I couldn't write a complete overview of drones even if I wanted to. But the first two chapters often get shortchanged in the current media feeding frenzy over a leaked drone memo.
Chapter 1. Finding and exploiting an opponent's weakness. On the surface, Chapter 1 has little to do with drones, but everything to do with the broader context in which they get deployed. The "war on terror," as Bush 43 put it, better framed by Obama as the mission to "disrupt, dismantle and defeat al Qaeda," relies on two sets of rules: the rules of law enforcement and the rules of warfare. Critics question whether the drone program properly follows those sets of rules. That's a fair question, and it's good that the press have gone beyond blandly reporting drone strikes as if they were a common weather pattern. We are great at law enforcement, and we are great at warfare; it stands to reason we should set high standards for both. But while our gumshoes and GIs have no peer, neither set of rules is well-suited to disrupting, dismantling and defeating al Qaeda, a mission whose merit no one questions.
The rules of law enforcement assume that all countries have the means and the will to enforce laws, and that the focus should be on building cases, not preventing crime.
The rules of warfare assume uniformed soldiers sponsored by countries. Deadly force gets employed based on split-second decisions by all, even the lowliest private.
Al Qaeda's asymmetric tactics fit neither set of rules. That is not a coincidence.
A common military strategy is to attack along the seam between two opposing forces, since coordination difficulties will ensure that that is the weakest point along a war front. In seeking to disrupt, dismantle and defeat al Qaeda, we employ two strategies, a military one and a law enforcement one. The weakest point is the seam between the two.
Indeed that seam is the weakness al Qaeda seeks to exploit. They don't wear a uniform, and they're not sponsored by any state. That would suggest they are common criminals, where the rules of law enforcement apply. But they gravitate toward countries where the rule of law is tenuous, at best. In that seam between the two sets of rules, we inherit the disadvantages of both and the advantages of neither.
Chapter 2 - Drone capabilities. Into this foggy milieu enter drones, unmanned planes carrying cameras and missiles, controlled by U.S. pilots in Nevada or another remote location. Before gnashing teeth over who is a legitimate target, when, under what conditions, and most of all, who gets to decide, one needs to read Chapter 2, which covers the four game-changing capabilities drones bring to the mission. These capabilities have profound implications for developing new rules that address the threats that have emerged out of the subtle weaknesses of the current two-pronged (law enforcement and warfare) regime.
Section 2.1: Drones are unmanned. Well, duh. But this has huge implications for rules of engagement. The critical problem with deploying troops to trouble spots is that they have to be given the right to defend themselves. If fired upon, fire back! Self-defense, however, means that people can get hurt, and it's not always the bad guys. Innocent civilians and your own troops can get killed. With drones, self-defense is irrelevant.
Section 2.2: Surveillance. Lost in the debate about deadly force delivered via drone is the fact that the most powerful "weapon" carried by a drone is a camera. In fact, most drones, which range in size from hand-held to a take-off weight exceeding a half dozen cars, are unarmed. The most famous drone strike, which targeted and killed Anwar al-Awlaki, was likely preceded by hundreds if not thousands of hours of drone surveillance. Regardless of the preferences of hawks as well as doves, we can expect the typical drone strike approval to require more intelligence.
Section 2.3: Time for deliberation. By providing staggering amounts of data over prolonged time periods, continuous drone monitoring imposes an obligation, moral if not legal, for extended analysis and deliberation. In World War II General Eisenhower would have rightly objected to having every shoot/don't shoot decision approved by an "informed, high-level official." No such excuse exists for drones, where the decision to strike can play out over hours if not days and involve gigabytes of data. Ultimately, if the right decision-making process involves approval from a dozen officials, digital drone technology supports it.
