The quadrilemma: appointing adjudicators in future investor–state dispute settlement

Concern with the selection and appointment of arbitrators has been central in the ‘legitimacy crisis’ surrounding investor–state dispute settlement (ISDS). The regime has been criticized for the outsized role of litigating parties in appointment, absence of transparency in the appointment procedure, potential for conflicts of interests, lack of diversity, and little emphasis on public international law competence. However, attempts to reform the selection and appointment of adjudicators involve confronting dilemmas, requiring trade-offs between different normative values. We therefore introduce a quadrilemma that captures the underlying values of independence, accountability, diversity, and procedural fairness that actors often seek to realize through adjudicatory design. We then set out seven idealized selection and appointment reform options under discussion in the ISDS reform process at UN Commission on International Trade Law (UNCITRAL) (from incremental reform through to new permanent mechanisms and removal of ISDS). The quadrilemma is employed to analyse their advantages and disadvantages of each model. In light of empirical and doctrinal evidence, it is clear that some reform options are more likely than others to optimize the quadrilemma. However, the effects are often conditional and sometimes there is a need for accompanying mechanisms. I . I N T R O D U C T I O N Concern with the selection and appointment of arbitrators has been central in the ‘legitimacy crisis’ surrounding investor–state dispute settlement (ISDS). The regime has been criticized † Coordinating author: Malcolm Langford 1 On the evolution of the crisis, see Susan D Franck, ‘The Legitimacy Crisis in Investment Treaty Arbitration: Privatizing Public International Law through Inconsistent Decisions’ (2005) 73 Fordham Law Review 107; Michael Waibel and others (eds), The Backlash Against Investment Arbitration (Kluwer Law International 2010); Daniel Behn, Ole Kristian Fauchald and Malcolm Langford (eds), The Legitimacy of Investment Arbitration: Empirical Perspectives (CUP 2022). VC The Author(s) 2023. Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons. org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Journal of International Dispute Settlement, 2023, 00, 1–28 https://doi.org/10.1093/jnlids/idad006

appointment reform options (from the status quo through to the creation of an appellate mechanism and removal of ISDS), and analysing their potential effects on the key concerns. These effects are predicted in light of existing scholarship on the selection and appointment in other international courts and tribunals 15 and a broad array of empirical evidence on the dynamics of ISDS.
However, we do so in the knowledge that such choices often involve trade-offs, alleviating one concern but exacerbating another. As Puig and Schaffer argue, in comparative institutional analysis, it is problematic if there is 'focus on the defects of a single institution while failing to apply the same rigour to its alternatives'. 16 Thus, for example, in the context of ISDS appointment and selection, mandating short renewable terms in a permanent mechanism might enhance the accountability of adjudicators, and therefore legal correctness, but compromise adjudicative independence and impartiality. We therefore introduce a model for considering trade-offs in design of international courts: the 'quadrilemma'.
The quadrilemma takes a departure point in the judicial 'trilemma' put forth by Dunoff and Pollack. They argued that international courts and their designers face trade-offs among three values, namely judicial independence, judicial accountability and judicial transparency. 17 Yet, a trilemma arises: it is only possible to 'maximize, at most, two of these three values'. 18 Thus, they assert that the 'trilemma provides a framework that enables international actors to understand the inevitable trade-offs that international courts confront, and thereby helps to ensure that these trade-offs are made deliberately and with a richer appreciation of their implications'. While we agree with this framework in general, we also take seriously its critics. The trilemma has been slated for overstating the importance of some values, occluding others, and providing overly narrow or broad definitions of some values. 19 In our view, a more complete model requires seeing a quadrilemma. This involves design trade-offs between four key judicial values: independence, accountability, diversity, and procedural fairness (including transparency).
Applying this model to the idealized selection and appointment reform options, we findunsurprisingly perhaps-no ideal solution. For example, increasing the number of adjudicators may increase geographic diversity but might complicate attempts to achieve case-based consistency (accountability) and reduction of costs (procedural fairness). To be sure, some models of selection and appointment are more likely than others to address states' concerns, especially those that involve greater institutionalization and centralization, and we do not hesitate to identify this. Yet, such evaluative findings must be treated conditionally and with some caution. Each finding should be interpreted in the context of the broader institutional solution to which each mode of appointment is attached and accompanying compensatory reforms, if any. For example, creating a permanent standing body with permanent members will, ceteris paribus, reduce ipso facto costs for litigating parties and shorten the duration of proceedings, but states could seek to reduce overall costs and length of proceedings with other policy interventions.
The article proceeds as follows. We begin by setting out the quadrilemma framework (Section II), then establish the seven stylized models of institutional reform and the corresponding selection and appointment choices (Section III). These seven models, loosely based on options under discussion, are: ISDS with no reform; roster(s) for partyappointment; institutional appointment of arbitrators; standing tribunal, but no appellate body; appellate body with first instance ISDS; standing tribunal and appellate body; and no ISDS. We then analyse the implications of each model for the concerns of states with ISDS (Section IV) and conclude by identifying the overall advantages and disadvantages of each model with a focus on the underlying trade-offs (Section V).

