Mediation and Arbitration: Who Decides When We Cannot Agree?

Mediation and arbitration solve different problems

In everyday language, mediation, arbitration, settlement and adjudication are often collapsed into one category: “bring in a third person”. The distinction is fundamental. Under the UNCITRAL model, a mediator assists the parties in attempting an amicable settlement and has no authority to impose a solution on them. Arbitration is adjudicative: the parties submit a dispute to an arbitrator or tribunal that decides it within the scope of the arbitration agreement.

Mediation is therefore not merely “softer arbitration”, and arbitration is not simply “stronger mediation”. In mediation the parties generally retain control over the substance of the final agreement. In arbitration they transfer part of that control in advance: they agree that a third party may decide a defined dispute even if one side later dislikes the outcome.

The first question is functional: do we need help reaching agreement, or do we need a decision because agreement is no longer possible? If the parties can still construct a mutually acceptable outcome, mediation may fit. If a dispute blocks shared infrastructure, contractual performance, repayment of a substantial sum or another matter that ultimately requires a determination of what applies, a binding route — arbitration or a competent public process — may be necessary.

The worst moment to design dispute rules is when both sides already believe they are right. Each will naturally prefer the forum, timing, evidentiary standard or decision-maker that looks more favourable. “How Can Communities Cooperate Without a Single Center?” therefore required inter-community agreements to define escalation in advance: technical clarification, negotiation, mediation and only then, where appropriate, arbitration or another independent process.

Imagine two communities jointly financing a wastewater treatment plant. A dispute may concern measurements, costs, delay, service quality or interpretation of the agreement. If the contract merely says “disputes shall be resolved amicably”, it works while amicable resolution works. Once it fails, the clause says almost nothing. A better agreement determines in advance when negotiation is considered exhausted, who can trigger the next stage and which disputes can ultimately receive a binding decision.

That is not an expression of distrust. It prevents trust from depending on the assumption that conflict will never occur. The parties choose the process while they still do not know which of them may later be claimant, debtor, harmed party or loser.

Mediation: control over the outcome remains with the parties

Mediation makes sense where there is room for a solution the parties can shape themselves. EU Directive 2008/52/EC describes cross-border civil and commercial mediation as voluntary in an important sense: the parties remain in charge of the process and may terminate it. National law may in some circumstances require an attempt at mediation or attach incentives to it, but this must not prevent access to the judicial system.

That distinguishes an obligation to attempt a process from an obligation to accept a solution. A mediator may help clarify interests, break a package of demands into smaller issues, test options for feasibility and record an agreement. The mediator should not convert personal preference into a ruling. The UNCITRAL mediation model expressly preserves party control over the process and outcome.

“After Harm: Restitution, Protection and the Limits of Punishment” adds another boundary: mediation after serious harm is not automatically appropriate. Where there is violence, coercive control, intimidation or such a severe imbalance of power that consent is not genuinely free, an apparent “agreement” can merely disguise coercion. Protection takes priority over the idea that every pair of parties must sit at the same table.

If mediation succeeds, another question follows: what legal effect will the settlement have? In the EU, national systems can under the Directive provide routes for the content of a written mediated agreement to be made enforceable. For certain international commercial mediated settlements, the Singapore Convention creates an additional enforcement framework in States Parties. Mediation therefore need not end as an informal promise only.

Arbitration: consent partly moves to an earlier moment

Arbitration works differently. The UNCITRAL Model Law defines an arbitration agreement as an agreement by the parties to submit all or certain disputes that have arisen or may arise in respect of a defined legal relationship. The central consent is therefore consent to the forum and mode of decision, not later consent to every individual award.

This explains the apparent paradox: arbitration can be voluntarily chosen and still binding. If the arbitration agreement is valid, a party generally cannot exit merely because a dispute has now arisen or because it suspects it may lose. The Model Law also allows an arbitral tribunal to rule on its own jurisdiction, while courts retain limited supporting and supervisory roles.