Section 2.4: Small, precise munitions. Conventional wisdom holds that the U.S. military seeks to build weapons that make bigger and bigger explosions. In reality, for the most part they are obsessed with building smaller weapons wielding less destructive power that limit collateral damage. This won't comfort people who follow in the footsteps of Anwar al-Awlaki, who everyone agrees got what he deserved. But it reduces collateral damage. Small, precise munitions make it harder to justify collateral damage as a necessary side-effect of hitting a legitimate target.
Given the two discreet sets of rules we can bring to bear, disrupting, dismantling and defeating al Qaeda is a difficult challenge. The unique capabilities of drones will help us meld the rules of law enforcement and the rules of warfare into a new hybrid set of rules that eliminates the seam al Qaeda is currently exploiting.
Update: Last year an article in the New York Times discussed some advantages of drones.
Update #2: Mark Bowden has a thoughtful article in the Atlantic.
Friday, February 8, 2013
Monday, February 4, 2013
How A/B web testing can unintentionally propagate stereotypes, including racism
Why does Google's ad network think people with names like Latanya and Rasheed have an arrest record?
Imagine the following scenario: An employer is Googling prospective job applicants. Some of those applicants have black-identified names. Due to his or her personal racism, the employer happens to be more likely to click on the InstantCheckmate ads that suggest “Arrested?” next to the black-identified names. And over time, Google’s AdSense algorithm learns that “ads suggestive of an arrest record” work better when associated with black-identified names. Voila.
Sunday, January 27, 2013
Robonomics
If, later this century, we had robots so intelligent and so dexterous that they can perform any and every job a human can do, how would that affect the economy?
This thought experiment, which I'll name "WALL-E" in honor of the animated movie about a future civilization that comes close, was prompted by Ray Kurzweil’s book How to Create a Mind. (Coincidentally, the book includes many other interesting thought experiments.) For centuries the economy has engaged in "creative destruction," where technological advancement destroys jobs for humans - buggy whip and slide-rule assemblers, for example - while it creates others: auto mechanic, computer programmer, etc. Robots won't suddenly become the intellectual equals of humans, as the thought experiment implies. As we humans continually improve them, we will busily find valuable things to do that robots can't, at least yet. Maybe we will keep doing that forever. And even if robots were every bit as capable as humans, I can't imagine electing one to be president, or wanting one to be a hospice director. Perhaps our ultimate competitive advantage over robots is our mortality - although, when we figure out how neurons can directly communicate with computers, it may get difficult to tell human from robot.
Nevertheless, the WALL-E question stands. Where's the "creative" in creative destruction if each new job technological advancement spawns can be done better by a robot, at lower cost? Would we humans get jobs building robots? Remember the premise: any - any - job a human can do, a robot can do better. So robots, not humans, would build robots.
Accepting the premise at face value, it's an understatement to say the concept of human employment would be fundamentally transformed.
A lot would depend on who owns the robots. Let's say for argument that there were as many human-capable robots as humans, ten billion of each. Let's assume that due to technological advancement these incredible robots, robots so advanced they can do anything a human can do, only cost $5,000 per year to run.
At one extreme, all ten billion human-capable robots could be owned by me. I could "hire out" the robots at say $20,000 per year, making a annual profit of $150 trillion dollars. Meanwhile, Adam Smith's invisible hand would drive human salaries down to $20,000 per year. If there's demand for more than ten billion workers, remember I can have my robots build more robots. So while it's possible for humans to find employment, Adam Smith ensures they can't find jobs paying more than what I charge for my robots.
At the other extreme, each human in this scenario could own one robot. That would still create downward pressure on human wages, but the profit from robots would be spread far and wide. Current ownership of robots, limited though the robots are, is concentrated in large corporations, and we can expect that pattern to continue.
Capital and human labor are the inputs into all goods and services. In my WALL-E thought experiment, perhaps some people would find jobs paying $20,000 per year, but for all intents and purposes, human employment has been wiped out. Goods and services are produced entirely from capital, and income is composed entirely of investment income.