I I. T HE Q UAD RI LE MM A
In their analytical framework concerning selection and appointment of adjudicators, Dunoff and Pollack invoke an underlying 'judicial trilemma' that seeks to capture the tensions in creating and managing international courts and tribunals: [T]he states that design, and the judges that serve on, international courts face an interlocking series of tradeoffs among three core values: (1) judicial independence, the freedom of judges to decide cases on the facts and the law; (2) judicial accountability, structural checks on judicial authority found most prominently in international courts in reappointment and reelection processes; and (3) judicial transparency, mechanisms that permit the identification of individual judicial positions (such as through individual opinions and dissents). 20 In their view, this trilemma creates an optimization challenge. It is only possible to achieve at most two of these three vectors due to the inherent trade-offs. 21 Thus, an international court can 'exhibit high levels of judicial independence and judicial transparency', such as through open voting and/or individual opinions. 22 However, this is only the case they argue, if individual judicial accountability is low, for example with non-renewable judicial terms so that judges feel confident about identifying their view publicly. 23 The heuristic of the judicial trilemma has been praised for its analytical and 'thought-provoking' clarity, 24 but also criticized on three counts. First, not all of these criteria might be relevant or highly relevant for the design of all international courts. De Burca comments that 'Judicial accountability would not be high on any list I would draw up of the values to be pursued or the features to build into the design of an international tribunal'. 25 Moreover, she comments that, 'transparency, while a reasonably relevant feature, would not necessarily be an indispensable one'. 26 Secondly, authors have criticized the exclusion of other important values from the list. Many of those named relate to issues of process. Hilman argues that key values include 'fairness of decision-making procedures, quality and thoroughness of judicial reasoning, and efficiency and timeliness of judgment' and even 'judicial competence and experience'. 27 Thirdly, the construction of each category has been critiqued. Some scholars, such as Keller and Meier, argue that the definition of independence and impartiality is too broad and 20 Dunoff and Pollack (n 17) 225-76, 225. 21 ibid. 22 ibid 226. 23 ibid. 24  demanding. 28 It requires 'judges to decide disputes upon the facts and the law, free of outside influences such as the preferences of powerful states' and 'independent of venal or ideological considerations'. 29 This latter demand is too onerous, and in their view a narrower and more precise definition of independence would be: 'the freedom of judges to decide disputes free of improper outside influences'. 30 Hilman argues that the definition of transparency is too narrow-focused on the authorship of dissents and separate opinions, rather than their mere existence. 31 She also argues that the framework is too state-centric and focused on courts with state-to-state dispute resolution: 'Individuals are less likely to be able to hold judges accountable to them than states are. Moreover, it is easier for judges to be entirely independent of individuals than to be totally independent of states.' 32 In our view, the trilemma framework provides a useful departure point for considering the inherent trade-offs in designing selection and appointment procedures. Moreover, some of these criticisms of the trilemma go too far, at least in an institutional design context. For instance, states and many other actors in the ISDS reform process are deeply concerned with adjudicative accountability, as expressed in largely legal rather than political terms. They worry that adjudicators might not pay sufficient attention to correctness and consistency in decision-making. 33 Likewise, the lack of dissenting opinions in ISDS has led many to speculate that awards are a function of compromise, and the dynamics of the party autonomy system, rather than independent legal reasoning. However, the critics are right to point out Dunoff and Pollack's narrow selection of underlying values and the challenge that this creates for applicability to adjudicative systems outside the WTO.
In our view, these weaknesses with the framework in the context of ISDS reform can be addressed by re-constituting the trilemma as a quadrilemma. To be sure, adding another vector sacrifices some simplicity for completeness. In a quadrilemma, some measures might maximize only one to two values; others three. Yet, it permits the optimization framework to operate more easily in the world of messy and real policy reform that involves multiple crisscrossing considerations. 34 In the quadrilemma, we maintain two of the values from the trilemma: accountability and independence. However, we make two key alterations. First, we recognize that a key value in reform debates concerning international dispute resolution is representativity. In many current adjudication reform processes, the primary concern is geographic and gender-based diversity, but it extends to language, legal system, age, experience, education and development level of an adjudicator's state. 35 Representativity or diversity is regularly viewed as important in contributing to the sociological legitimacy of the adjudicative bodies and improving the quality of outcomes through enhanced cognitive diversity. 36 28 Helen Keller and Severin Meier, 'Independence and Impartiality in the Judicial Trilemma' (2017) 111 American Journal of International Law (AJIL) -Unbound 344-48. 29 Dunoff and Pollack (n 17) cited by Keller and Meier, ibid 344-45. 30 Keller and Meier (n 28) 345. 31 Hilman (n 27) 365. 'With respect to transparency, the authors are focused on the narrow issue of whether a judge's position or vote on a particular issue before the court can be readily discerned as the test for whether a court is transparent. Their measure focuses on the existence (or not) of dissenting or separate opinions. While recognizing the normative debate surrounding judicial transparency, Dunoff and Pollack do not take on the issue of why it matters who authored a given dissent or separate opinion. In theory, so long as a court permits judges to issue separate or dissenting views, those views ought to stand for themselves, regardless of who drafted them. In my view, a court that regularly issues dissenting views on an anonymous basis ought to be viewed as a transparent court, but the Trilemma analysis would reach the opposite conclusion.' 32  Secondly, while recognizing that transparency is important, it is preferable to see it as one of many values concerning procedural fairness. It is one thing to know 'who' in the arbitral or judicial panel was in disagreement, and 'why' as Dunoff and Pollack focus on; but parties are likely to have a myriad of other concerns with procedural fairness. Many procedural fairness issues are central to the UNCITRAL WG III process: duration of proceedings (efficiency), tribunal and legal costs (resources) as well as other transparency issues such as the existence of third-party funding. Figure 1 illustrates this reconfiguration as a quadrilemma, a framework that can be used to assess the different trade-offs for selection and appointment in ISDS and arguably other international courts and tribunals. Like the trilemma, there will be one value that cannot be maximized even if the others are maximized or remain unaffected. In other instances, the quadrilemma can mean that more than one value is potentially compromised. For example, the need to ensure greater diversity in geography and professional background would speak for a large number of adjudicators, but this may reduce the efficiency of the system in terms of costs and the value of procedural fairness. Its effects on accountability and independence can also be ambiguous, for instance improving correctness but threatening consistency, and making adjudicators too accountable to individual states rather than the collective of states as the principal.

I I I . STYLI Z ED M OD ELS FOR SE LECT I ON A ND AP POI N TM EN T
There are six basic types of institutional reform scenarios in which adjudicators could be selected and appointed. As displayed in the following Table 1, these options are distributed along a rough spectrum of centralization, from rosters for party appointment through to a standing tribunal, appellate body, and the abolition of ISDS'. The rest of the  to each institutional form, namely who nominates, who appoints, the renewability and length of tenure, and the number of adjudicators. The current system is also included at the beginning as a comparator, which brings the total options to seven.
To be sure, these possible structures are not meant to be exhaustive. They only serve the purpose of illustrating the possible contexts within which selection and appointment of adjudicators would or could be made. To that end, we make no claim as to whether the alternatives we discuss are the only ones conceivable.

A. ISDS with no reform
The current practice for the selection and appointment of arbitrators in ISDS cases is dictated by the conditions provided in the specific dispute settlement provisions in the applicable international investment agreement (IIA). Typically, these so-called ISDS clauses will state how a claimant-investor can initiate an arbitration once pre-dispute conditions are met. For the clear majority of ISDS clauses, a list of options will be provided. They vary widely across IIAs, but will frequently permit the parties to bring a claim administered: (i) by ICSID; (ii) by an international commercial arbitration centre; or (iii) through an ad hoc procedure using the UNCITRAL Arbitration Rules.
The language of the specific ISDS clause that the claimant-investor invokes will invariably also affect the way the arbitrators for the case are selected.

ICSID arbitration
Many ISDS clauses include reference to ICSID arbitration as one of the choices that claimant-investors can select. It consists of two distinct institutional aspects: an institution administering the arbitration and a set of procedural rules. As to the latter, the standard practice is a choice of the ICSID Arbitration Rules, although there are several ICSIDadministered cases that have applied the UNCITRAL rules. 37 The default procedure for appointment is as follows: the claimant-investor and respondent-state select, respectively, an arbitrator of their choice and notify the ICSID Secretariat; and the parties or co-arbitrators (or via an indirect process including input from both) select a chairperson arbitrator to preside over the proceedings. This default procedure does not require that any of the arbitrators be selected from any type of roster or list, even though ICSID does maintain such a roster. For the majority of cases using the default procedure, the parties are able to select and appoint a tribunal of arbitrators without any use of a roster or institutional assistance. Not all appointments happen in this manner. While there are no known ICSIDadministered cases where the ICSID Secretariat has appointed the entire tribunal, respondent-states refuse occasionally to appoint an arbitrator. In those cases, the ICSID Secretariat can step in and make the appointment. Moreover, and more frequently, institutional appointments at ICISD occur when the parties or co-arbitrators are unable to select and appoint a tribunal chairperson. In these instances, the ICSID Secretariat will make typically this appointment from a list of potential candidates that is circulated to the parties. The list is narrowed down to a few candidates (or just one) and the ICSID Secretariat makes the choice. 38 The ICSID Secretariat, in making their selection, are not required to draw candidates off any of the rosters or lists that ICSID keeps.
The one restraint that is unique for ICSID-administered arbitrations is that the partyappointed arbitrators cannot have citizenship that is the same as the citizenship of the party making the appointment, unless the other party agrees.

International commercial arbitration centre
The second option found in ISDS clauses is reference to the selection of an international commercial arbitration centre to administer the arbitration. These might be the Stockholm Chamber of Commerce (SCC), International Chamber of Commerce (ICC), or London Court of International Arbitration (LCIA). In these cases, which is currently a distinct minority of the overall ISDS caseload, the selection of arbitrators follows the same default ICSID procedure. Like ICISD, none of these centres mandates that any party-appointed arbitrators or institutional appointments must be selected from a roster or list of potential candidates. However, the parties can appoint an arbitrator with the same citizenship as that of the party making the appointment.

