Bitlaw a legal system with a DSL for Cardano governance

Continuing off my discussions in the thread on a voice from the community on the Cardano Foundation board where I posted my discussion with Bertalan @bercinho I want to discuss with the community and learn together with the community more on the possibilities and perhaps also spark some interest into developing a DSL for governance on Cardano. We have Marlow that is trying to be a DSL for financial contracts but we do not have a governance DSL.

Here is my previous post:
There is much research on DSL for law. I think a great name for a bitlaw DSL would be to be named after Hammurabi (edit already taken https://github.com/mpoulshock/HammurabiProject so Ur-Nammu it is ! ) or after the first known existing law code called Ur-Nammu he created. https://en.wikipedia.org/wiki/Code_of_Ur-Nammu
The premise of the law is IF THEN statements.

Eystein_Hansen:

Bitlaw
I want to share one of my thoughts with you all on future governance and on legal issues. I call it bitlaw. Unlike common law or civil law systems of today it creates some new efficiencies that can only be found when you digitize not only the representation of law but the interpretation of law.

In a common or civil law system you the user of such rules are required to figure out when does this rules apply to me or you are enforced to make these rules apply to you by a great state apparatus costing society huge sums.

In a bitlaw system assuming it is dualistic (coexisting with common or civil law) many fields of law can be queried by the computer itself. For example in contract law there is many cases where users can simply agree on standards or what rules they will enforce onto contracts. Some laws are more absolute (most contracts in modern ages for example ban any form of agreement for slavery) and will still need interpretation. However many legal agreements and laws are straightforward yes or no or can be agreed upon to be straightforward yes or no and can then fit into a binary system of quarry. Also suddenly it can be computed and visualized for a user when a law apply to a said condition and it can be automatically enforced if users agree to this. Obviously this has many ethical issues but if done well I am sure it can be a good thing for society.

Tip of the iceberg law cases or database law cases?
Suddenly instead of having a professional class who interprets and theorize about law outcomes you can in many cases create databases of knowledge through millions or billions of contracts for filled. This is a far cry from tip of the iceberg type of law interpretations where only a small amount of cases ever reach the courts or other interpretive authorities.

A law for everything with AI?
It could possible also open an age old question if it is possible for a law for everything. Many have tried this historically and failed because it was simply not possible and out of this emerged the current law theories of interpreting laws based on an agreed upon system or standard and for even high courts to fill in the blanks where there was a legal void. But what if AI could in the future calculate all outcomes or even calculate on the fly outcomes. Ethically this still should be controlled by humans but the efficiency and scale of this would probably allow such a new way of legal thinking.

Layered law approach
Another aspect would be to have local laws vs universal laws on any such system. We know from human history it is very hard to agree to everything with everyone. So it is far better to agree on what we can agree with whom we can agree. With bitlaw suddenly it is far easier to organize such layered approaches where you can instantly say that only portion X of rule Y applies for this transaction or this contract Z. This seems a better approach than trying to make universal laws for every single subject and to every single user. Obviously still some rules needs to be national or even transnational like human rights but a layered approach allows for all of this and is certainly nothing new as this already is happening in paper form of law. The difference is in efficiency and potential ease of use of creating such rules.

Visual contracts in real time and developing over time
The baby steps of this would for me be contracts over time a new legal field in Contract Law. Contracts are no longer statics and are often updated. But this never happens in real time and there is no current system to quickly visualize and then formalize said agreements. Bitlaw combined with blockchain makes this possible. Suddenly you can both visualize Peder Aas agreeing with Marte Kirkerud on transfering X amount if Y condition and change this on the fly based on other parameters like for example previous contracts (because now we can build huge databases of this with all transactions) and due to blockchain there is no need for a 3rd party to verify. Also you can learn from contracts and gradually evolve them over time in a much faster and better rate than what we can do with current legal systems.

Computational Democracies
This again opens what I call computational democracies. In computational democracies humans are still the representatives but in many cases the need for a 3rd party is removed. Voting can instantly be verified creating a lot of voting opportunities previously not within human reach and creating huge efficiencies in the voting process itself. Combined with bitlaw governance can happen and change at a much faster rate where users agree upon change or scope of changes (only x in contract terms y) and we can build a much bigger database of governance history thus making us learn faster what is more optimal ways to governance.

In any case these are only some of my ideas so far and I know it is far out there but do think these things are possible. I welcome any criticism also as I want to see if these are realistically possible ideas or not?

