Software in the Real World

It’s hard to know how to react when someone you’re criticizing insists you don’t actually disagree. Tyler Cowen is a smart guy, so I’m going to take it as a compliment. And it’s true that very little in Chapter 3 of The Great Stagnation directly contradicts anything I say in my last post. Still, there’s a big difference in emphasis between his chapter and my post, and I think readers of The Great Stagnation are likely to get a misleading impression of how the software industry is affecting the “real economy.”

Cowen draws a distinction between “the Internet”—which he portrays as a kind of playground for white-collar workers—and “the revenue-generating sectors of the economy.” The impression that the software industry is largely producing new forms of leisure is important because a lot of people are concerned with their material standard of living defined in a relatively narrow way. Reading The Great Stagnation, one gets the impression that someone making $50,000 in 2010 lived about as well as someone making $50,000 in 1990 (in real dollars as computed by the BLS) except that the guy in 2010 had more fun when he turned on his PC.

The point of my last post was that the 2010 guy’s $50,000 goes farther because there’s a long list of products that he would have paid money for in 1990 that had free, or dramatically cheaper, digital substitutes in 2010. The money the 1990 guy would have spent on newspaper and magazine subscriptions, books and CDs, travel agents and maps, cameras and film development, and the rest is disposable income that the 2010 guy can spend on other “real economy” goods and services like restaurant meals, housing, travel, etc. Which is to say that the 2010 guy is wealthier, even if we only care about goods and services that were in the “revenue-generating sector of the economy” in 1990.

In Cowen’s account, the difficulty of measuring the wealth produced by Twitter and Facebook helps explain why those companies haven’t measurably increased GDP, but we still have to explain why the rest of the economy hasn’t been growing very fast. My point is that software might actually be reducing measured GDP by driving certain industries out of business. If that’s true, then it’s possible that the rest of the economy actually is growing at a healthy clip, but that this fact is being masked by the rapid implosion of a few unlucky industries.

Cowen’s focus on Facebook and Twitter also gives a misleading impression about opportunities for future software-driven growth. The low-hanging fruit of our generation is not just “the Internet,” but software powered by Moore’s Law. Moore’s Law made the modern Internet possible, but it also gave birth to the personal computer, various consumer electronics devices like iPods and smart phones, electronic financial networks, medical breakthroughs (e.g. medical imaging, computational genomics), and a vast array of embedded systems (computerized fuel injection in cars, airplanes with auto-pilot, industrial robots).

Looking at this list, it’s obvious that software innovation is not necessarily (in Cowen’s words) “interior to the human mind rather than set on a factory floor.” The output of a CAT scan is “interior to the human mind” in the trivial sense that it’s displayed on a computer screen, but the ultimate result—spotting a tumor, say, is clearly a “real world” result. I think people have a skewed intuition about this because the PC is the computing device with which they have the most direct contact. But as computers get cheaper and more powerful, they’re going to get embedded in an ever-growing list of devices, and the distinction between “Internet innovation” and “real-world innovation” will make less and less sense.

Self-driving cars are a good example of where things are heading. They will probably put millions of truck drivers out of work, lowering the cost of almost every consumer product. They’ll make taxicabs drastically more affordable, putting taxi drivers out of work and virtually eliminating demand for off-street parking. They’re likely to have significant environmental benefits, as consumers can order only as much car as they need for any given trip. They are likely to save thousands of lives by reducing accidents. They’re likely to transform the retail sector—how often would you drive to store if a self-driving Amazon-bot delivered your orders in an hour rather than 2 days? And of course they’ll have many other consequences we can’t anticipate.

The wealth created by self-driving technology will not be “in our minds and in our laptops” the way Facebook and Twitter are. By using these sites as exemplars of software innovation, Cowen dramatically undersells software’s potential for creating wealth in the “real economy” and producing our generation’s “low-hanging fruit.”

This entry was posted in Uncategorized. Bookmark the permalink.

24 Responses to Software in the Real World

  1. Brett says:

    That makes me think of the music industry. If it implodes, bankrupting the major labels in the process, the overall economic output measured from it might drop dramatically due to its smaller size in financial terms. But we might not necessarily be worse off, not if the new music business produces the music we want to hear, in the formats we want to hear it in.