UNICTRAL arbitration
In most ISDS clauses, the claimant-investor is given the choice to initiate an ad hoc arbitration that is subject to the UNCITRAL Arbitration Rules. The default arrangement, if this option is taken (which is about a third of cases), 39 is that the arbitration will not have any institutional administration. The selection of arbitrators in these cases is derived from the selection rules in the UNCITRAL Arbitration Rules. It follows typically the same process as ICSID-administered cases. If one of the parties refuses to make an appointment or if the parties or co-arbitrators are not capable of agreeing on a chairperson, the UNCITRAL Arbitration Rules require that an appointing authority be used. The parties are empowered to propose an appointing authority, which can include the Secretary-General of the PCA. The appointing authority then must circulate a list of potential candidates (not required to be drawn from a list or roster) to each of the parties, and the appointing authority will make an appointment based on candidates that have not been struck from each parties' respective lists.
We now turn to the six reform alternatives.

B. Roster(s) for party-appointment
The use of a roster to allow parties to select arbitrators for ISDS cases is one possible reform option. 40 Whether it modifies the status quo in practice depends though on how the roster is designed. Undoubtedly, there will be several structural and institutional conditions that must be in place to make a party-appointed roster system function. The most important are an institution to keep, update, and make available a list of arbitrators together with a mechanism to ensure that a roster system trumps the current default rules available to parties when selecting arbitrators under current ISDS clauses. With that said, there are several ways that a roster or roster could be used.

Types of rosters
Any roster-based reform will require that a choice is made about what types of rosters will be used, the conditions for nomination to a particular list, which institution will host the list, and how the parties will select from these lists.

a. Multiple rosters
One possibility is to keep the current lists or rosters that are maintained by the various institutions already administering ISDS cases. ICSID and the PCA, ICC, and LCIA all keep some form of roster or list. ICSID has a roster of 10 Members of the Panel of Arbitrators. In addition to this, there is a List of Designations by the contracting states to the ICSID Convention. A contracting state may nominate up to eight individuals, and these individuals do not need to have the same citizenship as the state appointing them. For both the Panel of Arbitrators and the List of Designations, the term is 5 years, and renewable. The PCA maintains a roster called the Members of the Permanent Court of Arbitration. 41 This includes designations by PCA Member States of up to four individuals (which do not need to have the same citizenship as the appointing state). The ICC also has members available for dispute resolution (currently, there are 178); 42 and the LCIA maintains similarly a list of members. The SCC does not have a roster or list that is publicly available.
Currently, the use of these rosters or lists by parties is not mandatory. In fact, many of the individuals on these lists (the ICSID and PCA lists) have not, to date, sat on any ISDS arbitral tribunal. Nonetheless, one reform option could be to permit parties (and institutions when they are required to appoint) to use the lists that are currently already available; and allow or require a party (or institution) to voluntarily select off a particular roster or a list of eligible rosters.

b. A single roster
Another possibility for a party-appointment-based system is to establish a new single roster that can be used by the institution established through the reform process. There are several ways this could be established and designed. As stated, at ICSID, there are currently two ways that arbitrators can be placed on the ICSID roster. The first is institutional nomination. The ICSID Secretary-General nominates and selects 10 individuals to be on its Panel of Arbitrators (which currently contains five women and five men). All individuals on the panel have sat as an arbitrator in an ISDS case. The second is state-based nomination. Here, the contracting states to ICSID nominate up to 8 individuals to be on the list or roster. Not all contracting states have made nominations and not all contracting states have nominated the maximum number of individuals.
A new single roster could be created for use by parties either on a mandatory or voluntary basis. The roster could be exclusively composed of nominations from contracting states to the new institution; or the roster could follow the precedent of the ICSID list and have a combination of the two modes. The next choice will be to determine the number of candidates that can be nominated, and the current rosters vary considerably in this respect. If only institutional nominations are permitted in a reformed system, it is likely that more than 10 individuals will need to be appointed to the list given the number of cases and chances of conflict of interests and non-availability. Likewise, if only contracting states are able to nominate, then (depending on the number of contracting states that sign up to the new institution) more than 8 individuals will likely need to be nominated by each eligible State.
For a single roster based on the retention of a party-appointment system for selection of arbitrators, 150 and 200 individuals would be needed in all likelihood.

Use of roster(s)
Once the determination on what type or types of roster is made, the next issue that will arise is how the roster or list is used by the parties to a particular ISDS dispute.

a. Mandatory selection from roster(s)
One reform possibility is that selection from the list is mandatory for both party-appointed arbitrators, as well the chairperson. An alternative is to allow the two party-appointed arbitrators to be freely selected, but that the chairperson must be selected off the list (whether the appointment is made by the parties or co-arbitrators or an institution).
If parties are required to select off a single roster or from the rosters already in use, it will be critical that these lists provide a good distribution of arbitrators. It should, for example, be geographically diverse, with an appropriate number of women, and all candidates should have the requisite expertise that parties (or the ISDS refrom process) require. Currently, it is doubtful that the rosters in place at the various institutions administering ISDS cases would meet such criteria.

b. Voluntary selection from roster(s)
Another possibility is to maintain the current status quo and use rosters only voluntarily by the parties to a particular ISDS dispute. This would not foreclose the possibility that a new post-reform institution could be established and create its own list or roster, however. If such a voluntary roster is used, it is less important that those on the roster have the relevant expertise or experience that parties would demand. Nonetheless, such a roster or list is unlikely to be frequently used by the parties or by the new institution in selecting arbitrators; much in the same way as the current lists or rosters kept by the PCA or ICSID are used infrequently. 43

C. Institutional appointment of arbitrators
The next reform option keeps ISDS in place but shifts the process for the appointment and selection of arbitrators decisively from the parties to an institution. While many of the design considerations in the use of rosters or lists for the maintenance of a party-appointment system of arbitrators will apply here, there are a variety of ways in which an institution could select arbitrators. One possibility, as mentioned above, is to institute a hybrid selection process. The parties to an ISDS dispute each select and appoint an arbitrator of their choosing (either from a roster or not) and then the relevant institution selects and appoints the chairperson (either from a roster or not). Another possibility is to have all appointments made by an institution. In the following sub-sections, we examine the manner in which appointments made by an institution could be structured, with a focus on who makes the appointments and whether the use of a roster is mandatory.

Who selects and how?
On the question of who decides, there are a number of options. One manner in which arbitrators could be selected is to vest authority in the Secretary-General of the arbitral institution established as part of the reform process. An alternative would be to have a standing committee within the institution that would be responsible for making all appointments. This would require consideration of the rules for such a committee which would cover: how arbitrators would be selected (eg through a majority voting process or an internal list process); the size of the committee and competence requirements; and the length of the terms for committee members.
The next question that would require thought is how arbitrators could be selected, which can range from the institution making all appointments with no input from the parties to permitting party input. As to the latter, the institution could, for example, circulate different lists of potential co-arbitrators and each party could rank their preferences; followed by a similar process for the potential chairpersons.

Mandatory or voluntary selection from roster(s)
The next question is whether a list or roster will be used by the institution in making appointments. The composition of such a roster would follow the same considerations that would need to be made for a party-appointment-based roster discussed in the previous section. Once the rules and parameters are set for establishing a roster, consideration must be given to whether the institution must select from the roster for all arbitrators (or the chairperson) or whether use of the roster remains voluntary.