And believe it or not I believe a baby step in all of this is actually if we can get the community on board as a say on these governance issues I am sure we will run into with the treasury system. We also need to accept that we need a layered approach and accept that with decentralization and globalization we will need different rules for different groups. But for sure the community needs to be part of this

First step for me now is to try to read up on the state of the art. I am aware there is much research on DSL and I am aware that there is projects like this in Etherium (https://legalese.com/) and several other projects. But I am not aware of any such projects for Cardano that specifically targets governance issues.

In any case for myself and perhaps also for the community I will post summaries and what my thinking and what I have learned from each article/book/webpage I read about the subject. Anyone is welcome to join the discussion.

Cardano community reads Computational Law Article 1
First out is an article from 1957 : “Symbolic Logic: A Razor-Edged Tool for Drafting
and Interpreting Legal Documents” By Layman E. Allen. (http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.456.4342&rep=rep1&type=pdf)

In the abstract it is argued that we have written laws on paper to try to clarify what we have meant verbally. It is argued a better approach would be to use a logic symbol language (Computer languages comes to mind!)
to avoid any ambiguity.

The author then goes on to symbolize IF THEN statements (again very much terms from the computer language world) mathematically with a line between a P that implies a Q (IF P THEN Q)

The article then discusses how to pulverize legal clauses into such IF THEN statements exemplifying with a IF A break of contract happens THEN it can justify a refusal of a contractual duty (obligation.) He further defines P as antecedants and Q as consequences.

The author then discusses conjunction or what in program terms are AND statements. (Roses are red AND violets are blue) and connects it with antecedants or consequences. One P can have many consequences (Q1 & Q2 for example) or one consequence can be caused by several antecedants (P1 & P2 etc) or even that conjuctional antecedants can give conjuctional consequences (P1 & P2 = Q1 & Q2 & Q3 for example)
He argues this can be represented program wise with AND statements. He then goes on to show that the more complexity you have the more ambiguity you have in interpretation and that the logic statements required for all pairs of outcomes will grow by a large degree as you add conjunctions.

Next he looks at coimplications. These are derived from implications. He argues that when P implies Q then Q implies P and that NOT P implies NOT Q as well. He then argues this can be represented with IF AND ONLY IF. He argues this is the case in many court cases where the court does not want to apply for example that IF NOT P THEN NOT Q but only that IF AND ONLY IF P THEN Q.

He then discusses disjunctions (This for me is how we have tree like structures in program languages) where for example IF P THEN Q OR Q2 (OR here is an exclusive injunction where it is either Q or Q2) but it does not say if it can be both specifically. (if P then can it be both Q1 and Q2 at the same time) and then the nuansens of logical impications of the different types of disjunctions and how to represent them in a more logical form using the keywords ONLY IF and COIMPLIES and what I understand to be ANDOR (&OR) for inclusive disjunctions.

He then discusses negations (NOT) and how every positive will have a negative connected to it and then the relationship between all of these logical connectors and finds with what I call mathematical logic a lot of statements that are true between all of these logical connectors. One example is a true relationship between NOT, ANDNOT, OR and AND/OR. “If the reader is either NOT tired AND NOT bored, OR is tired AND NOT bored, OR is NOT tired AND bored, THEN the reader is NOT tired AND/OR NOT bored”

The table in the article summarize quite well

The article is primary focused on using these logical implications for drafting of laws by humans and also argues the need still for human interpretations and for the symbolic language as a tool. However for me a key takeaway is that many legal relations can be represented logically and thus also in a computer language. You need a strong understanding of logics and you need to have a common language - A domain specific language (DSL) to represent the terms precisely but it seems entierly possible and the idea of a symbolic language is very old. The code of ur-nammu dating some 2000 years B.C already had these IF THEN statments. This has to do with the fact that computer programs are all based on mathematical logic and it seems to be the next step to try to codify governance or the rules governing our life also into such a logical language as a way to further and decentralize also these aspects of our life as we have already done with knowledge through the internet or with financial ledgers through the blockchain.

7 Likes

quarried? :face_with_raised_eyebrow:

Query. English is not my first language ^^ and no no open mining pits (Quarry :P)

2 Likes

Cardano community reads Computational Law Article 2
WISE CONTRACTS: SMART CONTRACTS THAT WORK FOR PEOPLE AND MACHINES by James Hazard and Helena Haapio (https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2925871) is an article from 2017 propsing a hybrid model between machine-readable and easily human-readable contracts.

