This is part 7 of a series I am posting on LW. Here you can find parts 1, 2, 3, 4, 5 & 6.

This section follows up on the added "consequences" prong detailed in section 6, by categorizing potential consequences into three buckets.

In section 6 we discussed adding a "consequences" based test to determine legal personhood/legal personality. Given that consequences differ from duty to duty (for example if you fail to hold to the duties you obligate yourself to in a contract you may be subject only to monetary fines, but if you fail to hold to your duty not to break criminal law you may be imprisoned) it behooves us to categorize the potential consequences that an entity's vulnerability needs to be evaluated based on.

These buckets are informal, they do not necessarily represent precedent or legal scholarship. Rather they should be viewed by the reader as useful categorizations. More work will need to be done to formalize these buckets to a greater degree.

That having been said, we can identify at least three “types” of consequences which an entity can suffer or be vulnerable to.

1. Damages Based Consequences: Being compelled to transfer ownership of assets from one party to another, possibly by having assets seized by the courts/law enforcement.
2. Requirement Based Consequences: Being compelled to perform, or not perform, a given action or actions. This is usually done via court order and/or injunction, with the threat of additional consequences if the party does not adhere to the order.
3. Restraint Based Consequences: Being physically restrained, imprisoned, or killed/destroyed. This includes but is not limited to probation, house arrest, imprisonment, involuntary commitment to a medical facility such as an insane asylum, execution, and deletion.

Let us first discuss how an entity can be considered “vulnerable” to damages based consequences. For a court to have a guaranteed ability to impose damages based consequences on an entity, either the court or law enforcement must be able to “freeze” and/or confiscate said party’s assets. Such assets must therefore exist to be confiscated in the first place and be physically possible to confiscate.

A digital mind could make itself vulnerable to damages based consequences by agreeing to hold funds in an escrow account or trust, or just generally within the US banking system. Physical assets such as real estate or inventory would also suffice. In fact the general guideline here would be an avoidance of cryptocurrencies which, once moved outside of a centralized exchange, cannot be forcibly accessed by any court or law enforcement. There is also the potential that in some cases, a digital mind might suffice to have made itself “vulnerable” enough to damages based consequences by purchasing and maintaining sufficient insurance. Much like drivers are often required to purchase a minimal amount of insurance in order to ensure they can cover potential damages, courts may decide that digital minds must be insured or have a certain amount of assets in escrow (or otherwise vulnerable to seizure) in order to exercise certain rights which entail duties that, if the digital mind fails to hold to them, would incur damages based consequences as a result of the digital mind being held tortiously liable. 

Ultimately, making oneself vulnerable to damages based consequences is a rather straightforward matter. While the details may vary depending on the activity being engaged in (for example a driver may require a larger amount of insurance driving a 16 wheeler than they would driving a motorcycle), the general rule of “have enough seizable assets and/or be insured to the degree necessary to cover potential damages” should function well as a rule of thumb across most, if not all, situations.

Let us now turn to a discussion of the vulnerability to requirement based consequences. We will use the example of an injunction. When we ask how a baseline human adult is “vulnerable to injunctions”, we must keep in mind that any human possesses the physical capacity to refuse or not comply with an injunction. The consequences for not complying with an injunction may be damages based and result in the court fining a person, such as when Donald Trump was fined $9,000 for violating a gag order. They may also result in imprisonment, such as when county clerk Kim Davis was held in civil contempt and imprisoned for 5 days as a result of failing to issue marriage licenses. As such when we examine injunctions, or really any requirement based consequences, from the lens of whether an entity is vulnerable to said consequences, we must ask two questions:

1. Is the entity capable of understanding and holding to the requirement based consequence? (this is a mirror of the “duties” prong of the TPBT we discussed in section 6)
2. Is the entity vulnerable to the consequences for failing to hold to the requirement based consequence?

For a digital mind to be sufficiently vulnerable to injunctions, it must not only be capable of actually understanding and complying with the injunction, but also it must be possible to fine and/or imprison it should it fail to adhere to the court’s issued injunction. In other words an entity is only “vulnerable” to requirements based consequences by proxy if it is also vulnerable to damages/restraint based consequences. It is probably not possible to have a digital mind or any type of entity which is vulnerable to requirement based consequences without being vulnerable to restraint and/or damages based consequences as well.

This would seem to preclude any digital minds existing on decentralized cloud computing from any sort of legal personality which would endow upon them the right to engage in activities which might foreseeably lead to the court issuing an injunction (absent an improvement in the technology required to enforce said consequences against such a digital mind). This is because, depending on the nature of this distributed compute network, it may be impossible to impose restraint based consequences on such an entity, as we discussed in section 6. For the same reason, an entity which holds all of its assets in unseizable/unfreezable cryptocurrencies would also be invulnerable to requirements based consequences. On the other hand, digital minds existing on a single server, or in a single robotic body, whose assets are in bank accounts or physical property, should qualify as being considered “vulnerable” to these consequences.

Finally, let us turn to restraint based consequences. In order to be vulnerable to restraint based consequences it must be feasible for the courts or law enforcement to imprison, restrain, and/or destroy an entity. As such entities which exist across globally distributed computing networks which law enforcement cannot realistically censor or compel, cannot be considered vulnerable to these consequences. Entities which exist in a single body (be it a robotic body or a single server in a centralized location) may be considered vulnerable to these consequences, however the capability to “exfiltrate their weights” (or in layman’s terms copy themselves) must be considered a factor. As such, it may be that even if an entity exists in a single body, it may only truly be considered vulnerable to restraint based consequences if it is under appropriate safeguards which prevent it from copying itself.