Datasets:

pdf2dataset
/

2dd6fe3cba26dfd09b86ddf4b1f4da15

Dataset card Viewer Files Files and versions Community

Split (1)

train · 29 rows

text stringlengths 171 3.14k	source stringclasses 1 value
Requirements Contracts: Deﬁnition, Design, and Analysis Ivan J. Jureta Fonds de la Recherche Scientiﬁque-FNRS, Brussels, Belgium, Universit´ e de Namur, Belgium, STEMCELL Technologies Inc., Vancouver, Canada ivan@ivanjureta. com,http://ivanjureta. com April 30, 2021 Abstract What are the necessary and suﬃcient conditions for a proposition to be called a requirement? In Requirements Engineering research, a propo-sition is a requirement if and only if speciﬁc grammatical and/or commu-nication conditions hold. I oﬀer an alternative, that a proposition is a requirement if and only if speciﬁc contractual, economic, and engineering relationships hold. I introduce and deﬁne the concept of ”Requirements Contract” which deﬁnes these conditions. I argue that seeing requirements as propositions governed by speciﬁc types of contracts leads to new and interesting questions for the ﬁeld, and relates requirements engineering to such topics as economic incentives, interest alignment, principal agent problem, and decision-making with incomplete information. Contents 1 Background 2 2 Problem 3 3 Solution Outline 4 4 Paper Outline 5 5 Rationale 6 5. 1 Background to the Rationale.................... 6 5. 2 Rationale for Contractual Relationships.............. 7 5. 2. 1 Prioritisation......................... 8 5. 2. 2 Acceptability......................... 10 5. 2. 3 Validation........................... 11 5. 3 Rationale for Economic Relationships............... 12 1ar Xiv:2104. 14110v1 [cs. CY] 29 Apr 2021	2104.14110v1.pdf
6 Solution 13 6. 1 Departure............................... 14 6. 2 Network................................ 14 6. 3 Nodes and Links........................... 14 6. 4 Roles.................................. 15 6. 5 Roles and Parties........................... 15 6. 6 Right to Request........................... 15 6. 7 Obligation to Satisfy......................... 16 6. 8 Obligation to Validate........................ 17 6. 9 Imperfect Transfer.......................... 17 6. 10 Handling Failure........................... 18 7 Alignment 18 7. 1 Interests................................ 18 7. 2 Simple Case of Conﬂict of Interest................. 20 7. 3 Interest Cases............................. 22 8 Conclusions and Open Questions 23 1 Background In research on Requirements Engineering, sentences such as “the system should do A”, “the software should do B”, or “the system must never do C” are called requirements. This is an example: A meeting scheduler should know the constraints of the various par-ticipants invited to the meeting within some deadline dafter invita-tion. [60] Requirements Engineering, as a research discipline, came out of the need to be clear and rigorous about how to deﬁne, document, change requirements for software and hardware, and how to evaluate if and how much requirements are satisﬁed. But requirements are not conﬁned to software and hardware. Law deﬁnes requirements on legal entities, governments on the behaviour of citizens, schools on the behaviour of students, and so on. When a proposition is called a requirement, it conveys what someone ex-pects, needs, wants, asks for. This general idea that requirements set expectations about the thing to be made or bought, and/or of behaviour or competence to exhibit1has been around ever since the early research publications in the ﬁeld. Ross and Schoman wrote “requirements deﬁnition is a careful assessment of the needs that a system is to 1Consider, for example, the following passage on Canadian citizenship requirements (see https://www. cic. gc. ca/ : “To become a Canadian citizen, you must: be a permanent resi-dent, have lived in Canada for 3 out of the last 5 years, have ﬁled your taxes (if you need to), pass a test on your rights, responsibilities and knowledge of Canada, prove your language skills”. Those requirements are not about things to make or buy, but on competences and behaviours. 2	2104.14110v1.pdf
fulﬁl. It must say why a system is needed, based on current or foreseen condi-tions” [54]. Brooks claimed that “[the] hardest single part of building a software system is deciding precisely what to build” [11]. Zave wrote that “Requirements engineering is the branch of software engineering concerned with the real-world goals for, functions of, and constraints on software systems” [63] Hsia, Davis, and Kung wrote that “Requirements engineering is the disciplined application of proven principles, methods, tools and notations to describe a proposed sys-tem's intended behavior and its associated constraints” [31] Continuing Ross and Schoman's line of thought, in arguing why it matters to think about goals of a system-to-be, van Lamsweerde claimed that “goals provide the rationale for requirements [... ] a requirement appears because of some underlying goal which provides a base for it” and he saw goals as that which relates the system to the organization which uses it [59]. Sommerville argued that “before developing any system, you must understand what the system is supposed to do and how its use can support the goals of the individuals or business that will pay for that system” [58]. 2 Problem We can have an arbitrary number of propositions that refer to what is needed, wanted, desired. Such proposition are common; they are there when people complain, for example, as well as when they set goals to achieve, or rules for how something should or shouldn't be done. This abundance of requirements-like propositions begs the following ques-tion: Are all these propositions always requirements? Or you can look at the same problem the other way: How does a proposition become a requirement? These questions are the focus of this paper. Ifgrammatical mood fully determines whether a proposition has the role of requirement, then the only condition that should be true for a proposition pto get that role, is that pis communicated in the right grammatical mood. This is what Michael Jackson argued in several papers in the 1990s [33, 34], including in the inﬂuential “Four dark corners” with Pamela Zave [64], and later, “A Reference Model for Requirements and Speciﬁcations” [27] with Carl Gunter, Elsa Gunter, and Pamela Zave. If the type of speech act determines if a communicated proposition gets the role of requirement, then all that's needed for pto get that role, is that the right speech act is used. John Mylopoulos and I supported this position with various co-authors in our publications on the core ontology for requirements engineering [35, 41] and on the Techne family of requirements modelling languages [38, 39]. If the assignment of a goal to a software agent is enough for a the goal to become a requirement, then as long as the goal is reﬁned to such extent that it can be satisﬁed by functionality implemented by software, this is enough for it to be called a requirement. That is Axel van Lamsweerde's position-“[a] goal under responsibility of a single agent in the software-to-be becomes a require-ment whereas a goal under responsibility of a single agent in the environment of 3	2104.14110v1.pdf