One versus three member tribunals
A final aspect in regard to the institutional appointment of arbitrators for ISDS cases would be whether a mandatory threshold rule could be established for the use of a sole arbitrator or a three-member tribunal. For example, one could require that all disputes with an investor claim less than a certain amount (eg 5 or 10 million USD) would only require a sole arbitrator. 44 For the appointment of the sole arbitrator, the institution could use the same mechanism chosen for the selection of the chairperson in a three-member tribunal.

D. Standing tribunal with no appellate body
The next option is to move away from ad hoc selection and appointment of arbitrators (whether institution or party-based) to that of a standing investment tribunal or court. 45 The organization of such a body court could take many forms, but in regard to the composition of the tribunal or court, the selection and appointment of judges would be significantly different than the current ISDS system. 44  In analysing this fourth option, we presume that no second-level review through an appellate mechanism would be created (see though option 6 below). We presume also that the awards or judgements rendered by such a standing tribunal or court would then be subject to similar rules currently used for the enforcement and recognition of arbitral awards according to the New York Convention. 46 In regard to the selection and appointment of judges to such a tribunal or court, the key considerations would be selection criteria, selection process, tribunal size and term lengths and renewability.

Criteria
The first consideration is the criteria to be used for those eligible for nomination to the tribunal or court. These would be established most likely in the instrument creating the body. All potential nominees would have to possess relevant expertise, standing, and experience. However, as we shall see, overly demanding criteria can limit other goals in selection and appointment design. 47 Further considerations for criteria are to determine whether gender balance would be mandated and whether each state would be required to only nominate its own citizens or could also nominate candidates with citizenship from third states. 48 There is also a need to consider whether a system for ad hoc judges to be eligible for appointment to specific cases involving particular states or issues should be established-a subject on which states have diverged significantly. 49

Nomination and selection
The next consideration would be to establish the rules for nominations and selections. Nominations could be made by states, nominees themselves, or third parties such as bar associations, universities etc. States are currently considering all of these options in the UNCITRAL reform process. For example, each contracting state to the new institution created as part of the reform process could nominate a set number of candidates to be selected by the institution, an independent selection panel, or states in a vote or by consensus. Current discussions in UNCITRAL point in the direction of an independent selection panel of former judges and lawyers/academics of high standing, which would be elected/chosen by states. 50

Size and structure
Another consideration is the number of judges and the structure of appointment. A simple conventional structure would be to have a set number of judges with a president and vice president. The number appointed could reflect the number of contracting states to the institution, or could be a set number in multiples of three so that chambers could be established to hear individual cases (eg a 15, 18 or 21-member court scepticism to a standing mechanism with full representation on grounds of cost and complexity, 51 but some were open to adjusting the size of the tribunal over time. 52 Another option would be to create a roster or list of court members that could be used by the institution in selecting panels of three judges for specific cases. This would permit the possibility of a much larger pool of judges to draw from. The same criteria and nomination procedures for a traditional standing court could be used to select judges on the list or roster. In terms of selecting judges for specific cases, the process could be random or fall to the secretary-general of the institution or the court's president.

Terms
A final consideration is term lengths and limits. There are many international courts that set terms for judges that could be used as models for a standing tribunal or court. 53 For example, judges could be appointed to terms that are 3, 6 or 9 years; and these terms could be nonrenewable or renewable once or twice. The appointments could also be staggered at 3-year intervals so that the turnover of new judges on the court would be gradual-although only a minority of international courts have staggered appointments. 54

E. First instance ISDS with appellate body
The next option would be an appellate body, established to review all first instance ISDS awards. Such a system might retain some traditional form of ISDS, either as it is currently practiced or as reformed through the use of rosters or institutional appointments (options 2 and 3 above). The scope of review of this appellate body's jurisdiction would have some effect on the selection and appointment of appellate judges. If the scope is de novo, then appellate body cases will take much longer than if the review is restricted and a greater number of judges would be needed and potentially with more varied competence. An additional consideration is whether review of first instance ISDS cases by the appellate body would be mandatory and automatic; or whether it would fall to the choice of the party or parties to request it. 55 The establishment of the appellate body and how it selects judges could follow the same design options as articulated for a standing first instance tribunal or court. One specific consideration for an appellate body that may have an effect on the manner in which judges are appointed and selected, is if an option for an en banc or grand chamber review is established for exceptional cases (as is the case for European Court of Human Rights). This would require that the size of the appellate body be set, even if judges are drawn off a list or roster for typical three member cases.

F. Standing tribunal with appellate body
The next major reform option could be the establishment of a standing first instance and appellate body. The selection and appointment of judges to both would follow the same set of considerations described in the two previous sub-sections (standing tribunal with no appellate body, first instance ISDS with appellate body).
The one additional consideration for this option is whether the standing first instance court and the appellate body remain separate institutions with their own rules and procedures for selecting and appointing judges, or if the rules would be the same. It is 51 ibid, para 35. 52  The quadrilemma 13 notable that in CETA the nature of the appointment of judges in the appellate body was left unresolved. 56 A related option would be to integrate the judges sitting on the first instance standing tribunal and the appellate body, regardless of the initial appointment procedure. For example, one president could be appointed to the standing tribunal and another to the appellate body, and made respectively responsible for selection to each body from the same pool of judges (or list or roster if used).

G. No ISDS
The last model that can be considered would be the removal of all formal adjudication from international investment agreements. Under such an option, selection and appointment of judges or arbitrators would not be an issue: it would be left to the relevant domestic appointment systems, whether it be permanent national courts or alternative dispute resolution (ADR), or inter-state dispute settlement procedures. For example, one submission to UNCITRAL notes that the Korean Office of the Foreign Investment Ombudsman (OFIO) provides 'investment aftercare to support investors who face grievances'; and the Brazilian Cooperation and Facilitation Investment Agreement model-with an investment ombudsmen and a joint committee of representatives from the home and host state-helps to both prevent and resolve conflicts. 57 Thus, questions of selection and appointment would relate to these distinct and often tailored solutions.
ADR may also be international, whether through mediation or expert/administrative review panels. Moreover, as South Africa noted in a submission, ADR can involve either 'conciliation or mediation, but it may also concentrate on a fact-finding exercise that makes it possible to narrow down the actual extent of the dispute'. 58 This could assist in resolving disputes more quickly and cheaply, as well as general conflict de-escalation. 59 The state also noted the role of arbitration institutions, which could propose simplified rules for ADR, provide for more flexibility in rules on conciliation, mediation and fact-finding, develop ADR capacity, encourage the inclusion of ADR experts in arbitrator lists, and provide logistical and secretarial support. 60 If such alternative mechanisms are adopted and there is a need for the selection and appointment of individuals to mediate disputes or sit on informal review panels, the various options provided in the above section could be used to select and appoint such individuals.

I V. I SD S C ON CER NS AN D STYL IZ ED M OD ELS
We now turn to analyse, evaluate, and partly predict the effect of these stylized models of selection and appointment on the four central values of procedural fairness, accountability, diversity, and independence. Each of these values is analysed specifically within the context of the specific concerns in the current UNCITRAL reform process.