So, in WALL-E world income inequality very likely would be higher than today, as human wages fall and robot "wages" accrue to a concentrated few.
How then, could anyone (other than me, of course) afford to buy anything? It wouldn't even be in my self-interest to own the entire robot labor force, because no one could buy the stuff my robots produced. (As one example of this consideration, Henry Ford deliberately paid his workers above-market wages so they could afford the cars he sold, allowing him to sell more cars.) The public policy lesson is that concentrated robot ownership could lead to higher levels of income inequality, and at high enough levels, income inequality creates a drag on the economy, hurting everyone. Like I said, a lot would would depend on who owns the robots.
Another consideration is the relationship between productivity and wages. For much of the twentieth century, productivity increases, measured by Gross Domestic Product (GDP) per person, have lead to wage increases. When each worker can produce more, companies, in the aggregate, have sought more workers, not less. The expanding economy spawns a new, more productive job for every job destroyed; companies discover they must offer higher wages in order to fill those new jobs, even if the job is no more difficult and requires no more skill. Other companies then have to raise wages in order to retain their workers. In an effect identified by Baumol, an economist, this can even lead to wage increases for jobs (hair stylists, perhaps) where productivity has not increased.
But in WALL-E world, productivity increases push wages down, not up. If technological advances (developed by robots, of course) allow my robots to produce twice as much at the same cost, such that it takes two humans to replace one of my robots, then that will force human wages down to $10K per year. And with robots in WALL-E world designing new, ever more capable robots, productivity growth would skyrocket.
I doubt we will ever be presented with a scenario as extreme as my WALL-E thought experiment, at least not in the twenty-first century. It's tempting to conclude that since the scenario won't occur, we won't have to contend with the issues WALL-E world raises. But these issues aren't binary. If a fraction of the scenario occurs, we will experience a fraction of the consequences. Hyperbolic extremes can illuminate subtle issues that exist in more realistic scenarios, but are hard to see. Although technology will fall short of the WALL-E thought experiment, it will continue to advance, and advance quickly. Thus, to some limited degree we will experience the effects discussed above.
In fact, income inequality has been increasing, a trend Tim Noah masterfully covered in a series of articles titled The Great Divergence. This could either be a passing fad, with income inequality dropping again soon, or the leading edge of a sustained climb, which if it occurs, would become the defining public policy issue of the twenty-first century.
And, in fact, we've also recently seen productivity growth coupled with stagnant wages:
Perhaps we will soon invent whole new job categories in an expanded economy, and the stagnant median wages we've recently experienced will turn out to be a temporary phenomenon in the long arc of history. But the WALL-E thought experiment suggests that with sufficiently advanced technology, steadily increasing productivity will force wages down. Could wages be on a downward trajectory from now on? In The Lights in the Tunnel, author and Silicon Valley entrepreneur Martin Ford says yes:
This thought experiment, which I'll name "WALL-E" in honor of the animated movie about a future civilization that comes close, was prompted by Ray Kurzweil’s book How to Create a Mind. (Coincidentally, the book includes many other interesting thought experiments.) For centuries the economy has engaged in "creative destruction," where technological advancement destroys jobs for humans - buggy whip and slide-rule assemblers, for example - while it creates others: auto mechanic, computer programmer, etc. Robots won't suddenly become the intellectual equals of humans, as the thought experiment implies. As we humans continually improve them, we will busily find valuable things to do that robots can't, at least yet. Maybe we will keep doing that forever. And even if robots were every bit as capable as humans, I can't imagine electing one to be president, or wanting one to be a hospice director. Perhaps our ultimate competitive advantage over robots is our mortality - although, when we figure out how neurons can directly communicate with computers, it may get difficult to tell human from robot.
Nevertheless, the WALL-E question stands. Where's the "creative" in creative destruction if each new job technological advancement spawns can be done better by a robot, at lower cost? Would we humans get jobs building robots? Remember the premise: any - any - job a human can do, a robot can do better. So robots, not humans, would build robots.
