Why the Wreckage of a Failed Treaty Tells You More Than the Treaty Ever Did
Picture yourself on day one of a multilateral negotiating session. You find your nameplate, open the background paper, and notice that the draft already has structure: operative articles, defined terms, a hierarchy of obligations. Someone wrote that before you arrived. The question worth asking, before the first delegation speaks, is who.
The autopsy is always more revealing than the surgery. When a multilateral treaty collapses, the instinct is to read the failure as a story about nationalism, or bad timing, or one difficult delegation. What the negotiating record actually shows, if you're willing to sit with the transcripts and the bracketed text and the quietly buried working papers, is something more structural: which interests the treaty was never architecturally capable of serving, regardless of whether it passed.
Take the Multilateral Agreement on Investment, the MAI, negotiated through the Organisation for Economic Co-operation and Development across most of the 1990s before collapsing in 1998. It is as clean a case study as international relations produces.
The Bracketed Text Nobody Reads
In treaty drafting, square brackets are the technical notation for contested language. A bracketed phrase means at least one delegation has not agreed. In the MAI's leaked drafts, certain brackets appeared and reappeared across years of negotiation: provisions for labour standards, environmental protections, and the right of governments to regulate in the public interest. These weren't late insertions from obstructionist delegations. They were there from early rounds, stubbornly unresolved, because the treaty's core architecture treated each of them as a potential constraint on investor rights, not as a parallel set of legitimate interests deserving equal standing.
That architectural choice was not accidental. The MAI was drafted inside the OECD, a club of wealthy industrialised nations, and its working groups were staffed heavily by trade and finance ministries. Development ministries were peripheral. Civil society groups had no formal seat. Developing countries, which would have been invited to sign the completed text, had no role in writing it. The brackets around public-interest carve-outs were not a negotiating impasse. They were a record of whose concerns were structurally subordinate from the first session.
This pattern recurs across failed multilateral instruments. The negotiating history of the proposed Plastics Treaty, which stalled through multiple sessions of an intergovernmental negotiating committee without reaching agreement, shows a similar geology: producer-state delegations fought to restrict the treaty's scope to waste management and end-of-life disposal, while small island nations and public health advocates pushed for upstream production limits. The gap was not bridged because the two positions were not symmetric disagreements. One concerned protecting a revenue stream; the other concerned basic environmental survival. A treaty architecture that treats those as equivalent bargaining chips will not resolve them. It will collapse, or it will pass in a form that one side reads as victory and the other reads as decoration.
What the Seating Chart Tells You
The physical and procedural organisation of multilateral negotiations is not neutral. Consider two delegations arriving at the same negotiating session: a mid-sized OECD country with a legal team of twelve and a full-time treaty secretariat back home, and a small Pacific island state with two generalist diplomats covering five concurrent negotiating tracks. Both delegations have one vote. Both appear in the participation list. Neither fact captures the asymmetry in drafting capacity, in the ability to propose and respond to text, in the sheer endurance required to sit through six two-week sessions over three years.
Take two hypothetical delegations to a fisheries treaty negotiation: one from a coastal industrial nation with a fleet-lobbying apparatus and a dedicated oceans ministry, the other from a landlocked developing state whose fisheries interest is confined to a single shared river basin. The industrial nation's concerns are baked into the baseline text from session one, because its experts helped draft the background papers. The landlocked state's concerns appear as a proposed amendment in session four, when the core architecture is already fixed. By session six, the amendment has been folded into a non-binding annex. The treaty fails anyway, but if it had passed, one delegation's interests would have been structural and the other's decorative.
This is not a cynical observation. It is a mechanical one. Drafting capacity compounds. Early text becomes the baseline against which all later proposals are measured as additions or subtractions, and the political cost of subtracting from an established baseline is always higher than the cost of never including something in the first place. The delegation that arrives at session one having already shaped the zero text has, in effect, pre-spent its negotiating capital at a discount nobody else was offered.
What People Get Wrong About Failure
The common reading of a collapsed treaty is that it represents a failure of political will, or that the parties were simply too far apart. Sometimes that's accurate. It misses a more uncomfortable possibility, though: that the treaty failed precisely because it was exposed.
The MAI's collapse came not from governmental opposition but from a leaked draft that reached civil society organisations in France, who publicised the bracketed labour and environmental provisions. The failure, in other words, was a transparency event. The negotiating record, once visible, made the structural subordination of certain interests impossible to sustain politically. That's a different kind of failure. It's a treaty that worked as designed until the design became public.
Ask yourself this: if the brackets had stayed hidden and the MAI had passed, would we be describing it today as a success of multilateralism or as a slow-motion expropriation of regulatory sovereignty dressed in the language of liberalisation?
The lesson for reading any failed multilateral instrument is to ask not just what the parties disagreed about at the end, but what was never in genuine contention at the beginning. Who drafted the zero text? Which forum hosted the negotiations, and whose members dominate that forum? Which interests appear in the operative articles and which appear in the preamble, that most decorative of legal spaces, where aspirations are acknowledged and obligations are not created? A preamble mention is the diplomatic equivalent of a get-well card: warm in sentiment, weightless in effect.
The Archive as Argument
Negotiating histories are not neutral records. They are, themselves, arguments about legitimacy. When a treaty fails and its drafting record becomes available, what you're reading is not just a sequence of positions but a map of whose assumptions about the world were treated as the starting coordinates and whose were treated as deviations requiring justification.
The MAI's working papers assumed that investor protection was the primary purpose and that regulatory sovereignty was a potential carve-out to be narrowly defined. You can hold that view, and serious economists do, but it is a view, not a natural law. The fact that it appeared as the architectural foundation, while alternative framings were positioned as proposals to be argued for and defended, tells you something the treaty's formal text never would have. That gap between foundation and position is where the real negotiation happened, before most delegations were in the room.
Failed treaties are genuinely useful objects, then. Not as cautionary tales about diplomatic process, but as fossils: the negotiating record preserves, in amber, the power geometry of the moment when the text was being written. The brackets show you where the pressure was. The preamble shows you what was conceded without being conceded. The seating chart shows you who had the stamina to stay.
If you want to understand whose interests a future treaty will actually serve, don't wait for the signing ceremony. The first draft already has the answer. Everything after that is editing.