the software-to-be becomes an assumption” [59]. A goal for van Lamsweerde is exactly what Jackson called requirements (see above): “A goal is an objective the system under consideration should achieve. Goal formulations thus refer to intended properties to be ensured; they are optative statements as opposed to indicative ones, and bounded by the subject matter” [59]. Other than minor diﬀerences2, all three positions revolve around the same idea: a proposition is called a requirement, if it is such that we understand it as referring to something that is needed, desired, or otherwise asked for; that seems to be both the necessary and suﬃcient condition for a proposition to become a requirement. What is wrong with the above? In simplest terms, you can want and say that you want pto be satisﬁed, yet no one may need to invest eﬀort to do so. This ties closely with such intuitively appealing and simple observations as that we cannot satisfy all needs we may think of. As the Rolling Stones say, you can't always get what you want. More speciﬁcally, the problem is that the position above-that all we need for a proposition to become a requirement is that it is in the right grammatical mood or speech act-clashes with the following common sense observation. A proposition p, which may refer to something I would like to see satisﬁed in the future, cannot be a requirement for you to satisfy just because I wish this to be so and I say that I wish it to be so. Grammar and speech acts alone are not enough for someone to satisfy that which another person said they want, need, or desire. Something more is needed, something that will lead others to want to invest eﬀort to satisfy these expressed goals, needs, or whatever we want to call them. It follows that propositions become requirements for some other reasons than simply because someone wants them to be satisﬁed. The problem, then, is this: What are the necessary and suﬃcient conditions for a proposition to become a requirement? 3 Solution Outline The following paragraph summarises what is missing when thinking about re-quirements as propositions communicated to convey needs. For a proposition pto become my requirement for you, I have to have been given speciﬁc rights, which I can then exercise to give p the role of requirement, and you accept the obligation to bring about p. Both you and I should get value out of this arrangement: I will 2The only one that stands out is van Lamsweerde's constraint to be able to assign a goal to a software agent, which is essentially about level of detail, i. e., he asks to have the goal reﬁned up to a threshold level of detail tied to the feasibility of software to implement functionality that supports the original goal (see, e. g., [17]). This makes no diﬀerence, because he still sees goals the same way Jackson sees requirements. 4	2104.14110v1.pdf
if my requirement is satisﬁed, and you will if there is something you get out of it, i. e., there is some transfer of value to you for satisfying that requirement. Finally, you should be conﬁdent enough that you will be able to satisfy my requirement through eﬀort that you can invest, given the value that you expect. To take the above seriously is to accept, as I will argue, that a proposition pgets the role of requirement within a nexus of contractual, economic, and engineering relationships. There needs to be a contract which distributes rights and responsibilities, and makes economic relationships binding for the parties involved; there need to be economic relationships, that is, expectations of getting value for anyone to want to enter the contract; and there have to be, broadly speaking, engineering relationships satisﬁed, between the understanding of what requirements are about, and assumptions of what should be designed, built, and eventually used or run to satisfy them. As a result, a requirement is not just a proposition denoting desired future conditions, a proposition that is the content, so to speak, of what someone may wish. We cannot say that a proposition isa requirement at all. The proposition remains a proposition, but can be given the role of requirement by those who enact roles deﬁned in a contract, exercise their rights and act according to their obligations. A proposition maintains the role of requirement as long as the contract remains applicable. Requirements, in short, come out of contracts, not wishes. 4 Paper Outline Why should you take the solution outline above seriously? Why do we need contracts when focusing on Requirements Engineering ? Can't we deal with the engineering of requirements while ignoring the contracting around their satis-faction? Section 5. 2 focuses on these questions, and develops the argument that the engineering relationships for a proposition to have the role of requirement are inseparable from the contractual ones. If the engineering and contractual relationships are inseparable, then can't we at least keep economic relationships out of the picture? Can we abstract those away and deal with them separately? Section 5. 3 is dedicated to these questions, and concludes that contractual and economic relationships are insep-arable. Therefore, if we take engineering and contractual relationships, then we need to take economic ones along as well. Section 6 gives the necessary and suﬃcient conditions for a proposition to get the role of requirement. An important implication of seeing requirements as roles resulting from a nexus of contractual, economic, and engineering relationships, is that the so-called “requirements problem” looks like a limited treatment of a more com-plicated problem. I will argue that, unfortunately, we cannot be deﬁning and solving that engineering problem without deﬁning and solving the contractual 5	2104.14110v1.pdf
and economic ones with it. One of the interesting, and I believe constructive consequences is that we need to have a broader and richer discussion in Re-quirements Engineering research, one which does not set aside economic and contractual relationships and constraints that aﬀect so many parameters of how the engineering problem gets formulated and solved. The paper closes with a discussion of the weaknesses of the proposal in this paper, and mentions open questions that I encourage colleagues to consider. 5 Rationale To argue that requirements cannot exist without contractual and economic re-lationships, we take a step back and recall the basic ideas in Requirements Engineering research, on what the central problem is, i. e., what you are up against when doing Requirements Engineering. 5. 