First the article discuss that contracts are hard to understand and that there even was a recent noble prize in economics showing how contracts are inherently incomplete. (You cannot possible account for everything in a contract…) The article argues that problems with contracts can be greatly reduced by contract visualization and design principles of “code is law” and they believe it can combine to something they call “wise contracts” :
image

In the paradigm they choose to try to implement this (Ricardian Contract paradigm) you need for full automation and legal enforcability 3 things: paramaters, code and prose.

Parameters are the particulars of each contract like prices, dates, quantities. So basically the variables. The author argues code can be forfilled by the technology of smart contracts and examplifies with Hyperledger, Corda or Etherium. For the prose the author suggest you can use the principles from software code to make universally codifyed contract prose and uses the data model of CommonAccord as an example of this. Basically you take a standard form and optinally visually represented supporting truly human-readable contracts and the author claims:
image

The authors then go on to discuss how to codify prose with five requirements for smart contract templates:

  1. Editing and methods for editing agreements.
  2. Transmission and standardized ways of doing this
  3. Stamping with protocols for executing contracts
  4. Binding ways to bind parameters (like price) to smart contract code. (So basically you would need some form of verification)
  5. Enforcability to make sure that the contracts can also be represented in some form accepted by legal authorities (so this also inplies that any system needs to be dualistic meaning it needs to live together with current legal system and be designed around and with current legal system.)

Article then discusses the importance of abstract layer through for examle xml or CommonAccord and how this can help build legal code / contracts as building blocks (basically same principle as lego) making non-lawyers able to rapidly add or adapt contracts.

Then it discusses visualization aspects and how computer should handle code while for humans it is best to visualize contracts in an easy to understand format and how layering can be used to give more and more detailed information if needed. (Notice also that layering could probably be used to differentiate between universal vs group or individual specific rules and present this easily visual as well.)

Transmission is discussed with examples like bitcoin or other blockchains and stamping can be done with cryptographically assured methods. Under binding it is suggested by simply having a hash you can bring the full context of a contract to bear meaning both very efficent data storage of contracts and for me it also symbolizes that it seems very possible for computers to store and learn from contracts very quicky.

The authors then discuss naming conventions and classifications and implications of different law systems across the world in how you layer and define these templates like based on CommonAccord. It also goes into discussion on definition of “things” like persons or properties like houses or cars. The article argues in a world more connected than ever it is already practically possible to make the law systems across the world “talk” with each other and suggest that contracts can incrementally improve so say for example first an US template is developed and then a Norwegian template is developed then an interaction layer between the two layers are developed. This learning capability and layered approach in my opinion opens a lot of possibilities.

The article then discusses how AI can develop more intelligent contracts for example deducting the relationship between things.

My key takeaways is that the article touches on many of the things I have been thinking about my self (I Had not read the article when I wrote what I wrote) and encourages me to believe others out there thing this is possible as well.

1 Like

Cardano community reads Computational Law Article 3
Formal Verification of Smart Contracts is a short article by several authors at Microsoft and Harvard (Bhargavan et al) in 2016 ( https://www.microsoft.com/en-us/research/wp-content/uploads/2017/08/solidetherplas.pdf )

Basically the article argues the importance of formal methods for smart contracts (IOHK does a lot of work on this as for example described in the video: https://www.youtube.com/watch?v=XEIfy8b1waw&t=6s) and examplifies with TheDAO contract exploit on the Etherium Virtual Machine. The researchers try to use a programing language F* to ensure a smart contract verification architecture. With this method the researchers discover some logical flaws in how solidity programs contracts and for me highlightning that any bitlaw system needs be programmed in a formally verified and very strictly logical programing language (haskell for example) Interesting also the article discusses a formal deductive logic verification program called WHY3 http://why3.lri.fr/ highlighting the fact that any computational law program needs to make sure its deductive logic is also sound and formally correct.

There are a couple projects that I recommend looking into that are relevant for your goal.

Ulex
An open-source legal system for special jurisdictions, online markets, ZEDEs, seasteads, and other startup communities. It combines tested and trusted rule sets from private and international organizations in a robust but flexible configuration.

Kratia
Kratia empowers these communities by enabling them with digital governance. It helps the communities grow, evolve and adapt by offering lego blocks for the decision-making process, that is, options to the way influence is distributed to the members, options to the way a decision is resolved once members have voted for proposals, and finally options to the automatization of the decision making process and how the decisions take effect.

Kratia is already being developed as a DSL in Haskell which means once finished it would be trivial to implement as a SC via Plutus for decision making. Ulex focuses more in the legislative side of things from a theoretical standpoint. These are two pretty big pieces though there would be a lot more needed for what you are aiming at.