A. Procedural fairness 1. Costs
The costs of investment treaty arbitrations can be substantial and their rise and rise over the past decade has attracted increasing concern. The result is that has a central place in the WG 56 Bjorklund and others (n 10 III reform discussion. 61 Moreover, scholars have pointed out the challenges for smaller investors, especially given that the costs of litigation can almost equal awarded compensation. 62 Reforming selection and appointment reform options can reduce these costs, although the potential varies according to scenario and context. Arbitration costs are divided between (i) legal costs: counsel costs and experts' costs; and (ii) tribunal costs: arbitrator fees, and tribunal, arbitral institution and hearing venue costs. 63 Table 2 shows recent studies on legal costs, which are the largest cost component of any investment treaty arbitration (besides an eventual damages award). 64 Using PITAD data 65 up to 1 February 2019, claimant's legal costs in 169 cases were 6.1 million USD, while respondent's legal costs in 177 cases were 5.2 million USD. 66 In addition, legal costs associated with the calculation of damages in successful arbitrations appear to have significantly risen, although it is challenging to measure as costs are not disaggregated in most arbitral awards. 67 However, in Tethyan Copper v Pakistan it was made transparent, with the claimant spending USD 22 million on legal fees and financial experts and Pakistan almost USD 10 million on the quantum phase alone. 68 As for the fees associated with tribunals, it was estimated in 2010 that 18% of the overall amount of any given investment treaty arbitration was the cost of the tribunal, but that ratio has now fallen below 10%, as Table 3 indicates. 69 Average tribunal costs are now in the vicinity of 1 million USD.
The question is then which of the reform scenarios for selection and appointment is likely to reduce tribunal and legal costs. Without any doubt, the three models that embrace permanent adjudicators (4, 5 and 6) will reduce the direct tribunal costs for parties. Both the investor and the respondent state will be able to take advantage of the standing mechanisms for free or relatively low filing and procedural fees. In options 2 and 3, the tribunal costs will remain, although legal costs may fall with option 3 given the amount of time that parties' lawyers use currently in the selection of arbitrators. 70 However, there are four important caveats in reaching this obvious finding on more permanent bodies. First, the state principals to the new institutional regime will need to pay for the judges/standing members. Thus, there will be indirect costs with a standing mechanism, although it is likely that developed states would bear more of this burden, thus reducing overall tribunal costs (including indirect costs) in a very substantial way for developing countries.
Secondly, the reduction in the overall cost burden is not necessarily large, even if it is significant. This is because the bulk of costs in ISDS relate to legal costs, as Tables 2 and 3 makes clear. Nonetheless, it might be possible to argue that legal costs could be reduced in a more centralized system. For example, a standing court may be empowered to institute case management principles that focus proceedings on key issues. 71 Indeed, there are real economic incentives for members of standing tribunals and appellate bodies to institute case management principles. If they are paid a fixed salary, the absence of such principles will increase workload. Permitting parties to ventilate many issues in a case, does not result in greater judicial compensation; it simply demands more time and, in accordance with judicial behaviour theory, one could presume that most judges seek to avoid excessive workloads. 72 However, a permanent tribunal is neither a clear necessary nor sufficient condition for reduction of legal costs. States would probably need to signal or require the need for case management approaches. In addition, the proposed Advisory Centre may be able to assist low-income states to reduce legal costs. 73 Thirdly, the conclusion that permanent bodies are more likely to reduce tribunal costs and possibly legal costs is partly conditional on cases proceeding at a relatively prompt manner. If a standing mechanism was inundated with cases-as occurred with the European Court of  Human Rights from the late 1970s 74 -then costs may increase for parties. This increase would or could result in slightly higher legal costs from delayed proceedings, but also potential 'opportunity costs' for both parties in terms of time used in the litigation which would only be partly reflected in any eventual compensation award. While it is unlikely that such permanent mechanisms will be overburdened in their first decade or so, given the likelihood that states would only gradually adopt the new system, some mechanism to adapt to increased volume of litigation should be considered. For example, one could envisage a requirement that the number of adjudicators be easily expanded-with mandatory funding-if an independent body concludes that the cause of delay is linked to workload. 75 Finally, even if arbitration costs are reduced through more permanent mechanisms there may be challenges in maximizing other values, especially independence and diversity, as we shall explore below.

Duration
From the time of registration of a dispute, the current average length of ISDS proceedings for an arbitration is almost 4 years. 76 Whether this period can be considered 'excessive' is contingent. It is partly a normative and legal question and empirically depends on the choice of comparators (some international and domestic courts are faster in processing; others slower). 77 In any case, many of the proposed reforms involving appointment could have a significant impact on the duration of proceedings. A sizeable portion of each ad hoc arbitration is devoted to the selection and appointment of arbitrators, including challenges to arbitrators and replacement. Yet, as we shall see, the magnitude of the average decrease in time for each institutional option is highly dependent on accompanying reforms.
Two of the principal causes of lengthier proceedings relate to the initial selection process and challenges to arbitration. First, the time taken to constitute an ad hoc tribunal is significant. A survey of the arbitral proceedings conducted by ICSID showed that the average duration of the constituting of tribunal was 258 days. 78 While Behn, Berge and Langford find that non-ICSID tribunals are constituted slightly more quickly, they also find that selection processes can take between 3 and 12 months. 79 Thus, the institutional reforms which involve a permanent body (options 4, 5 and 6) will clearly see a reduction of time devoted to this aspect of the case.
Could the use of a roster under options 2 and 3 also decrease the time taken to constitute a tribunal? This is a possibility, but depends on the appointment process. If parties retain the ability to choose the arbitrator or select from a shortlist (option 2), there is unlikely to be any reduction in the length of the proceeding. If an institution chooses the arbitrator from a roster (option 3), then this procedure could lead to a significant reduction in the length of the selection and appointment of an arbitrator, although not as much as a standing body as the institution must be in contact with arbitrators over availability, conflict of interests etc. The abolition of ISDS would mean that cases would be in the hands of permanently appointed domestic courts or domestic arbitration where relevant. However, whether this would be an improvement on duration of cases would be highly country-dependent. 80 Secondly, a statistical regression analysis of the key causes of delay identified four statistically significant determinants: bifurcation, arbitrator challenges, arbitrator replacement, and dissenting opinion. 81 A move to permanent bodies would affect directly two of these four factors. First, arbitrator replacement processes (due to illness, death, resignation or successful challenge) would not be relevant, given the existence of other available judges. Secondly, challenges to adjudicators would be much less frequent. Notably, the factor that had the largest effect on delay was arbitrator challenges. The use of a roster with institutional appointments could lessen delays from such challenges and replacements, although the process may be slightly elongated. A roster with party appointments is unlikely to have any effect on these two delay-causing factors.
It is important though to consider separately options 5 and 6, which involve appellate review. While the appointments would be speedier and challenges/replacements less of a problem, appellate review in and of itself may lengthen proceedings. The key question here is whether the average time taken to select arbitrators for existing 'appellate' processes (such as ad hoc ICSID annulment committee hearings, 1.91 years as at 2019) is longer than the average time for future appellate review. On one hand, appellate review would most likely cover a wider range of grounds for challenge and thus could potentially involve longer proceedings. On the other hand, there would be no time lost in establishing a tribunal. An appellate review body could be entrusted also with the power to retain the case and render a new arbitral award. 82 This option is not possible under the existing ICSID annulment system or other domestic court proceedings for annulment.