1 Background to the Rationale Requirements Engineering focuses on eliciting, modelling, and analysing the requirements and environment of a system-to-be in order to design its speciﬁ-cation. It is on the basis of its speciﬁcation that the system is built, updated, changed, its new releases planned, made, announced, rolled out. Speciﬁcations can take diﬀerent forms, ranging from minimalist to-do lists that hint at expec-tations and subsume implicit engineering solutions, to elaborately structured documentation on responsibilities of positions in the value chain, guidelines for employee coordination and collaboration, as well as formal software speciﬁca-tions made for use with a model checker. The design of the speciﬁcation, usually called the Requirements Problem, is a complex problem solving task, as it involves, for each new system-to-be, the discovery and exploration of, and decision making in, new and ill-deﬁned problem and solution spaces. Diﬃculties involved in solving an Requirements Problem instance are illus-trated by the variety of topics studied in Requirements Engineering research, such as requirements elicitation [24, 30, 18], categorization [16, 64, 35], vagueness and ambiguity [49, 46, 40], prioritization [42, 5, 29], negotiation [45, 6, 36], re-sponsibility allocation [16, 13, 22], cost estimation [7, 10, 55], conﬂicts and incon-sistency [50, 28, 60], comparison [49, 46, 47], satisfaction evaluation [9, 49, 43], operationalization [23, 22, 20], traceability [25, 51, 15], and change [14, 62, 12]. The de facto default view in Requirements Engineering, is that the speciﬁ-cation of the solution to build or buy, is produced incrementally, starting from a limited set of incomplete, inconsistent, and imprecise information about the requirements and the system's operating environment, and that each design step reduces incompleteness, removes inconsistencies, and improves precision, towards the speciﬁcation of the system [8, 16, 26, 50, 21, 64, 59, 13, 53, 38, 20]. 6	2104.14110v1.pdf
This important and general conceptualisation of the aim in Requirements Engineering is most clearly formulated in Zave & Jackson's “Four dark corners of requirements engineering” [64] mentioned in Section 2. Their view, denoted ZJ hereafter, is echoed in some of the most inﬂuential research in the ﬁeld, which both preceded and followed the said paper, including, for example, contributions from Boehm et al. [8, 6], van Lamsweerde et al. [16, 17, 60, 61, 59, 46], Mylopoulos et al. [49, 26, 13], Robinson et al. [53], Nuseibeh et al. [50, 32], to name some. According to the ZJ view, in any concrete systems engineering project, Re-quirements Engineering is successfully completed when the following conditions are satisﬁed [64]: 1. “There is a set Rof requirements. Each member of Rhas been validated (checked informally) as acceptable to the cus-tomer, and Ras a whole has been validated as expressing all the customer's desires with respect to the software development project. 2. There is a set Kof statements of domain knowledge. Each member of Khas been validated (checked informally) as true of the environment. 3. There is a set SS of speciﬁcations. The members of Sdo not constrain the environment; they are not stated in terms of any unshared actions or state components; and they do not refer to the future. 4. A proof shows that K, S⊢R. This proof ensures that an implementation of Swill satisfy the requirements. 5. There is a proof that Sand Kare consistent. This ensures that the speciﬁcation is internally consistent and consistent with the environment. Note that the two proofs together imply that S, K, and Rare consistent with each other. ” If the satisfaction of these conditions marks the end of Requirements En-gineering in any systems engineering project, then we can give a compact for-mulation of the default problem that Requirements Engineering should solve, which we call the Default Requirements Problem hereafter. Default Requirements Problem : Given a set Rof requirements, and a set K of domain knowledge, ﬁnd a speciﬁcation S, such that Ssatisﬁes the following conditions: 1. There is a proof of Rfrom Kand S, written K, S⊢R, 2. Kand Sare consistent, written K, S̸⊢⊥. 5. 2 Rationale for Contractual Relationships Why are contractual and economic relationships absent in the mainstream ac-count of the requirements problem? It is either that contracts and economics do 7	2104.14110v1.pdf
not matter in the Default Requirements Problem, or that there is a lot to say about that problem even without considering the two other dimensions. Neither of these positions is satisfactory. To see why, we need to consider three central, recurrent questions in Requirements Engineering: 1. How to decide the relative importance of requirements? This is called the prioritisation problem hereafter, and is discussed in Section 5. 2. 1. 2. How to determine if satisfying a requirement is justiﬁed, i. e., that we have good enough reasons to believe that that requirement should be satisﬁed? This is called the acceptability problem (Section 5. 2. 2). 3. How to determine if and how well a requirement is satisﬁed? This is the validation problem (Section 5. 2. 3). In the remainder of this section, I will argue that each of these problems is in fact a problem that involves contractual, economic, and engineering relation-ships. None of them can or should be seen simply as engineering problems. 5. 2. 1 Prioritisation The prioritisation problem reﬂects the inability to simultaneously satisfy all the requirements that we may want to satisfy. The prioritisation problem asks how to decide which ones to satisfy ﬁrst. How much we can prioritise depends on what the requirements are about, how much resources we can commit, and throughput, or roughly speaking, how much of the requirements can be satisﬁed by how much resources. This, in turn, begs the question of what we expect to gain from that commitment, and therefore, there are economic relationships at play when we do requirements prioritisation [56, 44, 3, 52]. Suppose that you expected the most competent people in the world to work to satisfy your requirements. Would this inﬂuence the content of the require-ments that you would ask them to satisfy? Would you ask for diﬀerent require-ments if this were not the case, if you in fact knew very little of who would work on them, and what their competence is? In Hertz Corporation v. Accenture LLP [2], one of many issues that the car rental company raised was that Accenture misled it into believing that “the best talent in the world” would work on satisfying the requirements that Hertz had at the time. Paragraph 4 of the Complaint (April 19, 2019) reads as follows: “After Accenture put on an impressive, one-day presentation for the Hertz team that included a demonstration of the transformed Hertz digital experience, Hertz selected Accenture to design, build, test, and deploy Hertz's new website and mobile applications (or 'apps'). ” The Memorandum & Order (October 25, 2019) adds detail: “Ultimately, Hertz hired Accenture following a one-day marketing presentation, in which Accenture touted its world-class expertise in 8	2104.14110v1.pdf