  2. MNP says:

    Indeed, I’d much rather look at how the software industry is going to create a permanent class of low-skill unemployed. Because automation is only going to increase there simply won’t be enough jobs for the people we’ve got.

  3. Jed Harris says:

    It seems to me that the kind of innovation you describe can’t be called “stagnation” without either a disagreement or a severe abuse of English. So I don’t find Cowen’s claim of agreement persuasive at all.

    His evidence that we’re in “stagnation” is all the unemployment, slow wage growth, and financial crises. But if these are caused by the shift to new, very low cost ways of doing things that’s hardly “stagnation”. And that’s precisely what he suggests at the end of his internet chapter: “We are coming up short on the revenue side, so it is harder to pay our debts, whether individuals, businesses, or governments. That situation means ongoing financial hardships, including crises of sovereign debt around the world.”

    Maybe calling it “The Great Dislocation” would be accurate. Would he agree to that?

    But that raises the question of what things look like after this dislocation proceeds further. And I get the feeling he doesn’t want to think too hard about that.

  4. Luis says:

    I think MNP nails it. For those who are employed, this is a huge boon, even if wages stay the same. But I sense that the jobs being lost in various places are not really being replaced in any meaningful way.

    The only hard numbers I’ve seen on this are from manufacturing, where the gross value of goods produced continues to rise steadily even as the number of people employed continues its decades-long decline. I expect this will be the case in other industries as well- the value of the encyclopedia is now bigger than it ever was, but no one is employed in printing it, writing it, etc.; value of the newspaper will actually be improved in the next few years, but there will only be three of them and none of them will employ a printing staff, etc.

    Don’t get me wrong- I’m a big fan of creative destruction, and think that overall this is a good thing. But we do have to recognize that (in concert with outsourcing) this trend is taking our economy in a direction it has never been before, and that very few people seem to be addressing seriously.

    By the way, if you really want to point at something Tyler missed in his book, outsourcing (both of manufacturing and technology development), was completely whiffed on- not mentioned as a cause or effect. Huge oversight, I think- it’s hard to explain downward wage pressure without it, I think.

  5. MNP says:

    Luis, the way this was addressed in the past was that innovations opened up new industries that — probably because they were new — ended up being labor intensive. But automation happens so quickly and eventually we will build so many increasingly cost effective robots that this level of employment doesn’t seem sustainable. And the thing is, it’s moving up the food chain to increasingly whiter-collar jobs. There are other factors as well: like Tim Lee said, things we consider to be expensive or high end now will be cheaper. I know Matt Yglesias has said that creating and watching stuff on the internet is going to end up being the cheap entertainment of the future so leisure will be cheaper.

    Also I think the manufacturing numbers are not less, they’re pretty much the same as 30 years ago, but the population has grown so much then that the works can produce so much more without having to grow the total number of manufacturing.

    There are a few places I’ve seen this addressed in science fiction, but they are based on strong redistributive systems that seem increasingly unlikely in the real world.

    So that’s why I would prefer to talk about this more.

  6. rj says:

    My point is that software might actually be reducing measured GDP by driving certain industries out of business. If that’s true, then it’s possible that the rest of the economy actually is growing at a healthy clip, but that this fact is being masked by the rapid implosion of a few unlucky industries.

    This makes very little sense. In every economy, some industries are doing better and some are doing worse. Top-line numbers like GDP and unemployment look at the economy as a whole. Saying that the growing economy is “masked” by imploding sectors is functionally the same as saying that everybody in America is rich if you just ignore the poor and middle class.

    The example of the $50k earner is instructive. The amount of media and communications he has access to has multiplied exponentially, but having access to out-of-town newspapers and celebrity tweets aren’t useful to someone whose wage has been stagnating and dollar-denominated costs have risen. The cost of rent, transportation, food, education and most everything else has increased. He may have better entertainment, but he’s still in a cash squeeze and it’s only getting worse.

    Counting free access to information as some sort of hidden GDP is like inheriting a huge storage unit full of canned goods – you can use some of it to cut food costs, but it’s hard to turn the rest of it into cash without spending enough time to make it not worthwhile.

  7. having access to out-of-town newspapers and celebrity tweets aren’t useful to someone whose wage has been stagnating and dollar-denominated costs have risen.

