The problem of 'should' statements in primary documents. General Court Policy amendment proposal to clarify that they are to be interpreted as imperative (and other suggested solutions)

Hello community.

In relation to the interpretative issues arising from the use of the words ‘should’ and ‘should not’ in primary documents, and considering an appropriate normative formula to propose a modification to the General Court Policy in order for these words to be interpreted in an imperative sense, as the responses I found were unsatisfactory, my research and reflections on the subject led me to develop what I share below. In addition to the aforementioned amendment proposal, I present some reflections, conclusions, and also suggest other lines of action to seek a solution to the problem.
The text may be a bit lengthy, but I considered including it in its entirety so that the process leading to the conclusions and final proposals can be thoroughly understood.


Option 1.

"The words “should” and “should not” contained in the arbitrable application primary documents are to be interpreted as follows: “Should” as imposing a requirement or obligation; “should not” as imposing a prohibition.
This rule of interpretation does not apply if the aforementioned primary documents expressly state that such words have a specific meaning or are to be interpreted in a certain way."

Option 2.

"The words “should” and “should not” contained in the arbitrable application primary documents which, due to their grammatical context, are logically admissible to be interpreted in an imperative or discretionary sense, are to be interpreted as follows: “Should” as imposing a requirement or obligation; “should not” as imposing a prohibition.
*This rule of interpretation does not apply if the aforementioned primary documents expressly state that such words have a specific meaning or are to be interpreted in a certain way."


Appropriate design of normative systems.

In very introductory and generic terms, the principles and objectives that should guide the design of any regulatory system include transparency, predictability (equivalent to legal certainty), celerity, among others. These characteristics make any system of rules reliable. Finally, it is reasonable to think that all the aforementioned features would lead to an increase in individuals or entities interested in submitting to such rules – that is, in the adoption/use/demand for systems operating based on a rule system with such characteristics.
Extrapolating this to the Kleros court system would mean that if its rule system – the primary documents of services/platforms/dApps integrated with Kleros and the Kleros court policies – possesses the aforementioned characteristics, it could result in a higher adoption of the protocol.

To achieve this, it is necessary for the norms to be designed and drafted in a technically appropriate manner.
This involves following various parameters of action, among which the following stand out:
a) Being written in plain language, in a concise, simple, clear, accurate, easily understandable, unambiguous, direct, coherent, grammatically correct manner, etc.; and
b) Being structured in such a way that the normative framework is as self-contained as possible, for example, by establishing explicit concepts and rules of interpretation.

These principles and parameters of action have the ultimate goals of a) facilitating the understanding and compliance with the normative framework by the community interested in it, b) facilitating and providing the greatest possible certainty regarding the interpretation and application of such regulations by the competent bodies (the jurors, in the case of Kleros), and, in this way, c) ensuring that the normative system is effective, and therefore, that the objectives sought to be achieved through its existence are fulfilled.

The words “should” and “should not” do not contribute to an appropriate normative design.

The words “should” and “should not” have many possible meanings depending on the grammatical context in which they are inserted. Moreover, many of these meanings are potentially applicable in normative texts [1], indicating, for example, imperativeness or discretion.
This makes these words potentially ambiguous, and as a result, they can generate, among other problems, interpretative divergences when applying the norms.
Various authors in the field of legislation drafting have expressed concerns in this regard[2].

Proposed regulatory modification. Purpose.

Regarding the proposal to incorporate a rule of interpretation for the words ‘should’ and ‘should not’ into the General Court Policy, it seeks to provide a higher degree of consistency/sufficiency to the normative system – comprised of the corresponding primary document and court policy – aiming to make it more suitable for addressing new issues that may arise in the Kleros justice system, especially when the primary documents do not seem sufficiently capable of processing and solving these issues due to design inconsistencies – for example, by containing ambiguous expressions like ‘should’ or ‘should not,’ as explained. In this way, the normative system can remain as close as possible to the principles and parameters outlined in points 1) and 2).
However, this purpose may not be realized due to the problems presented by the proposed normative technique.

Proposed regulatory modification. Problems it presents.

Also in relation to the proposed regulatory amendments, beyond what was expressed in point 5, in order to confront and (seek to) solve the problem of the meaning of the words ‘should’ and ‘should not’ as imperative or discretionary, resorting to the incorporation of an interpretation rule in the court policies, while it may have constructive aspects, is potentially problematic in the Kleros justice system – to a greater or lesser extent, depending on how it is designed.
The aforementioned is based on the following:

Option 1, being a non-ambiguous, closed rule that allows only one interpretation – at least from a literal perspective – would have the advantage of providing predictability/certainty about the functioning of the justice system (at least seemingly, due to what is stated in the paragraph immediately following).
At the same time, it has the disadvantage that, whenever the policy document does not expressly establish that the words “should” and “should not” have a specific meaning, in cases where the grammatical context clearly indicates that these words cannot be interpreted imperatively[3], there would be a situation in which an incoherent rule would emerge from the court policies in logical terms. This would create an interpretative problem: there would be a dilemma – at least theoretically – about whether the jurors, based on a strictly literal interpretation, must adhere to what the corresponding court policy establishes – which could lead to unfair and/or illogical decisions –, or if, faced with this normative incoherence, jurors, starting from a systemic interpretation, should make a reasonable decision – in a grammatical logic sense – that provides coherence to the regulation. Therefore, to rectify this normative inconsistency, a reasonable solution by the jurors would be required.