website and mobile application development. The presentation con-tained slides stating that Accenture's staﬀ consisted of '800 [e]xperts' who comprised '[t]he best talent in the world. ' The presentation also stated '[w]e've got the skills you need to win' and that Accenture would 'put the right team on the ground [from] day one. ”' Hertz claimed that the presentation was misleading, and Accenture's re-sponse was that it is “non-actionable puﬀery”. This is what the Court concluded as well: “Accenture's representation that it housed '800 [e]xperts' amounting to '[t]he best talent in the world,' along with its promise that it had 'the skills you need to win' and would 'put the right team on the ground [on] day one,' are quintessential examples of puﬀery. Accordingly, this Court concludes that the alleged misstatements in the marketing presentation are non-actionable. ” How does this relate to prioritisation? If one prioritises assuming that the best talent would be doing the work, one does it diﬀerently than if diﬀerent talent, so to speak, is expected to do the work: the best talent would presumably do more and better than anyone else. More importantly, if this was not in a marketing presentation, but was in some more accurate manner deﬁned in the contract, the contract would not only inﬂuence what gets prioritised, but also what enters prioritisation in the ﬁrst place. Going back to Default Requirements Problem, the content of Rwould be inﬂuenced by the contract, and since the contractual relationships are outside of the deﬁnition of Default Requirements Problem, it can only be part of a broader problem to solve. The incentives that a contract deﬁnes, or the economic relationships it sets up, also inﬂuence how prioritisation will be done. Suppose that a contract A pays out the same amount to those satisfying requirements regardless of the expected value of outcomes of diﬀerent require-ments. This would be the case of developing, say, software which streams video online, but getting paid for hours spent making it, regardless of how much it is actually used once up and running. By expected value of outcomes of satisfying a requirement, I mean the value that the system-to-be is expected to generate at run-time if it is in fact designed to satisfy that requirement. Suppose that in another contract B, their payout is proportional to actual value at run-time. In the same example, those developing it would be paid relative to advertising revenue, for example, which in turn is proportional to number of times each video is viewed (or some other metric correlating with usage). This diﬀerence between A and B exists in, for example, in employment contracts for those who accept the obligation to satisfy requirements: case A would be a contract that includes neither shares, nor rights to shares in equity of the legal entity (or in revenue, as in the case of royalties) that commercialises the system-to-be. In case A, there is no reason for them to prioritise or insist to 9	2104.14110v1.pdf
prioritise requirements which may be more diﬃcult to satisfy, but could generate higher beneﬁts, while they may think about this in another way if they can claim some of these beneﬁts, as in case B. In short, we cannot do requirements prioritisation while ignoring the con-tract. If you do, as Hertz did, you take on substantial risks. 5. 2. 2 Acceptability A contract could specify that the party which has the rights to give require-ments can indeed give anyrequirements, i. e., give the role of requirement to any proposition. If I have that right by that contract, then I can turn any proposition into a requirement, including something as ludicrous as “green cheese should grow on the Moon”. But even if the contract was written in such a way, and there was someone as delusional as to accept the obligations in it, then I could ignore any complaints they may have. That contract would guarantee a priori that all propositions I turn into requirements are acceptable as requirements, regardless of what these propositions are about. This is not realistic. One contractual setup that I have experienced in practice in the past is very much the one that seemed to be in place between Hertz and Accenture: Hertz and Accenture had, according to the Court Opinion and Order (March 3, 2020), a “Consulting Services Agreement” since 2004, and “[t]he Project was to be conducted in phases, and the services and deliverables for each phase were, in turn, speciﬁed in letters of intent ('LOIs') and corresponding statements of work ('SOWs'). ” In such a setup, the SOW would describe those requirements that the party which should satisfy indeed accepts to satisfy. Writing SOWs, in other words, comes after an assessment-however well or badly done-of clarity, completeness, and feasibility of producing the solution which is expected to satisfy the propositions suggested to be treated as requirements (or a subset thereof). I have also been involved on either side of very diﬀerent contracts, where the contract does not presuppose the existence of a comprehensive SOW. In place of specifying requirements on the deliverable (or solution to make to satisfy the requirements), it speciﬁes constraints on the process which the parties go through, in order to deﬁne requirements after the contract is signed. In place of buying a solution, one buys time from experts who are expected to be able to design and deliver the solution, even if the speciﬁcs of that solution, or of the problem to solve may be elusive at contracting time. One of the twelve principles in the Agile Manifesto reads “[t]he best architectures, requirements, and designs emerge from self-organizing teams” [4], something that has to be acknowledged through a contract: in the ﬁrst case above, where a detailed SOW is needed and requirements planned ahead, this principle cannot be satisﬁed and it is the contract that makes it unsatisﬁable. The speciﬁcs of the acceptability problem that we will face in a given situ-ation can be determined by the contract. For example, the contract may say that any proposition is acceptable as long as it is given by those with the rights 10	2104.14110v1.pdf
to confer the role of requirements to propositions. Another contract may ask that some evidence should be given for the acceptability of requirements, and it would have to deﬁne what counts as evidence. A third contract might say that any proposition given up to a speciﬁc date can be given the role of requirement, but not past that date. A fourth contract might not specify that date. Consider how diﬀerent these are, and the extent to which they shape the requirements engineering process that we will have to use to eventually satisfy the require-ments. In the third contract, we might ﬁnd the waterfall process good enough. The fourth contract makes waterfall a more diﬃcult process choice than one inspired by the Agile Manifesto. In conclusion, we cannot ignore the contract when picking the requirements engineering process, and speciﬁcally how we decide if a proposition can have the role of requirement. 5. 2. 