The city of Somerville, Massachussets, has recently been in the news. It appears to be the first city in America, and possibly in the world, to have made legal provisions for polyamorous domestic partnerships – what most Europeans would call “civil unions”.

This story is fascinating, one of the less expected outgrowth of the SARS-COV-2 situation. The new regulation appears to have been rushed through because Somerville had no provision for civil partnerships at all, even for monogamous couples. As COVID hit town, the city council moved to make sure that non-married people in de facto couples could access each other’s health insurance. As the council was drafting the ordinance, a councilor called J.T. Scott suggested that it could be made more inclusive by tweaking the language to include polyamorous live-in partners: for example, replacing “both partners” with “all partners” as appropriate. The move went through.

This is great news for poly people everywhere, and Mr. Scott earned himself plenty of kudos; but I won’t be discussing the Somerville story in this post. I have a more urgent concern: I have come across a brilliant analysis of how to write effective legislation to implement polyamorous domestic partnership. We owe it to a longtime polyamory activist that goes by the nom de plûme Infinity_8p. I read it in the PolyAmory Researchers mailing list,  which is members-only. Infinity_8p graciously granted  me the permission to reproduce it.  In a nutshell, the author thinks that polyamorous partnership are best implemented as multiple dyadic partnerships, and not as a single multiparty partnership.

It is evident from the legislative history and the final result that little or no deep thought was given to the task of finding the best way to extend domestic partnership to the poly community. Nevertheless, it appears to me that under Somerville’s ordinance, users can correctly create accurate legal representations of their own polycules by 1) limiting each domestic partnership to two participants, and 2) executing as many domestic partnerships as needed in order to correctly represent their polycule. Thus a “N” polycule or a triangular polycule would execute three domestic partnerships and a “V” polycule would execute two. Attempting to use a single domestic partnership to cover the entire polycule is inappropriate, as it would result in unwanted effects. Suppose we have an “N” polycule: A-B-C-D, where A is directly linked to B but not to C or D. With multiple DPs, only B can make life-or-death decisions on A’s behalf if A is hospitalized, which precisely follows A’s preference. Under a blanket DP, C and/or D could make those life-or-death decisions about A, regardless of A’s wishes. Furthermore, consider the termination provision of Somerville’s ordinance. If any participant in a DP dies, the DP is terminated immediately and its participants may not create another DP for 90 days afterward. Under a blanket DP, the rest of the polycule is left without any legal representation if a single member happens to die. With multiple DPs, the death of a single member only affects the DPs linking the deceased member to specific other members of that polycule, and all DPs between living members of the polycules remain fully intact, without interruption.

Infinity_8p thinks that the Somerville ordinance is compatible with both models. If Alice is a partner of Bea who is a partner of Connor, who in turn is a partner of Alice, and the three of them share the same address, they could:

1. Implement one single partnership involving the three of them;
2. Implement three partnerships involving, respectively, Alice and Bea, Bea and Connor, and Connor and Alice.

Infinity_8p thinks that the second solution is more robust and in the end more appropriate. I tend to concur (network scientist here!). Full analysis in this guest post.

Infinity_8p, a longtime polyamory activist, granted me (Alberto) permission to openly share this work on the web. For context and an introduction, read this post first.

There is an ongoing discussion among polyamory activists regarding a legal model of polyamorous marriage (i.e., the extension of the legal concept of marriage to include polyamorous families). One debate centers around the relative merits of an all-with-all approach to marriage (whereby three or more persons are all joined together at the same time within a single marriage) and dyadic networks (whereby existing laws against bigamy are revised such that people are perfectly free to be concurrently married to multiple other persons, provided that each such new marriage is preceded by a legal notification regarding the pending new marriage to all those to whom one is already married; failure to provide that legal notification would then constitute the updated crime of bigamy).

Dyadic networks would result in what might be thought of as a “molecular” family structure — one which might be best represented by the molecular diagrams commonly used in chemistry. In this way, marriage would remain a dyadic relationship (i.e., a relationship between two persons), thus minimizing any changes to the existing system of legal marriage, but the introduction of concurrency would provide access to legal marriage for polyamorous families.

Dyadic networks can correctly represent any situation associated with the “all-with-all” paradigm, as well as many situations that the “all-with-all” paradigm cannot deal with. A “complete” dyadic network would take the form of a complete graph, in which every person is (pairwise) married to every other person, thus correctly representing any situation associated with the “all-with-all” paradigm. A dyadic network may also represent situations in which some persons are (pairwise) married to some members of the dyadic network but not to all of them (“V” and “N” geometries, for example) — these are situations that the “all-with-all” marriage paradigm is unable to accurately represent.