2 Likes

Excellent I will be sure to check this out. Yeah many pieces of this is already in place but to make the whole puzzle of it would be quite some work. Smart Contract via Plutus sounds like a good way indeed. Lego block based contracts is exactly the type of visualization I want for some of these contracts and with option to go into more detail. So you can visualize quickly an overall term and then click on that to go into details if you wanted or needed.

So Ulex basically is used when there is a special legal jurisdiction instead of governed by all national laws. Interesting idea but I think most nations will not accept this and bitlaw will require to be dualistic and live alongside national laws. Just as for example EU law lives along national laws in countries like my own (Norway). Also we need to have on blockchain resulution mechanisms instead off judges etc off chain if possible. Kratia seems more concerned with members and resolution between said members. Again I believe it is better to go for a dualistic route where bitlaw has a semantic system that can produce a contract in any language and the terms used are in accordance with national laws and international law principles. But yeah for sure there are some use cases on for example more general resolution stuff like voting.

Cardano community reads Computational Law Article 4

This article is from Christian Prisacariu and Gerardo Schneider who at the time worked at the University of Oslo and is called A Formal Language for Electronic Contracts* and was published in 2007. (https://link.springer.com/content/pdf/10.1007%2F978-3-540-72952-5_11.pdf) Gerardo has done a lot of research in the field as can for example be seen here:http://www.cse.chalmers.se/~gersch/publication.pdf

In the article the authors look carefully into how to translate contracts into logic and proposes “extentions to the μ-calculus (https://en.wikipedia.org/wiki/Modal_μ-calculus) in order to capture the intuitive meaning of the deontic notions and to express concurrent actions” (I understood half of this but understand they are using math symbols to represent logic - another field I need to look more into.)

The article argues that there are several paths to translating contracts into computer language but believes the most promising one is through logic. They then specifiy that this does not have to be the branch of logics called deontic logic (https://en.wikipedia.org/wiki/Deontic_logic) but argues it has many important aspect that are needed like obligations, permissions and prohibitions. They then argue it is not an easy task to formalize such tasks (like obligations) and this has been a research topic for a long time starting as early as 1926. They then offer a historic review.

First came the so called Standard Deontic Logic where normatives like obligations and such where related to to quantifiers like all, some or no or modalities like possible or impossible Here one witnessed some logical paradoxes that needed to be avoided or representation issues like if norms should be aught to be or aught to do (represent actions of humans or represent state after the actions) . They also argue contract clauses by definition are violable because if not contracts would be useless. This implies that you have to consider contrary to duty obligations (CTD) and contrary to prohibitions (CTP). (Terms when duties or prohibitions are failed to be met.) It also summarize some other pitfalls that can happen in this type of logic and then propose some properties of a contract language:
image

They define then desirable properties of a language of contracts using μ-calculus (and frankly the symbol usage goes above my head and is something I need to study more carefully) and this tells me atleast that it is possible to represent this type of logic with mathematical symbols thus opening up a lot of mathematical methods to be used in scrutinizing and improving the logic. They also are pragmatic about the usage and claim that some of the philosophical discussions on deontic operators do not apply to the field of e-contracts for example discussions on logic for when one is NOT obliged to perform an action. Or what I believe is the math symbol term ¬O(a). At this point the article simply got way above my head and I realize I need to learn more on how the math behind it is representing the logic so that I can fully understand the article. If anyone out there have a solid understanding of this feel free to join the discussion :slight_smile:

Key takeaway after reading the article is that Math is a powerful tool to represent this form of logic and something I need to learn more about when advancing my knowledge in this field. Another key takeaway is from the summary at the end of the article where they compare approaches to defining a formal language for contracts and give a summary that while all the approaches carry some aspects of e-contracts none of them have managed to fully capture the intuitive properties while avoiding any logical paradoxes so far (in 2007.)

Cardano community reads Computational Law Article 5
A CNL for Contract-Oriented Diagrams by Camillieri et al (among others Gerardo Schneider) from 2014 (https://arxiv.org/pdf/1406.5691.pdf) examines a framework for creating contract diagrams (so basically how to visualize contracts) and uses a controlled natural language (https://en.wikipedia.org/wiki/Controlled_natural_language) together with a Gramatical Framework (https://en.wikipedia.org/wiki/Grammatical_Framework) to ease the scemantics interpritations (Seems like a good way for baby step starts of translating on the fly contracts to any language).