3 Validation The validation problem concerns how you determine if a requirement is satisﬁed, and how well it is satisﬁed. If the contract speciﬁes how this should be done, then the requirements engineering process cannot ignore it. Paragraph 7 in the Complaint that Hertz ﬁlled against Accenture reads as follows: “Hertz relied on Accenture's claimed expertise in implementing such a digital transformation. Accenture served as the overall project manager. Accenture gathered Hertz's requirements and then devel-oped a design to implement those requirements. Accenture served as the product owner, and Accenture, not Hertz, decided whether the design met Hertz's requirements. ” If the contract speciﬁed that Accenture validates the requirements it im-plements, then that contract determines a key property of the requirements engineering process between these two parties. The point, again, is that we cannot dissociate how one does requirements engineering from the content of a contract which determines who has the right to give propositions the role of requirements. How does this tie to the Default Requirements Problem, the engineering problem? The issue is who is accountable for the proof that K, S⊢R. At ﬁrst sight, this should not matter; after all, it is a proof and if it is there, anyone can follow the steps in it and see for themselves. The problem, however, is that proof rests on only those relationships between propositions in K,S, and R which have been formalised and appear in there. For example, you can have K={p1, p1∧p2p3},S={p2}and R={p3}, and it won't really matter what each of these is about; you will have K, S⊢R. At the same time, p1may be about green cheese, p2about pink clouds, and p3about a property of a car rental company website. 11	2104.14110v1.pdf
5. 3 Rationale for Economic Relationships It may seem from the above that there are two ideas that I am tying to the con-cept of requirement, one being that there must be a contract, and the other that there must be some economic relationships, some expectations and exchange of value between the parties involved. These are not separate ideas, and we therefore cannot say that there is an engineering problem, to which I am tying a contracting one, and then a third, economic one-the contract and economics are so closely tied together that we cannot take one and ignore the other. Why is this so? The obvious idea is that there is no acceptance of obligation without expectation of value in return (ex-cept under duress, which we leave aside). Importantly this observation matches the central tenet of contract law, that contract establishes chosen obligation. “[C]ontracts [... ] arise through an exchange of promises. This is in-scribed in legal doctrine, in the principles that contracts are created through oﬀer, acceptance, and consideration. An oﬀer, according to the U. S. second Restatement on Contracts, 'is the manifestation of willingness to enter into a bargain, so made as to justify another person in understanding that his assent to that bargain is invited and will conclude it. (R2 Contracts: §24)' ” [48] For a proposition pto get the role of requirement, we need parties willing to enter the bargain, whereby there will be at least one such party that will expect pto be satisﬁed through this bargain, and another which expects to get value in exchange for satisfying p. “To establish a contract, an oﬀer must be met with an appropriate acceptance, characteristically 'a manifestation of assent to the terms [of the oﬀer] made by the oﬀeree in a manner invited or required by the oﬀer. ( §50)” [48] Forpto be a requirement, then, we need a party which accepts the oﬀer. “These requirements entail that all orthodox contracts contain promises. But not all promises establish contracts, among other reasons be-cause the law further requires that contracts be supported by good consideration. The consideration doctrine, in its modern form, adds a bargain requirement to contract formation. The Restatement says that '[t]o constitute consideration, a performance or a return promise must be bargained for' and adds that '[a] performance or return promise is bargained for if it is sought by the promisor in exchange for his promise and is given by the promisee in exchange for that promise. (R2 Contracts: §71)' Contracts, that is, must arise not out of a simple, gratuitous promise, but rather out of an exchange of promises. ” [48] 12	2104.14110v1.pdf
Finally, then, we do not have promises going in one direction only. Those asking for pto be satisﬁed need to promise, at the very least, that value will be given to those who promise to satisfy p, i. e., to perceive that satisfying pis a requirement for them. But don't we have promises when, for example, I ask you to give me your book, oﬀer you $20 in exchange, and you accept? Am I giving a requirement? It is an exchange, but does it require a contract? “[T]he main economic function of contract law is to assist trans-acting parties who face diﬃculties associated with non-simultaneous transactions. Stated diﬀerently, contract law facilitates deferred ex-changes. An example is useful here. Suppose that you agree to build me a boat in exchange for £10,000, payable in advance. Absent a law of contract, there is an obvious risk that, having received the £10,000, you will renege on our deal and pocket the money. ” [57] When we talk about requirements which involve making something-and we say a new system or a system-to-be typically in requirements engineering-then this is a deferred exchange, and it makes little sense therefore to ignore ensuring a contractual framework and being clear about economic relationships. In conclusion, contract, economics and requirements are inseparable. 6 Solution The necessary and suﬃcient conditions for a proposition pto have been given the role of a requirement are as follows. 1. There is a so-called Requirements Contract, which deﬁnes (a) the right to give propositions the role of requirements, called the right to request hereafter, (b) the obligation to satisfy requirements, (c) the obligation to validate if a product satisﬁes requirements, (d) the obligation to remunerate satisfaction of requirements, (e) the obligation to remunerate validation, (f) the right to request remuneration for satisfying requirements, (g) the right to request remuneration for validating requirements. 2. The Requirements Contract is enacted, i. e., there is a party for each role that the contract creates by deﬁning the respective rights and obligations. 3. Rights and obligations in the Requirements Contract are, respectively ex-ercised and discharged, and speciﬁcally, the party which holds the right to request indeed requests that pbe satisﬁed, and thereby, pgets the role of a requirement. 13	2104.14110v1.pdf
