Askemos

How To Make Contracts Machine Executible

In Askemos we consider only "written" contracts, specifically in machine readable form. We want our virtual actors or avatars to automatically act accordingly. Hence the contract must include an eventually executible code.

In this section we derive the minimum set of conditions a system must meet to be eligible as a contract party. Independent of any particular implementation, we present an abstract, minimal and implied contract. The principle of inalienable privilege.

Standing on the shoulders of giants… On the relationship to J.J. Rousseau's Social contract.

Readers who buy into the rules E. Moglen demands in virtual spaces -independence of avatars from individual contracts and contracts in machine-readable form - (↗here) may skip the philosophy. To readers surprised by the analogy and method we recommend to read B. Schneier's essay on feudal security first. He uses the same analogy to illustrate how technology shapes society today. Assuming that there is an actual advantage in civil societies above "state of nature" we shall apply the recipe to networked security. In essence: create an "autonomous" thing, which can handle contracts.

Historic Remark

I originally discovered a highly similar structure of a "common sense based" global process governed by an initial contract as a logical result when I tried to come up with a security scheme for an autonomous distributed operating system. The structure I named Askemos. Being a computer science guy, I did not get there by the philosophical arguments, I wrote a prototype to test my understanding.

Later my dad traced the idea back to Rousseau. In other words: I read quite a bit of philosophical background literature about contracts and forms of government. Just this best (and oldest) match we have - I did not learn about until I published Askemos.

About The Method

The text here is no attempt to summarize Rousseau in general. Neither will we try to follow in the sense that we actually derive the model from his text.

The concept of the Social contract has been deeply build into the notion of constitutional state ever since. Any model adequate to handle contracting must therefore be able to express this "minimal" scheme.

We summarize Rousseau, focusing on the homogenious structure of the concept of the Social contract and the Askemos system.

It is beyond the scope of this text to follow Rousseau further than the basic concept as outlined here for comparision. Using the notion of abstract concepts enables us to focus on the logic. We'll use "actor" instead of "a persons spirit and rights" and "computer" or even "physical token" where Rousseau relates to physical human beings.

In Rousseau's text we further found the connection of individual rights balanced with the a social contract.

No claim is made how other concepts, let alone consequences are applicable.

The social contract in short:

1. In the state of nature there is no obligation. Without obligations there is no responsibility. Therefore the only property any individual has is self-ownership. But there is no safety for this either.
2. The only difference to the state of society is that each individual closes a first contract. The social contract.
3. From this contract derive rights and obligations wrt. others. However the purpose of the contract is mutual beneficial. Since there where no receiver left to get anything in return, no such contract can transfer said first property. Therefore the contract can only impose reasonable obligations, can be terminated by the individual and the person must gain a benefit from the obligations the others signed.

The philosophical question Rousseau asks is: Why and which obligations rational, free individuals could trade off and which benefits they receive in return from accepting the contract and thus owe obligations (to government)?

We will not discuss here whether or not the state of nature could be preferable. Instead we start from the premise that contracts are the rules of business. We accept "formally/evidentially enabling business" (by basing contracts on prior contracts all the way down ↗The Origin Of The Idea) as a desirable benefit from accepting the initial contract.

The state of nature in technical terms is comparable to an insulated computer running somewhere. While the human case is hypothetical because humans can't exist in isolation, the hypothesis in the technical case is about the question who build the machine and where does the power come from. Questions people rarely ask when using computers.

The state of nature is seen as not too bad after all. It is assumed as initial state and easy to return to (pull the network plug to feel safe).

However we realize that the single computer is all too fallible. It's a physical thing and those break. Worse: the token is controlled by some person(s) (administrator). Therefore confidence into a computer can not exceed confidence in these persons. We trust no trusted third party as much as our self. Hence the token must be owned and controlled by just one person.

Rousseau introduces the general will as the concept to keep an idea of rights and obligations "straight" in the society beyond the fallibility of the person. (We don't trust even our-self under torture. And we don't trust our self that we fully verified that our computer has no backdoor for intruders or viruses to exploit.)

A social contract comes down to a set of statements, which the peer promises to comply with. So far this initial act is unilateral.

This is a single-sided contract one can enter or exit any time. Insofar it's not yet a "real" ↗contract among several parties. And so is verification. No peer will ever consider the actual content of the contract from other peers for it's own decision making. (At least not yet, at this initial state - there's going to be a voluntary exception.)

The intend is to promise defending the "most general interest" all parties share anyway and get something in return.

The general will (common understanding) is controlled from those statements upheld by the peers/tokens/humans obliged to actively support (bear witness of) some process by means derived contracts.

Since the actual "text" of the initial contract in use by the token can change at any time, we need a minimum contract, which is assumed to be understood and accepted by all peers. (For it being simple and obvious; axiomatic in a way) Thereby we assure that even when the person fails to follow the contract the general will is consistent and correctly presents the actors rights and obligations.

1. By matter of fact the token is bound and allowed to temporary rest, suspend or permanently cancel obligations from this contract. Therefore never assume our actors to wholesale transfer their rights (violate grant semantics) even if the token appears to signal consent.
2. No contract ever changes.
3. By reference to common knowledge a language definition for derived contracts is included. The peer will support and execute such contracts under the restrictions 1 and 2.

NB: A token appearing to request a violation the first rule is handled as if where at rest. It's usually considered broken somehow.

Eventually we signed off the promise to behave mostly rational but keep the freedom to fail.

In return we got the least common understanding of leagally binding obligations. The concept of contracts and their limits.

For any given concern a person will be in need to allocate as many tokens witness as felt reasonable. About the selected tokens bearer they must trust to keep promise to the contracted extend in the case at hand.

A nice example: The Debian Social Contract.

A side note wrt. the events in December 2012: Even an agreement to permanently transfer the right to participate in governance would fail to take effect unless the majority of peers actively circumvent the restrictions in the first rule.

A Remark To The Security Concious

Assuming the reader might disregard the lesson learned for any reason, the contract in it's simplicity might appear almost useless. Nevertheless having an operational symbol (i.e. an evidently true statement describing a conceivable operation) for an event which "shall never happen" (or deemed illeagal) is quite useful one when processing communication for consent.

The security concious mind will anyway not continiue with the gain from the contract. It will try to find a possible attack to proof it void. …(TBD: insert best one here. Needs more research.)

Askemos.org comments please period open.