Consistency
In UNCITRAL WG III, states have been concerned that the lack of consistency in awards could negatively affect the reliability, effectiveness, and predictability of the ISDS regime and its credibility. 83 This would undermine ideas of the rule of law, general legitimacy in the system, and confidence in the stability of the investment environment. Thus, there is a widespread view that individual decision-makers must be made more accountable to systemic concerns. Consistency, however, is a multivalent term. It has different meanings as it relates to both outcomes and interpretation. 84 First, we may be concerned with outcome-based consistency: in the application of the law, adjudicators should treat like cases alike. There have been cases of claims based on similar facts, arising out of a single governmental measure, and brought under the same substantive and procedural rules, which have been decided differently by different tribunals. 85 Secondly, there is concern with jurisprudential consistency, ensuring doctrinal uniformity across arbitral decisions. 86 Finally, there is inter-system consistency, in which some states and scholars have expressed concern that investment arbitration has failed to ensure consistent interpretive approaches to interpretation and outcomes across different regimes in public international law, for example, trade, human rights, and environment. 87 81  This perceived lack of consistency has, at least in part, been attributed to the ad hoc nature of ISDS arbitration. Each tribunal is typically composed to decide one particular dispute and there is no formal doctrine of precedent in public international law. New methods of selecting and appointing adjudicators may help enhancing at least the first two types of consistency. To that end, ISDS Academic Forum members have illustrated how different reform options may assist. For instance, 'a permanent investment court or appellate body . . . can achieve, relatively quickly, a high level of consistency, coherence and predictability beneficial for both public authorities and business'. 88 However, as they note, the extent to which each reform model will contribute varies, and there are drawbacks with a strong focus on consistency. 89 It is clear that greater institutionalization will ensure greater outcome and jurisprudential consistency. That is one of the prime functions of an appellate body, which is found in options 5 and 6: 'rogue' or 'outlier' decisions could be appealed, as determined though by the scope of the appellate body's jurisdiction. 90 The presence of an appellate body would reinforce consistency also in the model of first instance ISDS with Appellate Body, as arbitrators or judges would be concerned with the risk of being overruled. However, outcome consistency in treaty interpretation would be highly dependent on a secretariat assisting the body.
Moreover, the creation of a standing first instance body (option 4) could also contribute to greater outcome and jurisprudential consistency. It would be formed by a stable group of appointed members. This would likely favour consistency by the very fact that adjudicators would be limited in number and judge on an extended number of cases. If standards for selection would also include relevant expertise on international law, issues arising out of the coexistence of different treaties and their interpretation according to the cases could arguably improve; although some might argue that there is the risk of greater outcome inconsistency if judges do not have sufficient commercial competence to grasp the facts of a case. 91 Again, the existence of a single secretariat in assisting tribunal members would also ensure greater consistency.
It is also possible that such institutionalization in options 4-6 would lead to greater coherence with other branches of international law: that is, inter-system consistency. However, such a scenario is only likely with clear instructions or a change in substantive law. The WTO Appellate Body, upon which many model a future investment appellate body, 92 has partly kept itself insulated from the rest of international law, including international investment law. 93 While others point to significant convergence in reasoning, 94 Johansen and Sachs are concerned that: domestic and international law and policy and relevant expertise, exacerbates concerns that any law developed by a new court or appellate body will be unduly ignorant of or unconcerned with non-investor rights and interests. 95 The other selection and appointment options are less likely to increase consistency and may potentially decrease it over time. According to Brown et al., so-called improved ISDS, which includes options 2 and 3, 'may lead to qualitative improvements in the ISDS system, and may have indirect benefits, but it would not directly lead to improvements in the consistency or coherence of decision-making'. 96 Thus, the greater introduction of rosters with party or institutional appointments may decrease the number of arbitrators but consistency may only be achieved to the extent that such a reduction occurs. Moreover, they conclude that our option 7, No ISDS, would be unlikely to lead to 'benefits from the perspective of unjustifiable inconsistencies as different national courts, which would not be bound to have regard to the others' decisions, would be ruling on the correct interpretation of IIA provisions'. 97 Yet, as we shall see, all these benefits of reducing the number of adjudicators raise questions for another vector on the quadrilemma-diversity (see section 4.C).

Correctness
An excessive or one-sided focus on consistency can also come at a price for another aspect of accountability, namely correctness. The existing system permits 'experimentation', and thus potentially correctness: 'correct solutions tend to bubble to the top over time, and higher quality reasoning is generated in the long term'. 98 In the ISDS reform context, correctness is also related to the concern with accountability in the design of international dispute resolution. How to ensure that the agency of adjudicators is appropriately constrained in the process and substance of decision-making such that they arrive at a legally correct decision. In particular, how do they avoid misidentifying and misapplying applicable law. 99 While consistency is a systemic notion, correctness is a decision-specific one, concerning the substantive quality of the legal reasoning in a specific case. For ISDS, one particularly important aspect of correctness is whether different treaties (or factual circumstances) warrant different interpretations. Here there may be trade-offs between a correct decision and the systemic consistency of jurisprudence. 100 As Alschner has pointed out, states have imposed a clear hierarchy between the two objectives in their UNCITRAL deliberations, 101 stressing that 'consistency should not be to the detriment of the correctness of decisions'. 102 In other words, states are more concerned about consistently incorrect arbitral decisions than inconsistent but correct ones. Interestingly, Alschner finds empirically that arbitral tribunals tend instead to prioritize consistency over one measure of correctness: case citations. He finds that three out of four citations, based on a dataset of more than 4500 references, connect to highly dissimilar IIAs. 103 While this empirical research indicates a potential problem with correctness, determining epistemologically what is a correct decision in individual cases is nonetheless demanding. Few, if any lawyers, have the time and super-mind of Dworkin's ideal judge Hercules. 104 Nonetheless, we can at least identify selection and appointment factors that may enhance the quality of legal reasoning and outcomes. We focus on four: tenure, representativity, institutionalization, and legal qualifications. 105 First, the establishment of renewable terms in a standing and appeal mechanisms could function as an incentive to correctly identify and apply applicable law. As Larsson, Squatrito, Stiansen and St. John put it: 'Renewable terms can improve what we refer to as judicial accountability as states can base reappointment decisions on the past performance of the judges.' 106 Thus, models of appointment which rely on long and fixed judicial terms (also for national courts in the no ISDS option) are less likely to increase legal accountability and correctness. It might also be argued that amongst the arbitration options (1-3), the current system of open appointments or large but functional rosters may include more incentives for accountability than a small roster where arbitrators feel more confident of reappointment.
To be sure, there is a risk that enhanced accountability simply leads to more political correctness, rather than legal correctness. Political correctness involves arbitrators responding to state preferences as to interpretation, procedure, or outcomes for individual reasons such as reappointment 107 and reputation. 108 This is not to discount necessarily the role of political correctness in maintaining the sociological legitimacy of tribunals and legal regimes-with accountability producing a collective effect on adjudicative behaviour. Indeed, some scholars and states in UNCITRAL WG III maintain that such legitimacy can be important for garnering state's consent to jurisdiction and compliance to dispute settlement and ongoing financial and budgetary support for its institutions. 109 Thus, reform models that make bodies more accountable may produce decisions that are both more legally correct and politically acceptable. But this is far from a given. The result may be itself a trade-off between a correct legal result and the political placation of parties.
Secondly, selective rather than full representation approaches to appointment can enhance legal correctness. This is because the accountability constituency will be diffuse. Larsson, Squatrito, Stiansen and St. John argue that: On full representation courts where each state controls the appointment of one judge, judges may be expected to primarily seek to maintain support from "their own" state. By contrast, if judges are elected through majority voting, they will need to maintain the support of a larger coalition of states. 110 Thus, in selective approaches, adjudicators are answerable to many in both selection and renewal. And such models, in which the numbers are constrained-especially an international court with many state members or appellate panel-may be more likely to lead to decisions which are correct, politically and possibly legally. Although, again, this can create a dilemma for maximising adjudicative diversity.
Thirdly, reform models of any kind that put greater emphasis on qualifications of adjudicators may improve the quality of correct reasoning. However, different types of qualification may relate to different types of correctness. It may be public international law competence with treaty interpretation; commercial and linguistic competence that is relevant to understanding and adjudging the facts of cases; or technical competence in the calculation of damages. 111 There are also trade-offs within a quadrilemma context. For example, the UNCITRAL Secretariat proposed, in selection processes, a simple and broad focus on reputation and competence in international law, with specific reference to investment law and dispute settlement. 112 This was received positively by states as it did not unduly limit the pool of candidates, which requirements for adjudicative experience or specific competence would have done. States reacted negatively though to the additional proposal of 'experience in or consulting governments including as part of the judiciary' om grounds of independence. 113 It could create a perception of bias even though the proposal was meant to focus on enhanced competence, that is independence would be compromised by too strong of a focus on accountability. Likewise, states were divided though over a requirement for language competences with some emphasizing its contribution to diversity; others its limiting effect on the pool of candidates.
Finally, greater institutionalization of ISDS may increase accountability, although the basis for such reforms is only partly a question of selection and appointment. Thus, for example, the mere establishment of an appellate body may enhance correctness as this is a key reason for its establishment-a second look. Likewise, a robust secretariat and more regular contact between and with adjudicators under options 4-6 may produce more correct or higher quality decisions. Yet, even creating space for greater engagement of third-party states, nongovernmental organizations and affected citizens to participate in proceedings, under all nonstatus quo options, may assist as there is an increase in concentrated expertise and more available information. 114 There is, nonetheless, some empirical support for these institutionalization claims. One study finds that the greater centralization of the WTO system explains partly why it has been more responsive (accountable) to state concerns than investment arbitration. 115 However, while centralization may increase consistency and political accountability, it is arguably only features that increase expertise and information flows as well as a second review that would more clearly have a strong chance of increasing legal correctness. But even then the claim needs to be nuanced. An appellate review body will increase dialogue and concentrate attention on correctness, but pressures for consistency may sometimes be at the cost of legal correctness.
As to the last reform option, the role of national courts and state-to-state dispute resolution should not be discarded in relation to correctness. While national courts will struggle with expertise about international law, they may have more expertise in relation to the facts, relevant domestic law, and the overall context. Thus, they may be stronger on factual correctness, although this strength would vary from court to court, country to country.