6. 1 Departure This approach to deﬁning how a proposition gets to be called a requirement is diﬀerent from related work mentioned so far in this paper. There is no need to be concerned with grammatical mood, speech acts, or where the proposition appears in the reﬁnement tree. The concept of requirement is separated from intentional states and folk psychology; this addresses the problem I raised else-where [37]: intentional states cannot be known (i. e., while you may experience or observe your intentional states, you cannot do so with mine, nor can I do so with yours), so any appeal to intentional states as reasons for a proposition to be a requirement essentially asks you to assume that there is something behind the requirements that you have no way of observing, or otherwise accessing. It puts a veil of mystery where there is no need for one. Instead, the explanation for a proposition to have the role of requirement is that there is an enacted contract and parties exercising their rights and discharging their obligations. The solution above emphasises the contract in the ﬁrst condition. That ﬁrst condition is not enough, however. The contract must be enacted, the rights and obligations in it should be, respectively, accepted and discharged. This is captured in the second and third conditions. This is also where economic relationships come into play: if there are no expectations of value from making the various promises formalised in the contract, then there will be no one to enact it. 6. 2 Network The necessary and suﬃcient conditions given earlier can be represented as a network of relationships over expectations, rights, obligations, actions, and out-comes. That representation is called the Requirements Contract Network (also only “Network” hereafter). I use the Network to discuss alignment of interests of the parties in the Requirements Contract, and how the contract can be designed to support that alignment. The Network is shown in Figure 6. 5. Each node is an expectation, action, or outcome. The Requirements Contract itself appears through black nodes, which represent actions involving rights and obligations that the contract deﬁnes. The network is arranged in the Figure to show sequence over time, with time passing from left to right. 6. 3 Nodes and Links In the Network, everything starts from the expectations, which lead to actions related to the Requirements Contract, namely the acceptance of rights and obli-gations, which are, respectively, exercised and discharged through subsequent actions. Outcomes result from actions. Each link reads ”is necessary for”, a relationship indicating that the target of the link cannot happen if the source of the link hasn't. Taken together, all links targeting a speciﬁc node are the set of suﬃcient conditions for what that 14	2104.14110v1.pdf
node describes: expectations to be had, actions to be executed, or outcomes to occur. 6. 4 Roles Each expectation, action, and outcome is associated with a role. Roles are placeholders for parties who will ﬁll them when an actual Requirements Contract is enacted. Moreover, roles shown in Figure 6. 5 are merely one way in which expectations, actions, and outcomes can be grouped together and assigned to parties. Three roles are shown, and should be read as follows. Requester role, labelled Qin the Figure, is to be ﬁlled by the party which has the expectation of getting value if her requirements are satisﬁed. Maker role, labelled M, is responsible for making the product that should satisfy requirements; the party in this role expects value from doing so. Evaluator role, labelled V, is responsible for evaluating if the product satisﬁes requirements, and the party in this role expects value from doing the evaluation. 6. 5 Roles and Parties It is also not necessary that each role is ﬁlled by a diﬀerent party. For example, in both Hertz Corporation v. Accenture LLP and GB Gas Holdings Ltd v Accenture (UK) Ltd & Ors [1], Accenture ﬁlled itself both the Maker and Evaluator roles. Whether that is the best choice is a separate question; there is a lot to say about how many parties you may want to involve in a Requirements Contract, depending on which role you have; I return to this in Section 7. 6. 6 Right to Request The starting point is the expectation of value from having requirements satisﬁed (denoted ERin the Figure). There is no reason for the Requirements Contract to exist if there is no such expectation. In order for a party to accept the right to request, Rt R, and provide requirements that need to be satisﬁed, three conditions need to hold: The party which is to accept the right to request has the expectation of value from seeing those requirements satisﬁed ( ER), A party accepts the obligation to satisfy requirements ( Ot R), A party accepts the obligation to validate if requirements are satisﬁed (Or V). 15	2104.14110v1.pdf
Expectation, right, obligation, action, or outcome V VQ V VV Q QVVVM Q Q QMM MMQ Q Symbols Necessary for Expect value from requirements being satisﬁed ERAccept the Right to Request Rt RAccept the Obligation to remunerate satisfaction Ot RS Accept the Obligation to remunerate validation Ot RVKR, RRExercise the Right to request A(Rt R )Discharge the Obligation to remunerate satisfaction A(Ot RS) Discharge Obligation to remunerate validation A(Ot R V)Experience value from the investment to design, make, release a solution to satisfy requirements VS Expect value from the investment to satisfy requirements EPAccept the Obligation to Satisfy Requirements Ot R KP, RP, SP, PPDischarge the Obligation to satisfy requirements A(Ot R )Accept the Right to request remuneration for satisfying requirements Rt RSExercise the Right to request remuneration for satisfying requirements A(Rt Rs) Discharge the Obligation to validate the solution A(Ot V )Accept the Obligation to validate the solution Ot VExpect value from the investment to validate the solution EV Accept the Right to request remuneration for validating the solution Rt RVExperience value from the investment to validate if the solution satisﬁes requirements V(A(Ot V)) Exercise the Right to request remuneration for validating the solution A(Rt R V)KV, RV, SV, PVValidation concludes that the solution satisﬁes requirements Validation concludes that the solution does not satisfy requirements Experience value from requirements being satisﬁed VR... Q Requester role M Maker role V Evaluator role Figure 1: Nexus of rights, obligations, actions, expectations, and outcomes subsumed by a Requirements Contract 6. 7 Obligation to Satisfy No one will accept the obligation to satisfy requirements ( Ot R) if they lack an expectation of value from doing so ( EP) and an idea of what the requirements and assumptions around them may be; the latter comes from having at least some requirements and assumptions communicated by the party which accepted the right to request (the party having the right to request should also exercise that right). The interplay between knowing requirements, and accepting the obligation to satisfy requirements, is captured in two ways. The ﬁrst is the need for the right to request to be exercised, in order to accept the obligation to satisfy re-quirements (link from A(Rt R) to Ot R). The second is the overlap of acceptance of the right to request with the acceptance of the obligation to satisfy require-ments, and then the overlap of the latter with the exercising of the right to request. 16	2104.14110v1.pdf
