Lee put this question of servitudes versus plain vanilla contracts very well when she asked me to “distinguish more sharply between two kinds of concerns about servitudes: problems that are uniquely presented by servitudes ‘running’ to other users (who may be remote in time and space), and public policy concerns that exist independent of these concerns about running.”
Taking my three categories of concerns (notice and information costs, the problem of the future, and harmful externalities) as a starting point, my short answer is that these problems can and do plague even restrictions that bind only a single generation of users. A fully dickered bilateral agreement can be misunderstood, it can become obsolete, and it can harm third parties.
The point I emphasize in the article is that servitude-like characteristics exacerbate and magnify these problems. For example, the fact that chattel servitude terms run with the transfer of possession of a physical object can cause special confusion because people are conditioned to associate possession with certain rights of use and transfer, and because the high salience of immediate possession is likely to swamp accompanying terms. Moreover, servitudes that reach through to subsequent possessors have the potential to affect more people, over a greater span of time, than bilateral contracts--thus multiplying information cost (and other) problems.
And, as Stew notes, the remoteness between the parties also makes clarification or negotiation of problematic terms unlikely. In the case of licenses offered by an identifiable and well-known entity like Microsoft it is at least conceivable that one could identify and communicate with a person at the company regarding the terms of its license agreements and possible alternatives. But many works distributed under other licenses are anonymous--with the claimed restrictions to be enforced by unknown licensors--or, to add even further complication, by thousands of unknown and unconnected authors who have contributed to a work that has been serially modified.
Note that the exacerbating factor is not the “running” of the servitude per se, but the remoteness, durability, scope, etc., that are correlated with running. Part of my goal in the article is to identify these characteristics so that we can recognize them even when they are caused by something other than “running” (perhaps, as Lee suggests, by "hard-wiring," a possibility to which I will return in a separate post).
The fact that the difference between servitude-like and non-servitude-like restrictions is one of degree does not mean that it shouldn’t have consequences in terms of the enforceability of the restrictions (or the desirability of mechanisms other than non-enforcement that might address the problems--registries that help solve information cost problems, etc.). Whether the costs imposed by a restriction or type of restriction outweigh the benefits (the topic of a previous exchange) of course depends in part on the magnitude of the costs. (And we also have to consider the costs--administrative costs, error costs, etc.--associated with whatever doctrinal, or regulatory, or self-regulatory, etc., intervention we might make in the especially problematic cases.)