C. Diversity
Diversity has been the subject of sustained discussion in the UNCITRAL reform process, given its implications for the quality of adjudication as well as perceived legitimacy and eventual compliance. An ISDS Academic Forum working group summarized the reasons for this as follows: It is commonly accepted that decision-making bodies should be inclusive and that decisionmakers should represent the diverse constituencies of the stakeholders subject to their decisions. This feature serves multiple purposes. Social science literature shows that diverse decision-makers are more likely to avoid cognitive biases and group-think in decision making. Moreover one or more decision-makers might have the cultural knowledge to understand the dispute in context. Diversity among decision-makers may improve the quality and rigor of the decisions they render, and in doing so effect or enhance the normative legitimacy of a particular system. Further, diverse decision-makers are likely to be perceived as capable of producing fairer decisions, which is likely to enhance the sociological legitimacy of a particular system. 116 The discourse on adjudicative diversity in ISDS tends to focus on the low number of appointments of women and nationals of non-Western states. 117 Female arbitrators represent roughly 10% of overall appointments in ISDS cases and non-Western arbitrators roughly 26% of appointments in ISDS cases (up to 2019). 118 Other potentially important diversity variables among international adjudicators include experience, legal and professional background, age, education, and religious and cultural background, as well the less tangible and measurable values such as their ideological preferences andworldviews.
Currently, the party-driven appointment process in ISDS curtails opportunities for diversity. Those appointing arbitrators in ISDS cases are often guided by the 'prior experience norm'. 119 New entrants are less desired as disputing parties are unsure of their likely positions on doctrine and facts. This norm results in repeat appointments of arbitrators that are Western, male and older-the so-called 'pale, male, and stale' effect. Thus, it is unlikely that maintaining the existing system would improve significantly diversity, absent specific reforms to appointment processes. Indeed, existing initiatives that do not rely on institutional intervention have only had a limited effect thus far in improving gender and geographic diversity among ISDS arbitrators. 120 Given this background on the diversity of adjudicators in ISDS, there is likely to be significant gains in diversity (at least in terms of gender and nationality) if the various reform options are considered. They would remove a completely unrestricted, decentralized, and ad hoc system of party-appointed arbitrators, but it would be dependent on new institution mandates and mechanisms for improving diversity.
Moving through the various reform possibilities, options 2 and 3 move ISDS away from an unrestricted system of party-appointment to one of institutional appointment (option 3) or institutional suggestion (option 2). Although the nature of the appointments remains ad hoc, it would be based on what could be a relatively large pool of candidates eligible for appointment. In terms of improving diversity, it is likely that option 3 of these two could have the most dramatic impact, so long as the institution making appointments is under a mandate to meet diversity criteria. Indeed, currently in those instances where institutions are asked to make appointments (usually to select the chair), there is more diversity compared with party appointment. 121 An even more clear-cut example is the appointments that ICSID makes in annulment proceedings. Backed by a policy that aims at increasing gender and geographic diversity, ICSID has appointed approximately 50% non-Western members and 20% female members. 122 Reform option 2 would require the selection of arbitrators from a roster, and to achieve diversity there would need to be at least some sort of requirement to address gender and geographic diversity. However, it may not result in any improvement in diversity in practice if the roster goes unused. If such a roster is mandatory though, then parties will make at least appointments from a more restricted (and more diverse) pool of arbitrators. Although, there is nothing that can be done to ensure that parties actually select in a diverse manner unless there is a further mandate in place: for example, there must be diversity in the presidential appointment if it is not achieved in the party appointment of wing arbitrators.
Through the establishment of a permanent and institutionally centralized system of adjudication, reform options 4, 5 and 6 permits more control over diversity. However, the degree of diversity may be limited by the number of possible members. An appellate mechanism would most likely have a maximum of 12-15 members, and a permanent mechanism only slightly more. Bodies of such size may be capable of gender diversity, but less so geographic, linguistic, and other forms of diversity. Thus, there would need to be an understanding of diversity that was not particularly focused on particular states or sub-regions. States have indicated that this should be clarified, with a focus on 'geographical representation' (at least at a regional level), 'balanced representation of gender', 'levels of development', and 'legal systems'. 123 Considerations of, or a mandate for, diversity would be relevant at both the nomination and selection stage. For the former, the greatest risk is that few women would be nominated. States may argue that there are not enough qualified women in the pool, as has been done in the nomination process for other international courts. For the latter, the risk is a lack of sufficient geographic representation when the selection panel considers the nominated candidates.
Overall, the issue of creating a system of adjudication that is at least gender and geographically diverse is challenging. Under the current system of ISDS, achieving diversity is limited by the practice of party autonomy and the prior experience norm. The reform options provide significantly greater opportunity to improve diversity, but each reform option has its own set of limitations. Roster-based reforms may not ensure that parties will select a diverse tribunal, unless this is mandatory. A reform option that would shift to a system of ISDS with all institutional appointments (options 3-6) is most likely to achieve desired diversity, even if none of them will secure full geographic and linguistic diversity. Removing ISDS (option 7) will arguably lead to greater geographic and linguistic diversity (given the number of state respondents and the consequent high number of national courts) with only some effects on gender (given that national courts are not unlike international courts in this regard).