But “the parties once agreed” does not end the inquiry. The validity of the arbitration clause, the actual scope of authority, an opportunity to present the case, arbitrator independence, arbitrability and public policy still matter. Several of these safeguards belong in “Fair Procedure Without an Assumed Monopoly”; this article establishes only the institutional boundary: an arbitrator may decide only because, and only to the extent that, a valid basis for that authority exists.

Forum choice is part of the agreement, not a technical footnote

The sentence “all disputes go to arbitration” is often not enough. A serious agreement should consider which disputes are covered, whether mediation is required first, who appoints the arbitrator, which rules apply, the seat of arbitration, the language, the law governing the substance of the relationship and the allocation of costs. UNCITRAL's Notes on Organizing Arbitral Proceedings specifically identify rules, language, place or seat, confidentiality, hearings and evidence among matters that benefit from timely attention.

The seat of arbitration is not merely the address of the room where participants sit. It is ordinarily the legal anchor connecting the proceeding to an arbitration law and to courts with supporting or supervisory functions. A physical hearing may occur elsewhere or remotely without necessarily changing that legal seat.

An inter-community agreement should also provide a fallback appointment mechanism. If each side is meant to appoint one arbitrator, what happens when one refuses? Who appoints a neutral chair if the first two cannot agree? A process that works only while both sides cooperate perfectly is not yet a process designed for a serious dispute.

Enforcement shows that a private forum does not live in a vacuum

If both sides voluntarily perform a settlement or award, coercive enforcement never arises. This article must address the harder case: one side loses and refuses to comply. This is where private dispute resolution often needs a bridge to a wider legal order.

The 1958 New York Convention is the central international framework for recognition of arbitration agreements and recognition and enforcement of foreign and non-domestic arbitral awards in Contracting States. The UNCITRAL Model Law similarly provides that an arbitral award is to be recognized as binding and may be submitted for enforcement. This is a major reason arbitration is useful in cross-border commercial disputes.

Enforcement is not an automatic magic seal. Recognition or enforcement may be refused in defined circumstances, including problems with the arbitration agreement, inability to present a case, decisions beyond the submitted mandate, procedural defects, non-arbitrability or public policy. The Singapore Convention likewise contains limited grounds for refusing relief on mediated settlements.

A decentralized forum may decide without one central adjudicator; it does not follow that it can always enforce without a wider legal environment.

Between communities, escalation is preferable to a new permanent ruler

“How Can Communities Cooperate Without a Single Center?” showed that communities can cooperate through agreements, consortia, federal bodies and polycentric networks without one superior pyramid. This article adds dispute resolution. Two or more communities can specify in their founding agreement that a dispute first receives technical clarification, then negotiation, then mediation, and only at the end — for predefined categories — independent arbitration.

The neutral third party need not be a permanent higher authority. It can be an independent expert, a panel selected from an agreed roster, an arbitral institution or an ad hoc tribunal created only for the particular dispute. The crucial point is that its jurisdiction comes from a clear agreement rather than a blank authorization to decide “whatever is necessary”.

“How Can Communities Cooperate Without a Single Center?” already separated a dispute about performance from a dispute about mandate. That distinction is decisive here. An arbitrator may determine whether an operator breached an agreed standard or how an existing contractual provision should be interpreted. It is far more problematic if an arbitrator uses an ambiguous clause to create a new political competence for a joint body that the member communities never granted. Some questions therefore need to return to the members, require an amendment or demand a new mandate — not a more creative adjudicator.

Contractual arbitration is not automatically free or fair

The word “voluntary” can conceal very different realities. Two roughly equal organizations that negotiate and choose a neutral forum are not in the same position as an individual who can obtain an essential service only by accepting a long standard-form contract requiring an expensive arbitral forum on the other side of the world.