    OK, but that’s not my point at all. The difference is that in 1990 the guy was probably paying for a newspaper, and in 2010 he was much more likely getting news for free on the Internet. Therefore his “dollar denominated costs” have gone down—he can take the $200/year he would have spent on a newspaper and spend it on something else like “rent, transportation, food, and education.”

    Now, if you think that the cost of “rent, transportation, food, and education” have gone up, but this fact has escaped the notice of the people who compute the CPI, then your quarrel is with the BLS, not with me. I think the opposite is more likely, but in any event pointing to other measurement biases in the CPI isn’t a refutation of the measurement bias I’m pointing to here.

  8. rj says:

    OK, but that’s not my point at all. The difference is that in 1990 the guy was probably paying for a newspaper, and in 2010 he was much more likely getting news for free on the Internet.

    But he’s paying for internet! He’s probably spending around $150/month on cable and broadband internet, versus $50/month for cable in 1990. Even if he steals his neighbor’s WiFi, the savings from not subscribing to a paper is not much of a dent.

    As for inflation, it’s undeniable that while it’s not ’70s-era high, it still exists and it is rising faster than wages for workers outside the top brackets. I’d dig around for the right BLS stats, but I’m late for lunch. In the meantime, let me Google that for you:

    http://www.businessweek.com/investor/content/feb2010/pi2010025_902249.htm

  9. He’s probably spending around $150/month on cable and broadband internet, versus $50/month for cable in 1990.

    I’ve lived in 3 different metro areas in the last 6 years and I’ve never paid more than $50/month for cable broadband. For most of that period I could have significantly reduced my monthly costs by switching to DSL. I bet there’s as much professional content available for free on sites like YouTube and Hulu as the 1990 guy was getting for his $50/month cable service–not to mention options like NetFlix’s $8/month streaming plan that wasn’t available in 1990.

  10. rj says:

    Well, I get cable and internet, as do a lot of people (and it costs $130 even without HBO, but that may just be NYC).

    If he goes to the trouble of rigging up his TV to his computer, foregoes all the sports events that come with cable but cost extra online and assiduously torrents all that is not available for free on Netflix and Hulu, he has saved … the cost of a newspaper subscription (while opening himself up to piracy lawsuits).

    He may also have dumped his landline for a cell phone. When I had a landline, I remember paying about $25 for a phone that lasted me until I cancelled local phone service (when I tossed it), $30 for an answering machine (same) and about $30-$40 a month for the service. Verizon’s cheapest plan is $40 a month, so there’s no savings there. Add to that the fact that cell phones become obsolete every couple of years – an additional expense avoidable by the most frugal shoppers, though most people pay at least a little extra for a nicer phone.

    Our $50k-earning friend got a lot more for his communications/entertainment dollar than he used to, but it’s still roughly the same amount of dollars. What he gets is more consumer surplus, which can’t pay the rent.

  11. Ragweed says:

    The cost of rent, transportation, food, education and most everything else has increased. He may have better entertainment, but he’s still in a cash squeeze and it’s only getting worse.

    Actually, he also has access to information that makes it easier to find the lowest prices on many commoditities, not to mention discounts available online (think Groupon here), but that may be of limited impact.

    But there is something interesting going on with education. The Internet is actually enabling an incredible amount of free and low-cost education. If I want to take a class in introductory economics at UC Berkeley, Brad DeLong is posting almost all of the material for his Econ 1xx classes online – reading assignments, lectures (actual video), assignments, exams, answers to the exams so you can self-check, etc. There are scads of other examples of professors who practically live-cast all their classes. Plus you have the academic blogosphere, where you can learn a heck of a lot by virtually listening in on top academics debating just about anything in a variety of fields (with links to the source papers and journal articles).

    What you don’t get from this, and what is increasing in price dramatically, is the degree – the letters after your name that prove to the rest of the world that you should have a certain level of knowledge and abilities. And degrees and credentials exist for good reasons, but in discussion of education we should make a distinction between the credentials and the actual education. I would posit that the former is more valuable as a gate-keeper, but the latter more relevant to long-term performance (ie. the degree will get you the job, but the amount of knowledge you have determines how well you do with it).