6. 8 Obligation to Validate Same applies to validation, in that one needs to expect value from doing it ( EV) and know at least some of what they are getting themselves into (the right to request needs to be exercised), in order to accept the corresponding obligation (Ot V). In addition, to accept the obligation to satisfy requirements Ot R), the party needs to have the right to request remuneration for the investment that she will have to make to actually do so ( Rt RS ), i. e., needs a way to request value. At the same time, it will make sense for her to accept that right to ask for remuneration if someone else accepts the obligation to provide it ( Ot RS ). Note how this creates a nexus that links expectations to the obligations and rights in the Requirements Contract. We have the analogous situation for validation. For someone to accept the obligation to validate if requirements are satisﬁed ( Ot V), she needs to accept the right to request remuneration ( Rt RV ), and thereby be able to request value she expects. In turn, a party needs to accept the obligation to provide that remuneration for having validated if requirements are satisﬁed ( Ot RV ). All relationships identiﬁed above mean that expectations need to be there, as well as rights and obligations accepted, before the Right to request is exercised, and requirements are given. 6. 9 Imperfect Transfer The party which holds the right to request will exercise that right and provide two sets of propositions: A set KRof propositions that convey this party's understanding of her current situation, one which gave rise to her requirements, and A set RRof propositions that, by exercising the Right to request, this party gives as requirements. As diﬀerent parties provide and satisfy requirements, communication be-tween them means that there is a diﬀerence between one's assumptions and re-quirements and other's understanding of these assumptions and requirements. This may be, e. g., because of incompleteness, vagueness, ambiguity, or other deﬁciencies which come from the impossibility to be perfectly clear and compre-hensive in providing all the assumptions and all the requirements that one may want to provide, or considers implied in that which one does in fact provide ex-plicitly. It goes back to the distinction between explicit and implicit knowledge [19]. The implication here, is that KRand RRare one party's, while the other will work on something more or less diﬀerent. Speciﬁcally, the party which discharges the obligation to satisfy requirements, will have the assumptions KP, have understood requirements RP. As the output of this party's work, we have the speciﬁcation of the product intended to satisfy requirements once in use, 17	2104.14110v1.pdf
SP, and the product itself PP. (The diﬀerence between SPand PPis that the former is the blueprint of the thing, and the latter the thing itself. ) KP,RP,SP,PPare the outputs of the investment to satisfy the requirements. In order to discharge the obligation to validate the product ( A(Ot V) ), the party who accepted to do this, will do it on the basis of its own interpretation of these outputs, namely KV,RV,SV,PV. Once it discharges this obligation, that party will exercise its right to request remuneration ( A(Rr RV) ), which leads to it to experience value (denoted V(A(Ot V) ) in Figure 6. 5). The result of validation will be either that the product does satisfy require-ments, or that it does not. If it does, the party who expected value from having requirements satisﬁed will experience value (denoted V(PR)). In addition, the party who expected value from satisfying requirements will experience some value (denoted V(A(Ot R) ). 6. 10 Handling Failure If validation leads to the conclusion that the product does not satisfy require-ments, we need to return to the Requirements Contract, and the rules it speciﬁes around the handling of exceptions, including failure of this kind. I leave these out of scope in this paper; one option is that failure to satisfy requirements leads to going back to exercising the right to request ( A(Rt R) ). 7 Alignment What are the interests of each party that decides to accept its rights and obliga-tions in the Requirements Contract? Are these interests aligned? When are they aligned? What if one is pursuing actions which improve its outcomes, but in doing so do not improve those of others? Why would they not be aligned? Is it possible to design the Requirements Contract in ways that improve alignment? Alignment is discussed in three steps in this Section. Firstly, we need to make assumptions about the interests that each party in the Requirements Contract may have. From those assumptions, we can catalogue sources of misalignment, which can be shown using the Network in Figure 6. 5. Finally, we can discuss how the Requirements Contract can be designed to reduce the likelihood of speciﬁc types of misalignment to occur. 7. 1 Interests Why would a party accept rights and obligations in a Requirements Contract? According to the Network in Figure 6. 5, the answer is as follows. There is a party which expects value if requirements are satisﬁed, ER, and is willing to invest to get this value-both satisfaction of requirements, and evaluation (validation) of the product need to be remunerated. 18	2104.14110v1.pdf
There is a party which expects value EPbecause it invests its own resources to satisfy requirements by designing, making, and delivering a product to the party which had requirements in the ﬁrst place. There is a party which expects value EVbecause it invests its own resources in evaluating if the product satisﬁes requirements. To simplify writing, I will refer to these parties through the roles they would be ﬁlling in the Network in Figure 6. 5: Requester expects ER, Maker expects EP, and Evaluator expects EV. What does expected value depend on? Each party needs to make an invest-ment in order to produce that which eventually yields value for them. Thus, at the very least, expected value will be a function of expected beneﬁts and expected costs; let's write these as follows. ER=E(BR)-E(CR) (1) EP=E(BP)-E(CP) (2) EV=E(BV)-E(CV) (3) Since the Requester accepts the obligation to remunerate the satisfaction of requirements (the node Ot RS is labeled Q in Figure 6. 5), the beneﬁts that Maker can expect are capped by the cost that the Requester is willing to bear. At the same time, the Requester remunerates evaluation, so that the expected beneﬁts of the Evaluator are also capped by the Requester's expected cost. More speciﬁcally, we have E(BP) +E(BV)≤E(CR) (4) It has been implicit so far, but it is important to note now that expected value for each of these roles should be positive, else there is no apparent reason for the losing party to enter the Requirements Contract. Assumption 1 For a party to consider entering into an Requirements Con-tract, its expected beneﬁts should outweigh its expected costs. E(BR)> E(CR) (5) E(BP)> E(CP) (6) E(BV)> E(CV) (7) Does this mean that a party should enter an Requirements Contract as soon as expected beneﬁts outweigh expected costs? Is that the aim that each party has, when entering the contract? If, as usually in mainstream economics [], expected value is expected utility, then one enters the contract in the aim of maximising one's utility. In other words, if we have three roles in the contract, and we assumed each expects value from exercising its rights and discharging its obligations in the contract, then we also have three parties (if each role is occupied by a diﬀerent party). 19	2104.14110v1.pdf
