My Twitter feed is alive with the sound of indignation about an ad agency at South by Southwest that is using the homeless as human 4G wireless hotspots. The idea is that you see a homeless person with a t-shirt reading “I am a 4G hotspot,” and then you pay them a small fee to get online. There is definitely something unsettling about this, but there is also something a bit off about a lot of the reactions I’ve been seeing to it. I’ll get back to that in a bit, after a detour through a familiar theme.

The blog post announcing the initiative is full of gobbledygook about “charitable innovation,” and it’s very unclear about whether this project is supposed to be a profit-making business venture, a charity project, or some utopian neoliberal combination of both. Whatever it is, there’s something undeniably creepy about it, in the way it turns people into infrastructure — e.g. “I am a hotspot” rather than “I’m running a hotspot.”

It’s also, naturally, an opportunity for people to project their anxieties about the desirability of capitalist “innovation.” But the homeless-as-hotspots plan highlights a point I’ve been trying to make about technology. Technical change comes in two forms, one that is designed to more intensely exploit labor, and one that is designed to replace labor. Which one will dominate depends, in large part, on the condition of labor itself.

In a recent post, I framed this in largely Marxist terms. But I want to return, for a third time, to the mainstream economics version of the same argument, which I still think hasn’t gotten enough attention. The work I draw on is a paper by Daron Acemoglu, “When Does Labor Scarcity Encourage Innovation?” It was published in the Journal of Political Economy, and an ungated version is here. It includes lots of mathematical formalisms, which I’ll admit I only barely followed; my years of taking calculus are well in the rear-view mirror. But the core of the argument is easily understandable.

Acemoglu is attempting to reconcile two different stories about the interaction between labor supply and technological change. Standard economic growth models posit that when labor is scarce and wages are high, adoption of new technology will be discouraged and growth will slow. But there are a variety of arguments to the contrary. Robert Allen has argued that the industrial revolution took off in eighteenth-century Britain because of the high price of labor there relative to other parts of Europe, which encouraged the invention and use of technologies that substituted machinery and energy for labor. The “Habakkuk Hypothesis,” meanwhile, claims that the US grew faster than Britain in the nineteenth century because labor was scarce (and therefore more expensive) in the US, which in turn encouraged mechanization and other labor-saving technology.

The contribution of Acemoglu’s paper is to contextualize these arguments in a general framework in which there are two kinds of technology: labor saving and labor complementary. In economic jargon, a labor saving technology decreases the marginal product of labor, while a labor complementary technology increases it. This means that with labor saving tech, businesses need to use less workers, while with labor complementary tech, they need more workers. This then leads to Acemoglu’s conclusion about the reverse causal process: what is the effect of labor scarcity (and high wages) on the adoption of technology? From the paper:

The main result of the paper shows that labor scarcity induces technological advances if technology is strongly labor saving, meaning that technological advances reduce the marginal product of labor. In contrast, labor scarcity discourages technological advances if technology is strongly labor complementary, meaning that technological advances increase the marginal product of labor. I also show that, under some further conditions, an increase in wage levels above the competitive equilibrium has effects similar to labor scarcity.

So here’s a riddle: which form of technology should we prefer, labor saving or labor complementary? Labor saving technology is consistent with high wages and tight labor markets. But it also, of course, leads to less jobs overall in the sectors where it is deployed. Which brings us back to the homeless people with hotspots. Let’s imagine, for the sake of argument, that this is a legitimately profit-making business venture rather than a weird kind of charity. (And note that even as charity, the project depends on its consumers viewing it as a kind of legitimate business, a way for the homeless to engage in “productive” labor.) Putting hotspots on homeless people has to count as a labor complementary technology. From the standpoint of the wireless company, the marginal product of a homeless person’s labor is much higher (i.e., it’s non-zero) once you’ve figured out that you can attach hotspots to them. So if you think that it’s bad when machines replace human labor (which is not what I think), then this is just the kind of technical change you should prefer.

But labor complementary technology doesn’t necessarily look so great once you’re face-to-face with the kind of labor it complements. In this case, it relies upon the existence of a cheap and exploitable labor force — something that’s obvious when you’re looking at a homeless person in a creepy t-shirt, less so when you order from an online retailer. And here’s where I think a lot of the outrage over homeless-people-as-infrastructure goes wrong.

I don’t recall seeing a lot of complaints about the problem of homelessness in Austin prior to this story. Which I don’t mean as some kind of “gotcha” — the world is full of horrible things, and it’s neither possible nor particularly helpful to try to talk about all of them all of the time. But to get up in arms about an ad agency exploiting the homeless as wifi routers strikes me as a peculiarly half-assed form of outrage. If they weren’t walking around as billboards for wireless service, Austin’s homeless and poor would still be homeless and perhaps a bit more poor. The fundamental problem here is not exploitation, but the condition of possibility for that exploitation, which is the fact that there are so many poor and homeless Americans in the first place.

“The misery of being exploited by capitalists is nothing compared to the misery of not being exploited at all,” goes the old adage from Joan Robinson. Then again, says Marx, “to be a productive laborer is not a piece of luck, but a misfortune.” In the short run, labor complementary technology may employ more people, which is better than them not being exploited at all. But in the long run, the jobs thus created tend to be terrible, and our real goal ought to be to channel technical change toward labor saving innovation.

This leaves us with the question of what the homeless of Austin can demand, if not the right to be walking 4G hotspots. Fortunately there is a simple solution to that. There’s nothing (economically) stopping us from just giving people cash; and as the housing activist Max Rameau likes to say, the cause of homelessness is that people don’t have homes, and we have plenty of those. So imagine what would happen if this pool of cheap, easily exploitable labor wasn’t available. A company that wanted to sell 4G wireless services might have to invest in more transmitters to fulfill demand. Or perhaps they would deploy robots to roll around the streets selling wireless access! This would not employ as many people, since it’s more a labor saving than a labor complementary technology. But it also wouldn’t create the grotesque spectacle of fellow human beings serving as pieces of infrastructure.