Accepting the premise at face value, it's an understatement to say the concept of human employment would be fundamentally transformed.
A lot would depend on who owns the robots. Let's say for argument that there were as many human-capable robots as humans, ten billion of each. Let's assume that due to technological advancement these incredible robots, robots so advanced they can do anything a human can do, only cost $5,000 per year to run.
At one extreme, all ten billion human-capable robots could be owned by me. I could "hire out" the robots at say $20,000 per year, making a annual profit of $150 trillion dollars. Meanwhile, Adam Smith's invisible hand would drive human salaries down to $20,000 per year. If there's demand for more than ten billion workers, remember I can have my robots build more robots. So while it's possible for humans to find employment, Adam Smith ensures they can't find jobs paying more than what I charge for my robots.
At the other extreme, each human in this scenario could own one robot. That would still create downward pressure on human wages, but the profit from robots would be spread far and wide. Current ownership of robots, limited though the robots are, is concentrated in large corporations, and we can expect that pattern to continue.
Capital and human labor are the inputs into all goods and services. In my WALL-E thought experiment, perhaps some people would find jobs paying $20,000 per year, but for all intents and purposes, human employment has been wiped out. Goods and services are produced entirely from capital, and income is composed entirely of investment income.
So, in WALL-E world income inequality very likely would be higher than today, as human wages fall and robot "wages" accrue to a concentrated few.
How then, could anyone (other than me, of course) afford to buy anything? It wouldn't even be in my self-interest to own the entire robot labor force, because no one could buy the stuff my robots produced. (As one example of this consideration, Henry Ford deliberately paid his workers above-market wages so they could afford the cars he sold, allowing him to sell more cars.) The public policy lesson is that concentrated robot ownership could lead to higher levels of income inequality, and at high enough levels, income inequality creates a drag on the economy, hurting everyone. Like I said, a lot would would depend on who owns the robots.
Another consideration is the relationship between productivity and wages. For much of the twentieth century, productivity increases, measured by Gross Domestic Product (GDP) per person, have lead to wage increases. When each worker can produce more, companies, in the aggregate, have sought more workers, not less. The expanding economy spawns a new, more productive job for every job destroyed; companies discover they must offer higher wages in order to fill those new jobs, even if the job is no more difficult and requires no more skill. Other companies then have to raise wages in order to retain their workers. In an effect identified by Baumol, an economist, this can even lead to wage increases for jobs (hair stylists, perhaps) where productivity has not increased.
But in WALL-E world, productivity increases push wages down, not up. If technological advances (developed by robots, of course) allow my robots to produce twice as much at the same cost, such that it takes two humans to replace one of my robots, then that will force human wages down to $10K per year. And with robots in WALL-E world designing new, ever more capable robots, productivity growth would skyrocket.
I doubt we will ever be presented with a scenario as extreme as my WALL-E thought experiment, at least not in the twenty-first century. It's tempting to conclude that since the scenario won't occur, we won't have to contend with the issues WALL-E world raises. But these issues aren't binary. If a fraction of the scenario occurs, we will experience a fraction of the consequences. Hyperbolic extremes can illuminate subtle issues that exist in more realistic scenarios, but are hard to see. Although technology will fall short of the WALL-E thought experiment, it will continue to advance, and advance quickly. Thus, to some limited degree we will experience the effects discussed above.
In fact, income inequality has been increasing, a trend Tim Noah masterfully covered in a series of articles titled The Great Divergence. This could either be a passing fad, with income inequality dropping again soon, or the leading edge of a sustained climb, which if it occurs, would become the defining public policy issue of the twenty-first century.