EU consumer law therefore does not treat every arbitration clause as untouchable merely because it appears in a contract. Directive 93/13/EEC identifies as potentially unfair terms those that exclude or hinder a consumer's right to take legal action or exercise another legal remedy, particularly certain requirements to take disputes exclusively to arbitration. In Asturcom, the Court of Justice confirmed that in defined circumstances the unfairness of an arbitration clause must be capable of examination even at the enforcement stage.

This does not make arbitration inherently unfair. It means that contract paper is not proof of genuine balance of power. Forum design should examine cost, distance, language, participation in arbitrator appointment, access to information and whether the stronger party repeatedly selects the forum while the weaker party cannot realistically refuse it.

Nor can every dispute validly be privatized. Arbitrability depends on applicable law; some matters may be reserved for courts or other public authorities, and public policy remains a limit on recognition and enforcement. THY-REALITY therefore does not confuse plurality of forums with the idea that every matter can be converted into a private bargain.

Independence belongs to both the person and the design of the forum

A third party is not independent merely because the parties do not know them personally. UNCITRAL models for mediators and arbitrators require attention to impartiality and independence and disclosure of circumstances that may create justifiable doubts. These can include personal, professional and financial relationships, prior work for one party or another interest in the outcome.

Conflicts can also be structural. If one party always chooses the institution, supplies most of its business or repeatedly appears before the same small pool of decision-makers while the other side has no comparable role, it is not enough to say that a particular arbitrator acts honestly. A sound forum reduces dependence through the way the third party is appointed, required to disclose conflicts and, where necessary, replaced.

This article stops here. Exactly how equality of the parties, the opportunity to present a case, evidence, reasons, challenge, appeal or judicial review should be protected is the next problem. That is the core of “Fair Procedure Without an Assumed Monopoly” on fair procedure.

Practical audit of a dispute-resolution clause

Before a community, cooperative, consortium or other organization signs an important agreement, its dispute clause can be tested with these questions:

1. Which disputes are covered? Contractual performance only, or other legal disputes as well?
2. When does the formal path begin? How much time is allowed for direct negotiation or technical clarification?
3. Is mediation mandatory as a step, optional or inappropriate? Who can end it, and when?
4. Which disputes proceed to binding arbitration? Are some matters reserved for a court, the member communities or another authority?
5. How is the mediator or arbitrator selected? What happens if one party blocks appointment?
6. How are conflicts of interest disclosed and addressed? Who can seek replacement of the third party?
7. Which rules, seat, language and governing law apply? Are they understandable and workable for both sides?
8. Are the costs accessible to both sides? Can the price of the process itself become leverage?
9. How are people, property, evidence or a shared service protected while the dispute is pending? Who may seek interim measures?
10. How will the outcome become enforceable? Is a mediated settlement simply a contract, will it be formalized, or is the outcome an arbitral award?
11. Where does the mandate end? May the forum interpret the agreement while lacking authority to create new shared powers without new consent?
12. What review remains available? Which minimum procedural safeguards cannot validly be waived?

A contract without answers to all of these questions is not necessarily useless. But it means that part of the rules will be designed only after a dispute has arisen — exactly when agreement on the rules becomes most difficult.

This article is not searching for a new universal judge. Its aim is narrower: show how people and communities can create multiple levels of dispute resolution without turning every conflict into political war or dissolution of cooperation. Mediation leaves the solution with the parties. Arbitration can provide a binding answer where such authority was validly agreed. Public legal systems remain relevant to enforcement, non-arbitrable matters, public policy and supervision of basic procedural boundaries.

This fits the THY-REALITY axis: distributed power does not mean absence of rules, while rules do not require one institution to possess the answer to every dispute in advance. A healthier arrangement matches the forum to the problem, limits its mandate beforehand and prevents a temporary adjudicator from becoming a new permanent ruler.

When we cannot agree, the question is not only who will decide. A more just question is: who may decide this particular dispute, on the basis of which consent, through which procedure and within what limit?

“Fair Procedure Without an Assumed Monopoly” will therefore take up the next task: even if the forum has been chosen correctly, what procedure must it follow so that its decision is not merely binding, but reached fairly?