CNL’s have for example been used for wikipages (article examines one such ACEWIKI) that needs to be translated automatically to many languages but the authors argue they are even better used for domain specific languages (DSL) since they generally do not need to deal with the full with of a language. (My notes they only need precise words for meanings on specific topics.) They then present C-O diagrams as a way to visualizing normative texts (laws / contracts) and a full ecosystem of how this can interact with automatic translations and visualization of contracts using web applications and web editors. (Note: This seems to be easily able to be adapted to say a Cardano Wallet backend like Daedelus)
image

The host language for all the tools mentioned (CNL & GRAMATICAL FRAMEWORK) is haskell and the translation interaction is summarized in the following picture
image

So for example a contract written in a controlled natural english language (only some keywords) is then parsed first with prettying up any difference in line spacing / punctuations and such and with a Gramatical framework for word placing and it actually also creates the haskel soure code that can read or show the object and then this is converted to the web application data type (an xml format called COML) to be read by editor / shown by webpage. So basically with this process you can have one user reading the contract in english and the next one in Japanese. They are both semantically correct and mean the same thing as it is done in a controlled natural language environment where only none ambiguous words are used. AFter this they also discuss and show implementation of a web editor of such contracts.

The article then discusses the clauses that can be used in such a CNL like for example required and then discuss similar aspects like the 1957 article on usage of conjunction but also on constraints like time constraints.

With the system you can both show with objects but also with verbalization the contract terms as is shown in the following picture
image

An interesting review then follows of using a classification system to rate the qualities of the CNL with the so called PENS system. (PRECICION, EXPRESSIVITY, NATURALNESS, SIMPLICITY) and its properties as written language for formal representation originating from academia for use in a specific domain.

Key takeaways after reading article: In a controlled natural language environment both visualization of contracts and on the fly translation of contracts between languages is possible.

1 Like

Cardano community reads Computational Law Article 6
In the article Reasonings about Partial Contracts from 2016 Azzopardi et al ( https://www.um.edu.mt/library/oar/bitstream/handle/123456789/22368/jurix2016-contracts.pdf?sequence=1&isAllowed=y) argues that while natural language techniques like deontic logic has been used to translate legal texts and in particular contracts they have suffered from incomplete coverage leaving parts of such contracts uninterpreted.

The authors proposes that one solution to the problem is making a parser that can identify what parts of a contract can be formalized with such logic methods and ranging from tagging text (note: Notice how this also can be used by databases of contracts to learn better what can be formalized in a contract) or identifying thresholds that trigger such identification for example a text with high amounts (1 million usd) would require a more strict treshold and could require to be individually interpreted instead of through say a formal method like deontic logic. They propose instead of a binary system with 1 true or 0 false you can introduce a quantity called unknown or undecided state (note perhaps this could fit well with the superimposed state of quantum computing?)

Based on this the article authors proposes a deontic logic with 3 states and then goes on to do calculus with math symbols on this (and again frankly I do not understand them yet so need to read up on this) but to my untrained eye they seem to mathematically prove logic relationships. They then go on to try to apply this into a word plugin to identify contractual language that can be formalized and contractual language that can not be formalized.

takeaways: Denotic logic based computer contracts or for that matter also based on CNL with denotic language has some limitation in what can be formalized and a first step into tackling this problem is to identify what can be formalized and what can not be formalized. Should be super useful with visualization to show what can be agreed upon there and then.

Cardano community reads Computational Law Article 7
Challenges in the Specification of Full Contracts is an article by Gordon J. Pace and Gerardo Schneider from 2009 ( http://folk.uio.no/gerardo/ifm2009.pdf ). In the article the authors argue to make a complete specification of full contracts (tolerated exceptions and reasoning about the contracts themselves so meta reasoning) can be achived with a combination of temporal and deontic concepts.

The authors offers a what I found a good summary of what deontic logic is
“Deontic logic is the logic concerned with moral and normative notions such as obligation, permission, prohibition, optionality, power, indifference, immunity and intention, among others”

They then discuss a “practical” deontic logic that accounts for some of the weakness of such logic when used on these more full type of contracts and when to use it and how to extend it also into for example systems that needs to be fault tolerant (for example internet communications)
image One part of this could be in a Service Level Agreement (SOA) to guarantee that services are rendered at the right time (if I undestood this correctly then one example comes to mind if say you needed in etherium to pay x amount of gas to initiate a smart contract as an example of a SOA)

The authors then also give a name to a specific form of deontic logic that is needed for formalizing contracts (for example for computer usage)
image

