First, Somin is not clear about how to employ the factors of equity and efficiency. He writes, "as a legal matter, there is no doubt that this potential use of eminent domain is constitutional . . . However, the fact that the Army's plan is constitutional doesn't necessarily mean that it is equitable or efficient. " This implies that equity or efficiency or some combination of the two ought to be a factor in takings decisions.

This argument potentially has some merit; after all, it may make sense to force government actors to act efficiently or equitably or both, and it is easy to criticize actions that are neither. However, it is also clear that any such test would require elaboration. How _much_ should equity and/or efficiency weigh in the decision? Should only one be required, or both?

And such an approach is not without its own costs. For example, a strong rule could be articulated to include both of Somin's suggested factors, along the lines of, "no exercise of eminent domain is permitted unless the government can show that the action is both equitable and efficient." That would give weight to both of Somin's factors. It would also seriously limit government ability to act. That may be a consequence that Somin or other commenters view as positive, but it would probably not be universally viewed as a positive result.

So, an initial question relates to Somin's implied endorsement of an equity/efficiency test of some sort. Just what kind of implied test is Somin proposing, and how does it work? (As we'll see, this matters for the analysis.)

(As an aside, Somin seems rather cavalier about the analysis of efficiency in this case. In particular, he hints that this taking may be inefficient and that other property may be equally amenable to this particular use. Yet, his blog post (which cites other portions of the WaPo piece) does not mention the multiple statements in the WaPo article about the proximity of _this_ land to a major military base that is one of the largest training hubs in the Western U.S. Even if that is not dispositive, it seems like relevant information in any calculus of efficiency.)

A second critique relates to Somin's apparent articulation of a rule favoring use of existing government property.

There remains the question of whether a facility of comparable quality could be built without resorting to condemnation. The U.S. government already owns hundreds of millions of acres of desert property in the Western states, much of which is not being used. Perhaps the Pentagon could build a new training facility on land the federal government already owns; if so, that would be far preferable to displacing private property owners.

Somin does not explicitly set out a reason why this would be preferable. His statement suggests an interesting potential rule: If the government already owns land which could be put to a particular use -- an acceptable substitute -- then it should not condemn private land for that use.

Again, that seems inuitively sensible -- why take private land for a use, if an acceptable substitute in public land is available? But again, it seems the devil may be in the details. For one thing, this rule seems to invite disagreement about the definition of acceptable substitute.

For instance, suppose that the government wants to build a road from Point A to Point B; that private land lies directly between A and B; and that a circuitous, snaking chain of government land could also be used to connect A to B. Should the government be required to use government land and the indirect route rather than taking private land for the direct route? Would this apply even if the result was a winding and difficult (possibly dangerous?) road, or the creation of extra miles of difficult driving? If it creates extra road-building expense? Is it really an acceptable substitute to build on government land if that means a road from Times Square to Rockefeller Center, via Queens?

The difficulty of deciding on acceptable substitutes ties in to a broader point: A general rule of using government instead of private land every time could itself be highly inefficient. Should the city really be proscribed from condemning my $100 acre of rural property to build its power plant, and instead forced to build that plant on a high-value ($2000) city-owned acre downtown? We're where the rubber hits the road (and back to question 1, really) -- what matters more here, equity or efficiency? If we truly force a rule of "use public land first" in all cases, aren't we giving a kind of subsidy (failing to take) to certain property owners, at the expense of the general public?

And to apply that question to this case: If it would cost $10 million to take the privately owned Colorado land, and $20 million for a next-best option of building an equivalent infrastructure on available public land in Nevada -- should we really apply a strict rule of "use public land first" and simply pass the extra cost on to taxpayers? (What if the next-best option costs $40 million? $100 million?)

I'm really unconvinced that Somin's use-government-land-first approach makes sense as a broad rule. It has obvious flaws from an efficiency standpoint; and depending on the size of the subsidy, it may not make sense from an equity standpoint, either.

Finally, I was surprised by Somin's analysis of subjective values.

Of course, property owners often attach subjective value to their property, and thus the value to an owner is often greater than market value. That's why the property is still in the hands of _these_ owners, after all. Courts (including the Supreme Court) and commenters have discussed this matter, and it has come up in cases like Lutheran Synod. And in general, courts have _rejected_ (on workability grounds, among others) the idea of paying subjective values, instead keeping to the rule of fair market value. (The general rule is subject to exceptions, such as where the property value is unascertainable or where paying market value would result in manifest injustice.)

Is Somin saying that all property owners should be paid subjective value? He initially seems to be headed that direction, suggesting that these property owners deserve over-market compensation, and writing that "market value compensation often fails to fully replace the owners' losses. If they valued the land at the market price or less, they presumably would have sold it already; their decision to hold onto it is an implicit signal that they place a "subjective value" on the property above its market price."

If Somin is really arguing that compensation should generally be subjective, he's got his work cut out for him. That would be a major change in takings law, and would affect vast numbers of takings, since many property owners attach some subjective value to their property.

However, Somin then scales it back: "In this case, subjective value concerns are particularly serious. Many of the owners' families have lived on the land for generations, and would lose most of their livelihood if forced to move. Even if the Court is right to hold that fair market value compensation is all the Constitution requires, this is one case where the feds should pay more."

Okay, so only _these_ owners deserve extra compensation. Still, it's not completely clear why. Perhaps Somin is implicitly making a replacement-value argument by noting the loss of livelihood -- that is, an argument that fair market value would leave these owners unable to replace the property, and that they should be paid replacement value instead. Unfortunately, that approach would go directly against Supreme Court statements that replacement value is not the legal standard. Or perhaps Somin is making a more limited argument, that market value here would result in manifest injustice. That's probably a tough one to convince a court. (Or perhaps an argument that there really is no adequate market for this particular property?)

Somin may be right in his conclusions that this particular taking is a bad idea, or that extra-market compensation should indeed be paid. But I don't find his current analysis, as set out in his post, to be convincing support for those conclusions.