    John

  12. Our $50k-earning friend got a lot more for his communications/entertainment dollar than he used to, but it’s still roughly the same amount of dollars. What he gets is more consumer surplus, which can’t pay the rent.

    But that’s not his only option. Our hypothetical consumer can get an entry-level cell phone, which provides much richer functionality for about the same cost, he can use Vonage, which offers comparable service for a lower cost, or he can use SkypeIn, which offers slightly inferior service for a fraction of the cost. He can take his consumer surplus either in the form of more content/functionality or in the form of cash.

    Same thing with video. Today’s cable is more expensive, but you get a lot more content for your money—the number of cable networks increased from 79 in 1989 to 280 in 2002, and presumably there are even more today. Conversely, you can take that consumer surplus in the form of cash by dropping cable and watching stuff online.

    You can’t compare the high-end information product of 1990 to the high-end product of 2010 and then complain that prices haven’t gone down, because the latter is generally a lot better. Rather, you have to compare the high-end product of 1990 with the 2010 product that has equivalent functionality. That product is almost always much cheaper, and in many cases even the free stuff is much better than the stuff people used to pay for.

  13. rj says:

    But that’s not his only option. Our hypothetical consumer can get an entry-level cell phone, which provides much richer functionality for about the same cost, he can use Vonage, which offers comparable service for a lower cost, or he can use SkypeIn, which offers slightly inferior service for a fraction of the cost. He can take his consumer surplus either in the form of more content/functionality or in the form of cash.

    I don’t think you understand my point. Clearly, 2011 man gets more value from his entertainment buck than 1990 man, but the amount of bucks spent remains the same, give or take a few dollars a month. The fact that we’re quibbling over $10 a month here or there is beside the point – he gets value, but he can’t turn that into cash.

    You can listen to a lecture online, but you can’t get a credential (and the income boost that comes with it) unless you pony up tuition, which is getting more and more expensive. You can get access to almost every newspaper in the world, but you can’t print them out and sell them because everyone else has the same access as you. The consumer surplus that comes from better entertainment and information (aside from some decreased transaction costs, per commenter Ragweed) can’t be translated into most of the things people need to get by. It’s as if I asked you for help paying the rent and you gave me Season 3 of Perfect Strangers burned to a DVD. Yes, it’s something that I didn’t have before and it has a nonzero value, but it’s beside the point when it comes to keeping a roof over my head.

  14. You can get access to almost every newspaper in the world, but you can’t print them out and sell them because everyone else has the same access as you.

    This isn’t complicated. The 2010 guy can cancel his newspaper and magazine subscriptions and spend the money saved on other stuff. He can send email and pay bills online and use the money he would have spent on stamps on something else. He can stop buying CDs, listen to customized music streams on Pandora, and spend the money on something else. He can cancel his phone service and make calls using Skype. He can stop getting film developed and spend the money on something else.

    Obviously, any one service doesn’t have a huge amount of money attached to it, but the aggregate value of all these services is substantial. I don’t know exactly how much the average 1990 consumer spent on typewriters, newspapers, magazines, books, maps, cameras, film development, camcorders, music players, VCRs and DVD players, encyclopedias, landline telephones, calendars, address books, clocks and watches, calculators, cable TV service, travel agents, travelers checks, etc, but it’s a lot more than $10/month.

  15. rj says:

    Obviously, any one service doesn’t have a huge amount of money attached to it, but the aggregate value of all these services is substantial. I don’t know exactly how much the average 1990 consumer spent on typewriters, newspapers, magazines, books, maps, cameras, film development, camcorders, music players, VCRs and DVD players, encyclopedias, landline telephones, calendars, address books, clocks and watches, calculators, cable TV service, travel agents, travelers checks, etc, but it’s a lot more than $10/month.

    Aargh. Missing it again. It’s true that we don’t spend money on VCRs and 8-bit Nintendos, but we still buy XBOXes, Blu-Ray players, cell phones, iPads, notebooks, netbooks, cable TV, cable/DSL internet, cell service, add-on data packages for cell service, dashboard GPS boxes, Vonage for landlines, Netflix subscription, digital cameras, digital media, etc… We still have most of the expenses we did before home Internet, but they’re going to different providers for different things (plus some new ones).