Option 2, being an open rule that allows more than one interpretation, has the disadvantage that it would not provide sufficient predictability/certainty about the functioning of the Kleros justice system. However, from another perspective, it could be argued that it would have the advantage of providing more predictability/certainty compared to the current regulatory regime – in which there is no interpretation rule like the proposed ones.
At the same time, it has the advantage of avoiding the disadvantage of Option 1.

Another disadvantage of both Option 1 and Option 2 is that Kleros courts could interpret rules as imperative that the drafters of the primary documents intended to have a different meaning – for example, establishing recommendations. This could pose the risk of contradicting the intended meaning by the drafters – and the services/platforms/dApps, through their governance systems – to the regulations.
If this scenario were to occur in practice, it could generate conflicts between the service/platform/dApp and its users and between the former and Kleros, which could affect trust in the Kleros justice system.

Proposed regulatory modification. Conclusion.

Considering what has been said before, I conclude that the issue of interpreting the words ‘should’ and ‘should not’ as imperative or discretionary in primary documents that do not expressly establish their meanings cannot be definitively resolved solely through a modification of the Kleros General Court Policy – nor any other court policy.


According to all that has been said, I understand that an appropriate approach leading to the solution of the presented issue regarding the words “should” and “should not” – and consequently providing more consistency to the normative system related to the Kleros justice service, as expressed in points 5 – requires, at least, five axes of action.

a) It is necessary to incorporate a rule of interpretation into the General Court Policy.
• It might possibly resolve problems arising from normative inconsistencies in the primary documents.
• By providing some predictability – or perhaps the appearance of predictability in the case of Option 1 – it would encourage clarification of the primary documents by services/platforms/dApps. In other words, if a service/platform/ dApp disagrees with the proposed interpretation rule and wishes to avoid its consequences, then what it should do is rephrase the wording of its primary document.
• It represents a partial and provisional solution (or attempt at a solution), as indicated in point 5, as it does not definitively solve the problem.
• Regarding the decision on which option presented would be the most appropriate – or another one different from these – information on how jurors have interpreted the rules in specific cases would be relevant (data-driven decision).
• With the reservations made in point 6, I personally preffer the formula of Option 2, as it seems less problematic than Option 1.

b) It is necessary to work with the jurors to improve their approach and judgment process, especially when the normative system of the Kleros justice system – the primary documents and court policies – is not clear/explicit enough to solve a specific case – normative gaps, etc.

c) It is necessary to adequately communicate the existence of the rule of interpretation to users of the Kleros justice system – in case it is incorporated into the General Court Policy – and explain its necessity.

d) It is necessary for Kleros to urge services/platforms/dAaps to clarify their primary documents, in accordance with what was expressed in points 1 to 4.
• In this regard, it is advisable to completely replace or reduce to the minimum possible the words “should” and “should not” with unambiguous alternatives and/or explicitly clarify their meaning in those same documents.
• Such reformulation does not necessarily have to occur after a conflict arising from discrepancies in the interpretation of the meaning of the analyzed words – as happened with PoH primary document after various disputes, such as Kleros Case #554 – but ideally in a preventive manner.

The first four axes of action mentioned would allow addressing the problem, reinforcing and organizing the Kleros justice system to the greatest extent possible – albeit with potential negative consequences, as discussed above.
However, ultimately, the problem persists: we are dealing with words that imply ambiguity and whose meaning is not determined in the norms.

e) Therefore, finally, because it addresses its cause, the adaptation/reformulation/redesign of primary documents – fundamentally to be unequivocal – is the only fundamental and definitive solution to the problem.


Redesigning the wording of the General Court Policy.

The words “should” and “should not” in the General Court Policy should be replaced with other words.
• Because, just as jurors might interpret these existing words in a primary document as non-imperative, they could do the same with these words in the General Court Policy.
• Because it would contribute to the coherence of the General Court Policy in the event that a normative incorporation similar to the ones presented is implemented.

Fell free to make any comments you want on the matter.
Feedback, reflections, ideas, etc. are very welcome to continue pondering and generating proposals to solve the stated problem.

Tranks for reading.


[1] As examples, the meanings of ‘legal obligation’, ‘moral obligation’, ‘strong recommendation’ and ‘suggestion’ can be mentioned.
[2] As examples, the following can be mentioned:

  • “'Should’ is to be avoided as ambiguous as it may mean ‘must’ or ‘may’”. (Xanrhaki, Helen. 2022. Thornton’s Legislative Drafting (sixth edition). Bloomsbury Professional. pp. 127).
  • “Should” […] can give different meanings to the verb it goes with, depending on the sentence. It can show that what the main verb expresses is mandatory, necessary or essential. But it can also give the verb it goes with other meanings, such as “it is likely that…” […] In order to avoid uncertainty, it is recommended that the use of the modal verb “should” […] be avoided". (Law drafting manual: A guide to the legislative process in Albania, pp. 69). Available at
    [3] Examples of the described type of rules are:
  • “The name should be the most commonly used name to refer to the asset. It does not necessarily need to be the official name given by project creators nor the one in the token contract.” (Token Curates List Policies, first paragraph).
  • “Face should not be covered under large piercings or masks hindering the visibility of facial features” (Proof of Humanity Registry Policy v. 1.3, Guideline #2, fifth subsection). Considering that in the context of Guideline #2, the modal auxiliary ‘must’ is used in the first four subsections, it can reasonably be interpreted that by using ‘should’ a different sense than ‘must’ was intended – that is, a non-imperative sense.