They then describe the syntax of such a type of deontic logic and then describe how sequence is needed to account for temporal aspects of contracts. (As in example you are required to first buy X then sell y at time z) And again from my limited understanding of mathematical symbols for this type of logic it seems they then show the difference in true mathematical statements where you account for temporal aspect or not account for it. They also discuss nuance difference between for example Sequence of obligations (SoO) and obligations of sequence (OoS)

They then discuss some logical paradox that can happen if user does not have a choice (you are obliged to send in form A or obliged to send in form B) instead of (send form A or B) that frankly I am not familiar with so also something I need to look more into (logical paradoxes)

They also discuss the use of repetition in contracts is then discussed with these temporal aspects and action aspects like choices. For example they show the difference between Contracts of Repetitions and Repetitions of Contracts and they discuss the term of unbound repetition. They continue with several other aspect of more advanced contracts like contracts that can have introspection and evaluate if clauses are active or not or even give power to cancel contracts if certain triggers happens.

Takeaways from this article is that when you move past more basic contracts and you add aspects like time or systems that needs to be fault tolerant you need to carefully consider any deontic logic implications this could cause. But if you do you could pave the wave for machine learning or AI to have self learning contracts that can adapt to a changing environment around it and have tolerance for such aspects as time or fault in services rendered (for example internet communication)

1 Like

Cardano community discuss ethical concerns 1
In a system where most right or obligations can be automatically changed, revoked or applied you need to make sure power is not misused.

Natural law (https://en.wikipedia.org/wiki/Natural_law) concerns itself with rights that come as a virtue of being human and is said to be incorporated into the united states bill of rights or for example the universal declaration of human rights.

I believe that any such digital law system like for example bitlaw needs to be more concerned with giving people rights and opportunities than taking it away and any form of punishment or costs needs to be limited to as large extend as possible. I believe this all stems from the natural liberty we all humans should enjoy and is an aspect of natural law. Richard Cumberland would summarize that natural laws can be reduced to “Benevolance towards all Rationals” and indeed if we focus on the good for others computerized laws could be a great benefit for humanity.

The trend in law is that we have managed to agree on transnational agreements now the more connected the world is. I think still it is naive to think we can agree on everything globally but I hope we can make more and more agreements between countries and between different cultures.

Free movement of goods (immaterial as well), freedom of speech and decentralized power (to the individuals) when possible seems to be some of the goals that could fit well with a search for liberty for men.

The ethics of laws is a huge field that I have not put much energy into but I believe this is an important discussion. With Marlow we will have financial contracts that can be written not only by coders. Who can control them? Can they be rewoked or are they final once out on the blockchain? I hope others can join in on the ethical discussions and try to think of any aspects of contracts in blockchain or even further as I propose governance law on the blockchain.

Cardano community discuss ethical concerns 2
Continuing the ethical discussion can Denotic logic store values behind decisions? And unless we have a yes answer to this ready is it not best to make such denotic legal systems layered so that different cultures or nations can use different values ?

The question asked is inspired by the role of rules in computational law panel discussion summarized here:
http://complaw.stanford.edu/readings/rules.html
image

Cardano community discuss ethical concerns 3
Perhaps some more inspiration both for the practicality of bitlaw and for ethical implications.


From this principle the author outlines that universality is possible. Anything in nature that is not inherently simple can be replicated by a computation of equal sophistication. For example the brain while incredible complex is according to this principle of computational equivalence possible to be replicated with an equally sophisticated computation. But it also means there is a limit. Computations can only match an equivalent complex system in nature. Not more than this. (Note: I believe then if you can replicate a complex system like the human intelligence then that system is itself able to further complexity so I disagree with this philosophical implication)

To offer some thoughts on this principle of universality one can examine the use of language in laws. The different languages are quite complex but there is among many language theorists a belief of universal language rules (for example Noam Chomsky’s theory of universal grammar https://en.wikipedia.org/wiki/Universal_grammar) and one of the article I showed in this thread (article 5) examines how one can with a system of a controlled language automatically translate the contract language to any other language thus creating universality. Basically this is in line with the thesis of Wolfram that anything that is not simple can be replicated by any system of equal sophistication. Since we do not yet have the full understanding of all language rules how they translate to each other universally the scientists used a more limited and controlled language and where able to show then that with this level of sophistication they had the knowledge to compute it and have it represented by a program. The core of this principle for me gives hope because it also implies that we can somehow translate values and ethics into these computer logics.