    According to the Department of Labor, in 2009, Americans spent $2,600 annually on entertainment, compared to $16,000 on housing, $7,600 on transportation and $6,300 on food. DoL says 1990 data is on CD-ROM somewhere, but I can’t imagine that it’s that much different as a percentage of income. Even if everything required to access and enjoy the digital world were completely free (and it’s most certainly not), it still wouldn’t make up for losing your job.

    To go back to your original post, you wrote:

    My point is that software might actually be reducing measured GDP by driving certain industries out of business. If that’s true, then it’s possible that the rest of the economy actually is growing at a healthy clip, but that this fact is being masked by the rapid implosion of a few unlucky industries.

    You still haven’t explained how more consumer surplus “mask[s]” a growing economy. There are always some industries that expand and others that contract. Aggregate statistics like GDP shows this as a wash unless the growing industries grow faster than the shrinking ones. Which they don’t appear to be.

  16. G. says:

    This is the first series of comments on a blog that I have ever felt informed after reading.

    I do think there is a serious problem if people lose their manufacturing jobs and cannot become trained for new industries quickly enough to support continued consumption. Even those new industries may just require fewer individuals to operate. If a car factory fires 1000 workers and replaces them with industrial robots, we might think “well, now the world needs more robots, so the workers can work in the robot factory.” But the higher efficiencies from automation mean that only 100 workers from the car factory can get jobs in a new robot factory. It seems like the automation from computers reduces the need to labor as much. This can be good for society as a whole, just like a tractor is better than an ox. But it still leaves the car factory worker without a job. The cars he used to make may now be much cheaper, but he has no salary to buy these new, inexpensive products.

    What do you do about a displaced worker as society becomes more productive and the benefits of production go to the factory owners?

  17. Luis says:

    Luis, the way this was addressed in the past was that innovations opened up new industries that — probably because they were new — ended up being labor intensive.

    Of course. And this was generally inevitable, because you quite literally couldn’t do much of anything without some combination of both brains and brawn.

    But automation happens so quickly and eventually we will build so many increasingly cost effective robots that this level of employment doesn’t seem sustainable.

    Robots for mechanical stuff, and computers for distributing/creating intangibles. (i.e., the guys who used to run the printing presses.)

    Yglesias seems to think this will be solved by having increased employment in personal services; i.e., the handful of us who remain “productive” in the traditional sense will all have a masseuse, we’ll spend more on high-end food, etc. I don’t think this is completely implausible (we’ve certainly seen a boom in food the past few years), except of course that we’ve produced multiple generations of people who think that these jobs are beneath them; and given how little pay and healthcare those types of jobs typically provide, they may well be right in some sense. There’s going to be pretty bad dislocation there; not just for individuals (that has happened before) but for entire cultural notions of what a “good job” is and what manual labor should pay (when was the last time that happened, if ever?).

  18. Luis says:

    (That last part should probably be “pure service industry labor” rather than “manual labor.”)

  19. Pithlord says:

    If there is one thing the problem of declining employment in existing industries due to increases in productivity is not, it is “new”. In fact, Cowen’s point is that there is less increase in productivity (and therefore less of this problem) then there used to be.

    As long as aggregrate demand keeps pace with improved capacity and ignoring transitional issues (a big couple assumptions, I grant you), then there will always be new jobs. Economics comes under a lot of sometimes-deserved ridicule, but we really do know that to be true – and it is obviously still counter-intuitive.

    One problem Tim is ignoring is that we get used to our improved entertainment. So the guy at 50K would feel poorer if he used the cheaper options for getting 1990-standard entertainment than he felt in 1990.

    Another point that I think is missed is that the rate of inflation will vary depending on the consumption bundle you use. Or rather, that’s not missed, but the implications aren’t pondered. Let’s imagine 2 people and 2 sectors (housing and entertainment). If Mr Brown’s consumption in 1990 is 80 percent housing and 20 percent entertainment, and Mr Pink”s is the opposite, and productivity triples in entertainment and stagnates in housing, and the central bank dilutes the currency by a factor of 2, Brown and Pink could double their nominal incomes, but Brown faces a much higher inflation rate.

  20. MNP says:

    rj, I understand what you’re saying perfectly. I don’t know why Lee doesn’t seem to.