D. Independence
A key driver behind the UNCITRAL reform process has been the perceived lack of independence of arbitrators. 124 This threat to independence may be from external influences, which is the classical definition of independence: 'the freedom of judges to decide disputes free of improper outside influences'. 125 In the case of investment arbitration, concern has been expressed about indirect influence through party appointment, 126 direct influence through interference by parties, 127 or the combined effect produced when arbitrators also act as legal counsel in other cases, through the practice of 'double hatting'. 128 The concern is also about internal influences that threaten the impartiality of adjudicators. Accordingly, arbitrators 'must be relied upon to carry out their public duties independent of venal or ideological considerations'. 129 Empirical research indicates that normative views and geography may affect outcomes. 130 Several reform options address directly independence and impartiality, although there are various trade-offs to consider. First, the international reform options under models 2-6 will all reduce the degree of party influence on adjudication. Roster systems in option 2 will potentially have the least effect, as they only limit the pool of arbitrators, and arbitrators will know which party appointed them. Institutional appointment of arbitrators (model 3) and judicial appointment (4-6) will clearly distance adjudicators from the litigating parties. Model 7 (No ISDS) would have the same effect in relation to use of non-ISDS international arbitration and domestic courts, with two exceptions. 131 State-to-state based arbitration would maintain a model of party appointment and, possibly in some cases, a government may be able to appoint, in a domestic dispute with an investor, judges favourable to its position before an investor's case is heard. Second, the movement to judicial forms of adjudication (models 4-6) will decrease the likelihood of double hatting and increase transparency in appointment processes. This can have the effect of enhancing the appointment of members which are free from non-legal influences.
However, while the influence of the litigating parties may decrease with these reform options, the overall influence of states (in the plural) will increase. This is partly the point of the reform. As was discussed in Section 3D, greater state involvement in selection potentially ensures greater consistency and correctness (legal or political) in awards by enhancing the degree of accountability to the system's principals. Yet it creates a tension within the quadrilemma that is largely unavoidable: 'independence and accountability are features that are in conflict with each other: the more independent judges are the less accountable they will be, and vice versa'. 132 For example, renewable terms may decrease independence even if it enhances accountability and correctness. Notably, this tension also arises with all types of judicial alternatives, whether international (models 4-6) or through the No ISDS approach of accepting the greater involvement of domestic courts (model 7). Thus, there are some clear policy implications for independence and accountability in the choice of the models.
Nonetheless it is arguably possible to develop mechanisms of selection and appointment that would seek to optimize the attainment of both independence and accountability. For example, Larsson, Squatrito, Stiansen and St. John point to the use of screening committees for appointments to various international courts, noting for example on the Court of Justice of the European Union: While causal effects of the panel on judicial independence and performance have not been subject to much empirical scrutiny, it is worth noting that the Panel's reports have led to the rejection of several candidates and some observers posit that it has strengthened domestic appointment procedures in member states. 133 Such mechanisms have been met recently with a favourable response in the UNCITRAL reform process. 134 Thus, any move to stronger judicialization could or should be accompanied by greater attention to the types of influences states may have in the selection and appointment process. In addition, there may be compensating mechanisms, such as providing adjudicators with greater financial autonomy in the operation of the institution, higher thresholds for removal or suspensions of judges, regulation or prohibition of double hatting, and lessening of ability of states to interfere with interpretation post-decision. 135

V . CO NC LU SIO N AN D KE Y T RAD E-OFF S
Proposals for new approaches to the selection and appointment of adjudicators (ISDS) have flourished in the current multilateral reform process at UNCITRAL. However, the concrete design process raises difficult trade-offs for states in seeking to address their key concerns with the existing system of dispute settlement. These trade-offs can be understood as operating within a quadrilemma of accountability, independence, diversity, and procedural fairness-where it is virtually impossible to maximize all four vectors with a single proposal. As Section 4 shows, this quadrilemma provides an opportunity to identify many of the tradeoffs for the UNCITRAL process and the potential need for compensatory mechanisms.
Overall, several key trade-offs tend to be particularly prominent in the UNICTRAL ISDS procedural reform process. First, there is often a tension between procedural fairness and accountability: selection and appointment in more institutionalized and centralized dispute resolution models will most likely reduce costs and duration of proceedings. However, the desire for greater accountability in the form of jurisprudential consistency and legal correctness will naturally point to appellate review. This may mean that the length of proceedings and some costs would not decline significantly. Thus, case management reforms would be essential to any process but to the extent that they do not compromise other aspects of procedural fairness.
Secondly, there is a frequent tension between independence and accountability. A move to less party control over appointment will enhance the independence and impartiality of ISDS adjudicators. The result may be, though, that adjudicators may be less answerable to litigating parties but more answerable to states for their decision-making, threatening independence. Mechanisms to make adjudicators more answerable, for example through renewable terms, may decrease the likelihood of legally correct decisions but make them more politically correct in problematic ways. This trade-off is difficult to resolve, but approaches that combine non-renewable terms with possibility for state influence in other areas or renewable terms with high degrees of financial and interpretive autonomy may be one way to crack that specific nut.
Thirdly, diversity demands can conflict with reforms designed to improve procedural fairness, independence and accountability. The need to ensure greater geographic and gender diversity and variance in professional background would speak for a large number of adjudicators, whether through a pool or on a court. However, this may reduce the efficiency of the system (procedural fairness) and jurisprudential consistency (accountability), the latter through the possibility for diverse outcomes and an over-emphasis on the role of individual states being represented in the adjudicative system. Some scholars have also argued that if the result is less representation of adjudicators with a commercial law background (by addressing other diversity concerns), there is a risk of inefficient proceedings and incorrect outcomes. 136 There may be also a tension with correctness (accountability), but it is less clear: if there are a fewer number of adjudicators, they may be less accountable to individual states but more accountable to all states. The solution to these dilemmas may be ensuring representative-enough models of selection and appointment, for example, through minimum gender quotas, regional representation, and broad definitions of diversity. Moreover, the magnitude of some of these dilemmas can be problematized. For example, enhancing diversity may increase independence through cognitive diversity.
In summary, trade-offs are frequently endemic in institutional design, and ISDS is no exception. This is not to say that some reform options are not preferable to others in light of the concerns and underlying values. It is that ideal solutions are rarely present. Indeed, one might go further and point out the dangers in doggedly pursuing one value at the expense of the others. For instance, as Lon Fuller argued, the simultaneous pursuit of all aspects of the rule of law is unlikely to be feasible. 137 It thus 'becomes necessary to pursue a middle course which involves some impairment' of the relevant values or 'desiderata'. 138 Arato applies this insight to ISDS and the reform goal of consistency, and argues that while it is 'surely a prime virtue, no legal system enshrines it perfectly . . . Nor should any legal order strive for mechanical consistency'. 139 To be sure, identifying the optimal point or area of 'compromise' is challenging. 140 Roberts and Taylor highlight the particular and deep epistemic challenges of being able to predict the effect of any reform choice: ISDS is a complex adaptive system. 141 Nonetheless, identifying the dilemmas clearly-and acknowledging the unlikelihood of maximization of all values-is a first step. Such transparency helps fuel a deeper dialogue and spur follow-up empirical analysis to identify several different 'good enough' mixes-that make sense in the context of investment disputes and seem politically viable in the contingent context of WG III. It also communicates to those actors who, in the future, may evaluate the reforms, providing guidance as to what, at least, was intended.