1 Like

Cardano community reads Computational Law Article 8
In the article Automatic Conflict Detection on Contracts by Fenech et al ( http://hjem.ifi.uio.no/gerardo/ictac09-contracts.pdf ) the authors move the focus from a single contract to the interaction between contracts and the detection of any potential conflict between them.

The authors argue in today’s Service-Oriented Architectures (SOA’s) there are often decoupled services that collaborate often even within an organisation and certainly outside a local organization. For example if the deadelus wallet had Dapps they would have SOA by the POS staking and also by any user of the Dapp and perhaps even between DAPPS and the DEADELUS wallet.

The authors build upon the denotic logic electronic contracts (as discussed in article 7 here) with trace semantics. While again I do not understand the math symbol usage providing the logic and truthfulness of the statements this should be verified by anyone that is well versed in this field. In short the article proposes there are 4 conflicts that can arise.:
image

Note: Unlike before in contract history with these computational contracts these conflict detection can be run quickly and with a possible infinitive database. It seems very usefull to have a framework for such conflict detection in adapting for Contracts over Time and where contracts evolve over time and learn from each other.

The authors after describing the conflicts that can arise and showing this mathematically then goes on to do a case study of a contract for an airline company and in relationship with the ground crew. They then translate this into deontic logic statements and run the scemantic trace to see if there is any conflict. They find that one is both obligated and forbidden to give a boarding pass is a conflict and then they change a clause:
image
For me this highlights how I believe contracts in computational law will be able to adapt to each other more. The user ofcourse needs to be in control of such changes but when you can quickly access any conflicts and in a database of all relevant contracts it becomes much easier to have contracts that evolve over time.

Key takeaway: Contracts often do not live in a vacuum and databases of all relevant contracts and conflict detection is both important and possible. This should also support a new type of contracts that evolve over time.

I know nothing about the law, but here you enter my territory. I’ve studied this general area for many years. Of course we can and already do program rules deriving from values and ethics. But values and ethics themselves, the fundamentals from which such rules are derived, cannot be found outside of minds that are embodied, and embedded in social settings. It is the context that gives them (and everything else) meaning.

Sorry for the abrupt, know-all tone, but it is impossible to properly present a philosophical argument in this sort of setting. If you like, take it that this might be a serious issue, something not to be overlooked.

1 Like

Absolutely it is a serious issue. The ethics of computational law is of massive importance since it can be used for incredible damage as well as incredible good. We could so easily have Orwell societies with this technology.

At the heart of your statement i believe is a problem in how much complexity do we need to compute to make sure we have laws that create value and benefit and are ethical by both local, national and transnational standards. (We might not need to understand the WHY’s of ethics only the HOW’s.) The principle of computational equivalence would suggest any form of ethics is possible to code and we simply have not yet a complex enough understanding of the natural principles of it. From a pragmatical viewpoint (where I am at) we need to make sure we encourage good from these laws (that’s why I discuss natural law) and that the humans are still in control of how the laws evolve and are assisted by the computers to reach new heights of understanding. For example contracts over time evolving by computers still need human approval and this should always be there.

Basically for me what I am saying is that no I think you are wrong in the long run and with paradigm shifts this too will be possible (at least if you believe the principle of computational equivalence) and yes you are right here and now so we need to be pragmatic about this. He who says he can and he who says he cant are both usually right :slight_smile:

But if you have details into why you believe it is not possible then feel free to discuss. This is open to all and I welcome that.

Oh and why do I believe it is true that we can compute anything in nature? Because math that lies behind computer logic is a tool to describe nature. The better the tool gets the more we are able to describe nature. And then anything as complex as what happens in our minds is also part of nature. I remember 10 years ago I was discussing with my friends in it (one who is now PHD in IT and one who is now working with AI) that we will be able to make an AI that thinks more the way human thinks, all we have to do is replicate the way we think into computer code. I remember some comments where this is silly it is too complex. We cant replicate. I said why not we can code what the senses experience, we can code what it is like to learn, we can code memory etc. And low and behold this is already now happening with the new chess AI that does far less calculations but crushes say Stockfish because it uses more intuitive thinking methods. Another example is in how Watson now is starting to understand language and learns through language. (So we have started to codify something as complex as language into math logic.) We are simply getting better at replicating what is in nature. The human mind is a part of nature.

You’re assuming that ethics is what philosophers call “a natural kind”: a type of thing that exists independent of the observer. I’m saying that it’s not. It is essentially subjective. (Actually, inter-subjective, but that’s another technical term, and “subjective” is close enough.)