    He’s getting more for the money, but because the actually amount of money going out is the same, he doesn’t actually have more liquidity to spend on things like rent/food/clothing/energy and these things are all increasing in price.

    Luis, people I know who work in IT, when they “create a program to automate X” like database sorting or whatever, refer it as making a “robot” so I meant AI as well. As for jobs there are a few things. I am sympathetic to Yglesias on this (though on little else) because I realized that I actually spend very little on entertainment. Good internet, a good game console and a good TV to watch it on are pretty much all I need to but to be entertained. Other than that I can just do stuff outside like play sports which is pretty cheap and then I can either do any of those activities with friends or just chat with them. Also cheap. I also have no desire to have a house and those responsibilities. But not everyone is like me, and some people want kids or something so it’s not only that more people will have to make do with less, for somethings less might not just be possible unless you want to really lower quality of life standards to stuff we see in developing countries. Better than the historical human norm? Yes, but not what you could call great.

    The thing is, even service jobs are being automated. Like that restaurant that is replacing some of its waiters with computer ordering. You still need people to take the food out to you but you need less of them now. It’s an expensive initial investment but as long as the machines work, it’ll save you money on wait staff in the long run.

  21. cryptozoologist says:

    i do not share tim’s optimism for automatic, driverless vehicles. while potentially saving fuel, time, and enhancing safety, the liabilities are staggering. i don’t know about your part of the country, but where i live we have lots of deer and other woodland creatures that often hurl themselves into the paths of oncoming vehicles. so imagine a family on a bucolic outing when a deer lunges in front of their car resulting in injuries and/or fatalities. the manufacturer of the car, the automatic driving system, the car dealer etc. would all find themselves defending against enormous law suits. even if one presumes that such systems might be successfully installed within a less litigious society than the usa, there is no compelling reason to believe that we would be motivated to pursue those benefits for ourselves. just look at single payer health care and high speed rail for two examples. allow me to also point out that simply improving the software of the above referenced automatic driving system is not at all ‘low hanging fruit.’ software engineering is an emerging discipline, but it is far from an exact science. one simply cannot predict if a piece of code that depends on complex streams of sensor data will do the ‘right’ thing under extreme conditions (see http://en.wikipedia.org/wiki/Halting_problem).

    at least until ‘the singularity’ arrives

  22. cryptozoologist: your concerns are well founded, and I think these kinds of difficulties will probably delay the introduction of driverless vehicles. But the benefits are so overwhelming that I think it’s almost inevitable that they’ll get adopted eventually. One likely route is that other countries that are less democratic and more robot friendly (think Japan, China, or Singapore) will adopt them first and demonstrate their benefits. Our irrational fear of death by machine is likely to be trumped by our irrational fear of being left behind by foreigners. 🙂

  23. john says:

    I understand rj’s point, but it boils down to an empirical one: Do people spend more or less on “entertainment” than in the past? My sense is less, at least among my age cohort. I don’t buy $100 worth of CDs per month anymore. I bet that I spend less on entertainment now than I did in the past, even though I make more money now. But people like me could be balanced out by the people who buy a lot of XBox games or something.

    For what it’s worth, Tim seems to claim that because of all of the missing industries we’re now spending less overall. I don’t think that’s true; money once spent on cameras and maps might now be spent on food. But shifting that money around isn’t necessarily an economic wash.

    Let’s assume that people are spending less or that they are spending the same on industries that employ fewer people. There should be excess cash somewhere. But just because someone, somewhere is saving money, that doesn’t automatically mean that any lost jobs are instantly re-created elsewhere in the economy. Just because there is surplus cash, doesn’t mean there’s anything worth spending it on, or worth investing in. It is entirely plausible that measured GDP could shrink (and unemployment rise) even while employed people’s standard of living (and “wealth”) keeps going up. I think this ties into Tyler’s basic point: That we’re just running out of new stuff to invest in. Industries that go away because they’re emphemeralized don’t get replaced.

    The motto of the RepRap 3D printing project: Wealth without money.

  24. langlang33 says:

    When I originally commented I clicked the -Notify me when new comments are added- checkbox and now each time a remark is added I get four emails using the same comment. Is there any means you possibly can take away me from that service? Thanks!

Leave a Reply

Your email address will not be published.