Cohen, D. (1995). Law, Violence, and Community in Classical Athens, Cambridge, England: Cambridge University Press, pp. 3-24 (chap. 1).
This chapter begins by laying out the two broad schools of thought explaining the origins of legal institutions. Functionalist accounts of institutions treat society as an organism, assuming that any institution that persists does so because it fulfills some societal need. But functionalism prefers stasis to change, regarding stasis as a sign of a healthy society, while most historians and political scientists see change as a better sign of societal health. This approach appears inconsistent both theoretically (i.e., it argues that internal conflict is a mechanism for preserving social order) and empirically (i.e., even in places where functionalism predicts conflict, the conflict appears dysfunctional and breaks down social order). The other major alternative to functionalism is evolutionary approaches, which view institutions as growing out of unmet societal needs. The problem with evolutionary approaches is three-fold. First, it assigns agency to nonsentient systems like "society." Second, it "sees the law as the principal instrument of social order in overcoming the centripetal impulse of the anonymity and individualism of modern societies" (p. 14). Finally, it views the development of institutions as a continual and digital path toward progress, with each stage of development representing a stepping stone toward a goal. Cohen is skeptical of both these explanations for the origins of institutions to regulate violence.
Instead, Cohen prefers a process-oriented explanation of institutional development that starts from the assumption that state mechanisms for dispute resolution were alternatives to private ordering, rather than displacements of them - alternatives that may not have been as effective at resolving disputes. This approach has the benefit of seeing institutions in a societal context; viewing legal disputes not as the culmination of a dispute, but as simply another maneuver in a dispute that sometimes erupts into violence and sometimes uses informal, nonviolent strategies; and both constitutes and is constituted by normative practices that define the proper scope of institutional control and create another arena for the playing out of agonistic processes that define political relationships.
My take: I agree with Cohen that functionalism doesn't work as a mechanism for describing the development of institutions. This is one case where the "half an eye" argument proves helpful. Functionalism presumes that the current state of an institution is how it came into existence; it is abundantly clear from even the most rudimentary of examinations that institutions change. Their prior incarnations fail to cope with certain problems; they reorient themselves to address those needs; the new institution doesn't sufficiently cope with other problems. In other words, the development of institutions is an iterative process.
What I am less certain of is that process-tracing from the litigant perspective is enormously helpful in tracing the development of judicial systems. I will grant, as Cohen argues, that where private participation determines the system's ability to regulate violence, then litigant process may be one part of the analysis. But institutions, as he points out, both prescribe rules for participants and are prescribed by the existing norms participants bring into their processes. Thus, it cannot be enough to limit one's perspective to that brought by participants. Instead, the process must be viewed as iterative - institutions have their own processes, which are effected by participants, and then the new processes are brought to bear on future participants.