste" in the world. Instead of brilliant minds being wasted tilling fields in rural India or Africa, they will have the tools to meaningfully contribute to humanity. Doing more with the (biological) capital you have is growth.

Do automated tutors solve the issue of Harvard having limited admission slots? No, but you can solve that problem by other means. Getting into Harvard, after all, isn't so much about getting the education as sending the signal that you have abundance intelligence and drive, and it's that signal (not the English criticism classes) that opens up career and marriage opportunities. But there are other ways of generating those signals that don't need to cost $200,000 or more. One simple example is an IQ test, and I'm sure you can think of others. Start a company, for instance.

III. New Frontiers of Growth (more speculative)

Robot vision and AI systems exist today and its easy to see how they can be developed just a bit more to deliver value in many sectors if you just take 10 minutes to think about those sectors (and not rely solely on trending news topics). But there are also technologies that are more speculative, and offer the prospect of amazing growth in the future.

Biotech is one of them. CRISPR technology is only a couple years old, so it's hard to judge just how transformative this is going to be, but initial prospects suggest "very" transformative. Gene transplants won't just fix people with bad mutations, but will give ordinary humans super-powers, such as the ability break down and excrete coronary or amyloid plaques. Lengthening telomeres and rejuvenating our immune systems and supply of stem cells will extend healthy life by decades. Algae will be given the genetic machinery to produce drugs, useful industrial materials, and even edible proteins that could replace all cattle farming with a few bioreactors. If living decades longer in great health and being able to easily feed 10 billion humans isn't economic growth, I don't know what is.

3D Printing is a big deal. I'm surprise Tim didn't think about this. It won't replace mass-produced injection-molded plastic, but it will allow entirely new abilities by taking advantage of micro-geometries and reducing the complexity of aerospace parts. SpaceX and Blue Origin are already using 3D printing to make engines that traditional technology simply cannot make, which is how SpaceX is able to advertise the Dragon spacecraft's ability to land mass on any solid planet or moon in the solar system. Boeing, Lockheed Martin, and Airbus have also started using 3D-printing to replace complicated parts in terrestrial aviation with simple but previously-impossible-to-make parts, so this matters for stuff here on Earth too. Microlattices, for instance, cannot be made by any other means and offer amazing physical properties.

Speaking of space access though, having access to all the energy and physical resources of the solar system is sort of a big deal. Like how "discovering the New World" was a big deal for Spain. A single and not terribly large platinum-group metal asteroid would have more valuable metals and rare earth elements than have been mined from Earth in all of human history to date, and all you need to refine them out is a big magnifying glass. The asteroid Ceres has more water than Earth does. A single Bishop-ring habitat spun out from a carbonaceous asteroid would have a internal surface area about the size as India. Solar power is rather abundant, to say the least. And so forth. You might think these are fantasies, but that's because you haven't thought about how cheap reusable rockets are going to be, or what efficiencies using the energy and materials already in space will bring.

There are also a lot of companies out there trying to develop next-generation nuclear energy, both fission and fusion. We only need one of them to succeed to allow for nearly limitless energy growth. And once one of them succeeds, we will find a way to use that energy, just as software has found ways to use all those transistors that Intel, NVidia, and ARM keep making.

IV. Time and Labor Saving Consumer Devices

As for saving the time of average consumers, currently the biggest time wasters at home (not just for me, but in general population surveys I have seen) are cooking, commuting, yard work, and laundry. At least the first three of these are about to go away. Cooking will be replaced with robotic kitchens and drone delivery. High quality foods (including cheap fruits and vegetables!) will be deliverable from a smart phone app, so average folks will have the equivalent of a personal chef. (And on the restaurant side, much of the kitchen staff costs will go away) Commuting of course is about to be zeroed out with self-driving cars. You can spend that time working on productive tasks or relaxing instead. Yard work will be replaced with robotic lawn mowerstree trimmers, and weed removers.

Also it's worth pointing out that these things aren't going to be cheaper just because they replace human labor with machine labor. They'll also allow new business models that make do with a lot less hardware than we have today. You don't need your lawn mowed every day, for instance, so an entire street should share of single robot lawn mower or subscribe to a law mower service that sends out unattended mowers just as near-future Uber will send out unattended taxis. Self-driving cars will be shared cars. Tree trimmers will certainly be shared. That's a once-a-year thing after all.

As for laundry ... I don't know. Maybe one day. We're probably more likely to invent disposable clothes made out of paper before we make a robot that folds clothes. Or not. Hard to say. It's a hard problem.

V. Conclusion

At this point you probably think I'm a hopeless technological optimist, but to me this is just common sense. The timelines are fuzzy, and everything turns out to be a little (or a lot) harder than first imagined, but this sort of stuff does come true eventually. Predictions about cheap solar power have finally come true. Predictions about calling services from your smartphone have come true. Predictions about the increasing power efficiency of smartphones and graphics cards have come true. Machine learning is already exploding and delivering real results, most visibly with self-driving cars.

To come back to the topic of Tim's post, I again suggest that it's shocking that someone who allegedly reports on technology can underestimate the consumer surpluses and new 'super powers' that are about to descend on us. I'd suggest for anyone who thinks along the same lines to do the following exercise: 1) Make a list of all the sectors our economy spends money on, and 2) imagine how robots can replace that. It's going to happen.

No comments:

Post a Comment