in which I put on my management consulting hat

The concept of agency cost is a useful idea of economic theory which points to a real phenomenon and problem: that of aligning the interests of agents with those of their principals. Roughly, the problem goes like this: agents or employees will (often enough and until people are as naturally ethical as they ought to be) be happy to perform poorly at their job as long as they will continue to be paid; the distinctive problem, then, is to somehow manage the agent to ensure that they do their job well. But any such management comes at a cost. It always costs something to implement review or incentive structures which ensure that agents will be motivated to act always and only in the principal's best interest. Hence, the basic problem of agency cost.

Still, some modes of agent supervision and incentivization will be more cost-effective than others. Hiring a full-time inspector to constantly watch and inspect the work of each employee might be extremely effective in ensuring that employees don't slack off, but it is usually too cost-prohibitive to be a viable option. Encouraging employees to work in teams may be a more cost-effective model, with the thought that there will tend to be group policing against free-riding slackers. (But if the entire group goes bad, then there is another set of problems.) Performance bonuses are yet another popular way of ensuring appropriate high effort.

Performance bonuses can cause their own separate problems, however. For example, bonuses paid out at the end of a year for work performed during that year encourage short-term thinking in employees who are motivated to do the flashiest projects that will get the most management attention for the greatest bonus. Of course, this isn't making any news: arguably, the present financial crisis was in part caused by agents who had unfortunately short-term views. From the point of view of someone who stands to receive a hefty bonus for getting deals done, the natural incentive is to get lots of deals done and to do so sooner rather than later even if the quality of the deal is sub-par; if the deal goes south in 5, 10, or 20 years, there will typically be no clawback of the employee's bonus.

If I am right that there is in some cases or industries this phenomenon of performance bonuses causing these types of distortions in agent performance, then I have two ideas for fixing the problem. Those ideas are these:

1. Make bonuses subject to progressive clawbacks contingent on the failure of the employee's work-product. So, to stick with the financial example, if the deal goes bad within 1 year, 90% of the bonus is returned; if in 5 years, 10% is returned. (Made up numbers, but you get the idea. The clawback percentages could be calibrated to create the optimum amount of incentive to act foresightfully.)
   
2. Provide some portion of the performance bonus in a deeply retrospective manner. E.g.: after five years, your deals over that period will be reviewed and you will receive some bonus depending on the success of your work over time. Incidentally, deeply retrospective performance bonuses will encourage long-term commitments from employees to a particular principal, which will further enhance the principal's position.

The examples given are loosely from the financial industry (since that industry has caused so much economic wreckage from so much lack of foresight), but I think the principles might work in other spheres as well. Of course, this idea has its costs -- reviews at the space of five years are not costless, and success-criteria would have to be defined in advance -- but it may be a cost worth paying for the incentive to longer-term thinking which it provides.

the killing of al-Awlaki

So, a drone strike has killed Anwar Al-Awlaki, an American citizen, al-Qaeda member and radical Islamic cleric who had been hiding in Yemen. The strike was a long time coming; Al-Awlaki has been on the chief hit-list for over a year, at least. This event has led Glen Greenwald to bemoan the "new" power of the American president to "murder" American citizens by targeting them for assassination without due process.

To call the power in question a "new" power is somewhat polemical. Does anyone doubt that, at least in a classical battlefield situation, the Commander-in-Chief has always had the power to target enemy commanders (or soldiers), whatever their citizenry? You throw in with the enemy, you take your risks. So then the point is, I think, either that:

a) the war against Al-Qaeda and its members is, in a relevant moral sense, different from the aforementioned classical battlefield situation- (perhaps we say -- it is not "really" a war at all, but I've never been convinced by the arguments here, and, anyway, I don't think the point is really to wrangle over the legal definition of "battlefield") (the point must be to find and then draw a distinction based on the relevant moral sense of things) or

b) even if the war against Al-Qaeda and its member is not in any morally relevant sense distinct from other sorts of wars, still there ought in any case to be more or different process when determining a hit against an American citizen, even if he is, say, an enemy commander beyond the reach of traditional legal process- (but then let's have a concrete alternative proposal)-

I don't think a can be maintained on reflection (although not everyone agrees, obviously), and I'm sympathetic to b in some moods, but I have yet to hear a workable alternative concrete proposal. The question would be: what process would we trust more than the current one (which basically amounts to ex post political review -- the President can be punished by the electorate in one way or another if they think he oversteps). And what process would be appropriately sensitive to the operational demands of a dynamic battlefield situation? What do we do? Top-secret trials in abstentia? That indeed might be an improvement-- I wouldn't necessarily oppose it-- but even that has its downsides. There would still need to be secrecy in the case of top-secret evidence, and the defendant would still not be "confronted" by the evidence against him. From the (dogmatically) civil libertarian viewpoint, there would be much still to oppose. From the military perspective, the process could be unduly cumbersome (remember the old stories about letting bin Laden get away shortly after 9/11 because some functionary lawyer advised against a targeted strike?- even if apocryphal, the story has a moral).

But then what? Do we really think that there should be a de facto per se ban against deliberately killing any enemy combatant who has American citizenship and is effectively beyond the reach of American law? I don't especially, and I doubt most others do on serious reflection. I'm not (as Greenwald says) "cheerleading" here. But I'm frustrated with civil libertarians who are long on hand-wringing and short on any constructive alternative proposals, and who anyway tend to give no indication of appreciating the real, practical difficulties.

(Maybe Al-Awlaki was indeed no very serious threat; but, then again, maybe he was, and the evidence which shows it could not safely be made public. The assumption to make when trying to develop policy is the latter, because that is the assumption which brings out the real difficulties. (If you want a recipe for myopic policy, it would be to ignore real difficulties and practical problems as a rule.) In this vein, I agree that the question is not about Al-Awlaki particularly, but is bigger-- it is about the sort of policy we ought to have in a certain sort of desperate case.)

puzzles of patent policy

Read this article in the New Yorker (from 1993) for a still-relevant essay about the fundamental problems of contemporary patent policy. This article was the inspiration for a movie called "Flash of Genius" (not especially well reviewed). I haven't seen the movie, but from the trailer and reviews, I get the sense that the fimmakers missed some of the point of the above New Yorker article. The film is presented as an underdog story: heroic inventor against big, impersonal corporation which committed outright theft of inventor's efforts. The problem of patent policy, however--and a problem fairly highlighted in the article--is considerably harder to face squarely: how to make appropriate sense of the idea of an "inventor" at all, especially where nearly all inventors and inventions would be practically impossible except in some wider social context. The current American standard judges patentability in relation to what is "obvious to a person reasonably skilled in the art," but this standard, it seems, manages to give less guidance than good legal standards ought. But what to do then? As a recent episode of This American Life makes clear, the stakes are high; bad patent policy means restricting innovation by charging tolls on innovative companies, at the potential cost of driving some of those companies out of business. (And there is Google's $12.5 billion acquisition of Motorola for, in large part, its patents. That's billions of dollars spent, essentially, on a litigation strategy to protect itself from patent infringement lawsuits. That's billions not spent developing new products and technologies-- investment that would have fueled real economic growth and jobs.)