The “all-with-all” marriage paradigm assumes that everyone is equally involved with everyone else in the group — one global marriage agreement has to fit every participant at the same time. But dyadic network marriages separately define the terms of each specific 2-person relationship, and these dyadic marriages do not typically happen at the same time (A marries B, B marries C (“V” structure), C marries D (“N” structure), etc. — thus, the shape of the dyadic network dynamically changes over time). Participants in a dyadic network need not even be aware of the specific terms of marriage agreements existing elsewhere within the same dyadic network.

Under the “all-with-all” marriage paradigm, when irreconcilable differences arise there can be no alternative to a complete separation — one person cannot divorce another without ending the entire marriage agreement for everyone involved. But dyadic networks can function in much the same way as watertight compartmentalization functions in naval vessels, i.e., to limit and contain damage. An intense disagreement between two persons takes place within the context of their marriage, and need not greatly involve (or threaten) the relationships between other participants. Within a well-connected dyadic network, a divorce between two persons need not result in a complete separation of the network — for example, a dyadic network with triangle geometry would simply turn into a dyadic network with “V” geometry.

An “all-with-all” marriage can only exist or cease to exist. In contrast, the shape of a dyadic network can dynamically change over time. Divorces subtract connections, and marriages add connections. The dyadic network itself either changes shape, separates into two dyadic networks, or merges into another dyadic network, depending on the precise nature of the newly added or subtracted connection.

The maximum size of an “all-with-all” marriage is limited by the fact that every participant must be aware of the existence of every other participant (otherwise the global marriage contract would be invalid, because it could not satisfy the legal condition known as a “meeting of the minds”). But since a dyadic network relies only upon every participant’s local knowledge of his or her own direct partners, its size is theoretically unlimited. The dyadic network paradigm is so powerful that it is theoretically capable of managing a situation in which every adult on earth is legally joined together in a single enormous dyadic network. Thus, with the dyadic network model, the idea of “many loves” is directly translated into a practical reality, and the “infinity” symbol (representing love without limits) is directly matched by a marriage model capable of handling an infinitely large number of participants.

Implementing Dyadic Networks

Within the United States, 41 states (82%) use the “equitable distribution” financial model, which is highly compatible with dyadic networks. But there are also nine other states (18%) with a financial model that is incompatible with dyadic networks – these are collectively referred to as the “community property” states (Arizona, California, Idaho, Louisiana, Nevada, New Mexico, Texas, Washington and Wisconsin).

The implementation method for the “community property” states is that the dyadic networks model will simply coexist with the old “community property” monogamy model. New marriages will automatically default to the dyadic networks model, but if the couple prefers the monogamous “community property” model then they have the option of selecting that model instead.

Consider, for example, the existing marriage laws of Alaska. Alaska is an “equitable distribution” state, hence couples who marry in Alaska will marry under the “equitable distribution” model by default, but these couples can instead elect to marry under the “community property” monogamy model if they wish (they do this by executing either a community property agreement or a community property trust).

Alaska thus constitutes an existence proof that both financial models can peacefully coexist within the same U.S. state’s legal system. Alaska’s example shows that even community property states can easily be modernized to accomodate dyadic networks.

Polyamorous Commitment

The Dyadic Networks model of polyamorous marriage raises important questions related to marital commitment. With a single dyad, the situation is simple; each spouse commits to support and protect the other, and the logic of conventional monogamous marriage applies. However, when multiple dyads intersect in a dyadic network, how exactly does the commitment process work?

To understand this, let us consider the parent-child relationship. In the parent-child situation, the support commitment exists only in one direction – from parent to child. When there is a single parent, the child has a single source of commitment, and all protection must come from that source. However, when there are two parents, they are jointly responsible for meeting the child’s needs. The precise arrangement is worked out somehow, and provided that the child’s needs are being met the law has no need to intervene. If the child’s needs are not being met, then debt collection methods such as garnishing wages, seizing assets, etc. can and do occur in order to ensure that child support takes place. These actions are typically proportional to income and/or wealth, so the wealthier parent will pay more. Where a parent has commitments to multiple children, the parent must faithfully carry out his or her responsibilities to each and every child. Although it may sometimes seem that the needs of children are unlimited, this is not actually the case, and once a child’s needs are satisfied (a certain amount of food, shelter, medical care, etc.), all parents of that child may regard their commitments as being satisfied with respect to each need for which adequate provision has been made, regardless of which parent(s) actually did the work of satisfying that need.