Brains are part of nature. Mind is a concept that we find it useful to apply to some of the functions of brains, in a social context.

I’m going to leave it there, this is too deep for any forum discussion. If you’re interested DM me for my research MSc dissertation “Mind, Matter, Meaning and Information”.

So I believe what you are talking about is the individual ethic while I am discussing the ethics that can be agreed between persons (aka LAWS). Ofcourse for example to a person who commits a criminal act there can be many reasons for it individually and it is going to be shaped by any observer. The question is often however what signals it sends to society (so basically the group vs individual ethics.) And while you cannot find something that is 100% true for everyone (binary yes no) you can find something that a majority agree is mostly correct. For example most of us believe in freedom of speech. I do not see why this cannot be discovered and coded in computer law as well.

And even then I believe you could still code it. You just would need to apply it to every single person you where considering the ethics for. This is in line with the principle of computational equivalence. While quite complex and involving every single persons individual perspective it is possible if a model then included every individuals perspective. Perhaps with AI we can learn through AI more about each individual self and have a neutral 3rd party observer that gives us the ability of introspection into each of our own minds. With the raw data power you can have from huge databases and machine learning you can start to see ethical patterns you have not seen before and evolve group ethics.

Yes it is going on my read list :slight_smile:

Edit just read article at least.
https://www.triple-c.at/index.php/tripleC/article/view/323

Familiar with the subject as I have read some of these theories through my psychology studies as in for example the hard problem of consciousness and the term of Qualia. At the core for me this is basically a question if you believe the world is deterministic or not. If everything can be traced back to a cause and effect it is possible to replicate. For me this is the case. The mind is caused by neurons firing in the brain. The system is just too complex for us too understand fully yet. However while deterministic it is still randomized but not from the system itself but the interaction between different systems and the fact that they are not binary choices but branching choices. If I decided right now to write UGABUGGA I can do that, so I have many choices. However my words are limited by my knowledge , my history, my general mood, the context they are written in and many many many more factors. But the heart of the theory of computational equivalence is that if we knew all the factors we could compute all my choices. Maybe that lists is some millions but it is still possible to represent computationally. We could then offer a tree like structure of possible outcomes. Gigantic. But still possible.

And that brings me back to the fact we do not need this. We do not need the WHY’s of ethics as we can do that ourself to ask why do we need this and this type of ethic as it involves ourself as humans. But the HOW’s so that we make sure computational laws stay relevant to current group ethics is very important. It is easily possible to imagine create a world where everything is tracked and punishment is dispensed automatically. But do we want such a world? HECK NO. So we make sure laws focus on increasing rights not limiting rights and not tracking us but giving us more liberty. At the core of this is ofcourse WHY we want that and that is a subject we can do as humans and do not need to compute into the computer. However we can learn more about WHY through seeing how contracts are formed and with the databases we can now create that we never have created before of for example governance contracts.

Then in the article you discuss for example empathy. While it will take us great pains to get to a point where computers understand why to be emphatic we can create rules that in general will create empathy. However I believe for the discussion of computational laws that focus on contracts over time and governance it is a sidetrack and more needed in AI research.

The discussion on intentional information however is interesting. I like the analogy used with the paint left on a painting only showing the physical information and not the intent that was behind the painting fully. But has this not more to do with the tool used (paintbrush and paint) limiting the translation between intent and representation than it has to do with it not being possible to understand intent? And to further the anology instead of using one painting how about we do a billion paintings and then analyze this for common features and intents. Could we learn something we have never learn before or perhaps so deeply in ourself that any one of us could not reach it through individual introspection but only through a computer 3rd party analyzing it we learn the pattern ? This is already happening many places. We have AI learning to scan CT scans and finding new signs of tumor that not any single doctor or even group of doctors discussing and writing papers about have found. But because computers do one thing damn well (compute) they can take massive amounts of data and look at it in a way we have never looked at the sensory data around us before. So again my answer is short term no we cannot hope to learn these things. Long term I feel history has shown humans understand more and more complex interactions in nature and at its heart I believe these are complex interactions of nature we simply need better tools to understand. Intents is one such thing.

Back to the law and governance discussion. While we cannot understand the intents of a contract we do not need to do so to effectively make governance contracts. We can make limited controlled natural languages (CNL’s) that can be a tool to convey our intent faster - easier - and better than ever before. We need only to understand how we can make sure there are limits on bad intents for society as a whole, not that the machines understand fully the intent of each individual user.