And, in fact, we've also recently seen productivity growth coupled with stagnant wages:
The economy’s failure to ensure that typical workers benefit from growth is evident in the widening gap between productivity and median wages. In the first few decades after World War II, productivity and median wages grew in tandem. But between 1979 and 2011, productivity—the ability to produce more goods and services per hour worked—grew 69.2 percent, while median hourly compensation (wages and benefits) grew just 7.0 percent.(I suspect average wages have done better than median wages, as wages for high income earners have probably grown.)
Perhaps we will soon invent whole new job categories in an expanded economy, and the stagnant median wages we've recently experienced will turn out to be a temporary phenomenon in the long arc of history. But the WALL-E thought experiment suggests that with sufficiently advanced technology, steadily increasing productivity will force wages down. Could wages be on a downward trajectory from now on? In The Lights in the Tunnel, author and Silicon Valley entrepreneur Martin Ford says yes:
A simple application of common sense should show us that there is some threshold beyond which the overall economy will become too capital intensive. Once this happens, lower prices resulting from improved technology will no longer result in increased employment. Beyond this threshold or tipping point, the industries that make up our economy will no longer be forced to hire enough new workers to make up for the job losses resulting from automation; they will instead be able to meet any increase in demand primarily by investing in more technology. ...
What might we expect to happen if the overall economy were approaching this tipping point, beyond which industries would no longer be labor intensive enough to absorb workers who lost their jobs to automation? We would probably expect to see gradually rising unemployment, stagnating wages and significant increases in productivity (output per hour of labor) as industries were able to produce more goods and services with fewer workers. That sounds uncomfortably close to what actually occurred in the years leading up to the current recession.Even if technological advancement falls far short of WALL-E, it will greatly improve our future, just as it has improved our lives in the past. But our economy may already be experiencing side effects from accelerated productivity growth: greater income inequality and stagnant median wages. If economic growth doesn't keep pace with productivity growth, jobs and wages will decline, a serious problem at least until WALL-E world makes human labor obsolete. With the technological frontier moving as fast as it is, we need to either create new jobs at a much faster pace, or transition to an economy where human labor doesn't play the critical role it does today. (People might still work, but it would be more of a choice, influenced in part by factors other than income - labors of love.) Either way, we will need more than the bromide of education reform; we will need an entrepreneurial spirit our forefathers could never imagine.
Friday, January 25, 2013
Working poor have the highest marginal tax rates
Edward J. McCaffery, USC professor:
[The] highest marginal tax rates in America do not fall on the highest incomes, like [Phil Mickelson, a professional golfer], but on certain of the working poor, many of them single parents, who are being taxed at rates approaching 90% as they lose benefits attempting to better themselves.
Thursday, January 24, 2013
General Martin Dempsey meets the new Army
General Martin Dempsey, later to become Chairman of the Joint Chiefs of Staff, meets the new U.S. Army:
Dempsey took command of the Army's 1st Armored Division in June 2003, when Iraqi insurgents were starting to target American troops with sniper fire, grenades and roadside bombs. As he prepared for a trip outside his headquarters, he took a moment to introduce himself to the crew of his Humvee. "I slapped the turret gunner on the leg and I said, 'Who are you?' And she leaned down and said, I'm Amanda.' And I said, 'Ah, OK,' " Dempsey told reporters at the Pentagon.
Thursday, January 17, 2013
Whoever came up with the idea the earth is 6,000 years old, give him or her a break
Frankly I'm impressed that a writer living a couple of thousand-ish years ago, centuries before the scientific method, swagged the earth's age and was accurate to within six orders of magnitude.
Wednesday, January 9, 2013
The strength of weak ties
The above could pass for a Jackson Pollock painting, but it's actually a map of my current network, courtesy of LinkedIn Labs' InMap tool. (Update, January 9th, 2013: the tool appears to be down at the moment.)
Social network analysis took a huge leap forward with Mark Granovetter's 1973 paper, The Strength of Weak Ties. The principles Granovetter described have both helped me and allowed me to help others. The punchline: your "weak" connections might be the most valuable ones.
Subscribe to:
Posts (Atom)