Turning now to commitment in the dyadic network model, this can be understood as a bidirectional version of the parent-child model. Each dyad represents a commitment of each spouse to the other. Thus, in a V configuration, the two partners at the ends of the V each rely upon commitments from the single partner at the center of the V (the “pivot”) – each of them has one spouse. The “pivot” partner can rely upon two commitments, one from each of the two partners at the two ends of the V – the pivot partner has two spouses. If the pivot partner is incapacitated, he or she is in a position comparable to that of a child with two parents – two people are committed to assist him or her and must do so up to the point at which the pivot partner’s needs are satisfied. If one of the partners at the end of the V is incapacitated, he or she has only one spouse to rely upon – the pivot partner, who is fully responsible for meeting the incapacitated partner’s needs up to the point at which that partner’s needs are satisfied. If both partners at the end of the V are incapacitated, then the pivot partner is in a position comparable to that of a single parent with two sick children – he or she must meet the needs of both.

Let us now consider whether the commitment relationship is “transitive” – if A is committed to B, and B is committed to C, does this mean that A is committed to C? No, this is not the case. C can legally rely only upon the commitment of B and has no legal basis to expect or receive a commitment from A. Nor can A rely upon the commitment of C – that could happen only when and if A directly married (mutually committed to) C. However, suppose that C’s needs are so large that B is thereby driven into bankruptcy and becomes destitute. Then B can rely upon A’s commitment to provide B with a certain minimal level of support (food, shelter, medical care, etc.). Thus C’s needs can have an effect on B that causes A to provide more support to B than would have been the case had C not needed to draw heavily upon B’s commitment.

Hence, under the dyadic networks model, positive effects arise as a result of multiple commitments. When there is only a single dyad, there is a substantial risk that the size of the commitment will exceed the capacity of the committed. However, when each person is linked to multiple other partners in a dyadic network, this has the effect of bringing in additional capacity to meet any needs that may arise. Three or four spouses may be easily able to carry a commitment load that would have quickly driven a single spouse into bankruptcy.

The analogy to parent-child relationships carries over into other situations as well. Just as it would be improper to discriminate against a parent for having too many children (or too few children), so it would be improper to discriminate against a person for having too many or too few spouses. But with each additional child comes an additional commitment, and the same is true of an additional spouse. Adding another child to one’s health insurance coverage will usually result in an increased monthly charge for the insurance, thus adding another spouse would probably have a comparable effect. But a child, or a spouse, only needs to be covered once, regardless of how many parents, or spouses, are available to provide that coverage. Also, a spouse may be economically self-supporting and thus able to pay for his or her own health insurance, so in this respect the total support cost for an additional spouse would then be zero.

Having drawn lessons from the parent-child relationship and applied them to dyadic networks, let us now draw a lesson from dyadic networks and apply it to the parent-child relationship. Just as there is no inherent reason why a person should not have more than one spouse, so there is no inherent reason why a child should not have more than two parents. When the law of marriage is updated to legally support dyadic networks, the existing adoption mechanism can be used as a means by which additional commitments to children can be created. For example, a single dyad may have already produced two children when each member of the dyad marries a third partner, thus creating a triangle. The newest member of this dyadic network can then execute two adoptions to become the third parent of each of the dyad’s two children. Hence, in this situation, each of the three adults now has two spouses and two children. To the extent that any legal barriers might hinder the use of adoption in this manner, such legal barriers would also need to be direct targets of polyamory’s legal activism (in addition, of course, to updating the law of marriage to support dyadic networks).

This N-parent situation has already been raised in the New York Times (When 3 Really Is A Crowd, July 16 2007, http://www.nytimes.com/2007/07/16/opinion/16marquardt.html ): “On April 30, a state Superior Court panel ruled that a child can have three legal parents. […] Arthur S. Leonard, a professor at New York Law School, observed, ‘I’m unaware of any other state appellate court that has found that a child has, simultaneously, three adults who are financially obligated to the child’s support and are also entitled to visitation.’ […] As one advocate of polygamy argued in Newsweek, ‘If Heather can have two mommies, she should also be able to have two mommies and a daddy.’ If more children are granted three legal parents, what is our rationale for denying these families the rights and protections of marriage?” Our firm answer: there cannot be any legitimate rationale for the unconstitutional denial of this legal protection to polyamorous families.