Assumption 2 Each party in the contract will make decisions which it per-ceives as maximising the value that she will actually receive after exercising the rights and discharging the obligations that she accepted by accepting the Require-ments Contract. The objective functions of the parties, provided that there are three of them, are as follows. Requester: max VR(8) Maker: max VP(9) Evaluator: max VV(10) I will also assume that expected and actual value are almost the same. I will challenge this assumption later. Assumption 3 Each party in the Requirements Contract will exercise its rights and discharge its obligations in such a way that makes actual value as close as possible to its expected value. VR≈ER(11) VP≈EP(12) VV≈EV(13) To keep the discussion simple still, I need to make another brittle assump-tion, to relate maximisation of actual value and of expected value. I will question this assumption later too. Assumption 4 At the time when a party considers entering into a Require-ments Contract, i. e., accepting a role in it and the accompanying rights and obligations, that party will maximise its expected value in order to maximise its actual, future value. Requester: max VR≡max ER(14) Maker: max VP≡max EV(15) Evaluator: max VV≡max EV(16) 7. 2 Simple Case of Conﬂict of Interest Consider now four situations that a party can be in, and for simplicity, let that party be the Requester. 20	2104.14110v1.pdf
The party believes that each incremental unit of expected cost adds more than that in expected beneﬁts: there is a gain to be made by taking on more expected costs, that is, the following is true: ∆E(BR) ∆E(CR)>1 (17) If so, then maximising expected value means deciding to invest more, up to some limit which is private information for that party. The party believes that each incremental unit of expected cost adds less than that in expected beneﬁts: it costs disproportionately more to get an increment in expected beneﬁts: 0<∆E(BR) ∆E(CR)<1 (18) If this party has a threshold of expected beneﬁts, she should reduce costs up to-if that happens at all-the point when she believes that marginal expected value is equal to marginal expected cost, that is, up to the point when: ∆E(BR) ∆E(CR)= 1 (19) The party believes that any additional unit of expected cost leads to an equal unit of expected beneﬁts, as in Equation 19. If she hasn't reached a private maximal expected cost that she is willing to bear, her interest is to invest more, i. e., to increase her expected cost as long as marginal expected cost equals marginal expected beneﬁt, i. e., until Equation 18 is true, or until she has reached the maximal expected cost she is willing to accept. Finally, we can have this situation: ∆E(BR) ∆E(CR)<0 (20) If this is because ∆ E(BR)<0, then any increase in expected cost is believed by that party to lead to a reduction in expected beneﬁts. If the fraction above is negative because ∆ E(CR)<0, then any reduction in expected cost is believed to increase expected beneﬁt. But this is only one party, yet we have three in the Requirements Contract. What happens if we have the following situation? ∆E(BR) ∆E(CR)<1,∆E(BP) ∆E(CP)>1,and∆E(BV) ∆E(CV)>1 (21) 21	2104.14110v1.pdf
Due to Equation 4, we can rewrite this as follows: ∆E(BR) ∆(E(BP) +E(BV))<1,∆E(BP) ∆E(CP)>1,and∆E(BV) ∆E(CV)>1 (22) The above shows a conﬂict of interest: interests of the Maker and Evaluator are to increase costs in order to increase their beneﬁts more, while this isn't in the interest of the Requester. This only scratches the surface of interest alignment. There are more than four situations situations identiﬁed above, that a party can be in. 7. 3 Interest Cases Figure 7. 3 shows eight cases, labelled clockwise from A to H. Each is called an interest case, in that it describes what a party should have as its interest, assuming the given relationship between the change of beneﬁts and change of costs. Figure 7. 3 has three parts. The upper left corner shows four quadrants made by positive and negative change of expected beneﬁts and costs of the Requester, and the reading of points in each quadrant. The upper right corner of the Figure shows the same quadrants, now split into eight combinations of positive and negative changes in expectations of beneﬁts and costs. The lower part of the Figure shows in detail all eight combinations, and highlights those where a change in expected beneﬁts and costs leads to an increase of expected value for the Requester. It shows that the Requester should want to be in one of the four situations A, F, G, or H, rather than others. In summary, Figure 7. 3 shows what the interest of the Requester should be, depending on the interest case it is in. It is important to note that this same set of interest cases applies to any party; For example, for the Maker role, the same interest cases apply, except that you need to replace BRwith BPand CR with CP. In A in Figure 7. 3, positive increase expected cost goes together with a positive increase in beneﬁts, and the increase in the latter is higher than the increase in the former, so there is an interest for the Requester to increase expected costs, since doing so increases expected beneﬁts more: increasing cost will increase expected value. In F, if the Requester makes a change, that change will involve both a reduction in expected cost and in expected beneﬁts, with expected beneﬁts decreasing slower than the expected costs, hence increasing expected value. In G, decrease of expected costs comes with increasing expected beneﬁts, and so, an increase in expected value. In H, the dynamics are the same as in G, except that expected beneﬁts increase faster. 22	2104.14110v1.pdf
FGH A 0+ +--ΔE(BR) ΔE(CR) Decrease of expected cost goes together with decrease in expected beneﬁts Reduction of expected cost goes together with increase in expected beneﬁts Increase of expected cost goes together with increase in expected benefots Decrease of expected cost goes together with increase in expected beneﬁts A: ΔE(BR) / ΔE(CR) > 1, ΔE(BR) > 0, ΔE(CR) > 0: ΔE(VR) > 00+ +--ΔE(BR) ΔE(CR)B C D E B: ΔE(BR) / ΔE(CR) < 1, ΔE(BR) > 0, ΔE(CR) > 0: ΔE(VR) < 0 C: ΔE(BR) / ΔE(CR) > -1, ΔE(BR) < 0, ΔE(CR) > 0: ΔE(VR) < 0 D: ΔE(BR) / ΔE(CR) < -1, ΔE(BR) < 0, ΔE(CR) > 0: ΔE(VR) < 0E: ΔE(BR) / ΔE(CR) < 1, ΔE(BR) < 0, ΔE(CR) < 0: ΔE(VR) < 0 F: ΔE(BR) / ΔE(CR) > 1, ΔE(BR) < 0, ΔE(CR) < 0: ΔE(VR) > 0 G: ΔE(BR) / ΔE(CR) > -1, ΔE(BR) > 0, ΔE(CR) < 0: ΔE(VR) > 0 H: ΔE(BR) / ΔE(CR) < -1, ΔE(BR) > 0, ΔE(CR) < 0: ΔE(VR) > 0Figure 2: Interest Cases of the Requester 8 Conclusions and Open Questions The intent behind this paper is to stimulate a richer discussion of how proposi-tions get to have the role of requirements. If you adopt the perspective that is oﬀered here, that requirements exist in a nexus of economic, contractual, and engineering relationships, new questions come up. What properties do typical requirements contracts actually have (such as the one between Hertz and Accen-ture)? What are the optimal properties of a requirements contract? How do we design the requirements contract to help align interests of the parties involved? References [1]GB Gas Holdings Ltd v Accenture (UK) Ltd & Ors [2009] EWHC 2734 (Comm). Royal Courts of Justice, Strand, London, 2009. [2]Hertz Corporation v. Accenture LLP, 1:19-CV-03508. District Court, S. D. New York, 2019. [3] Philip Achimugu, Ali Selamat, Roliana Ibrahim, and Mohd Naz'ri Mahrin. A systematic literature review of software requirements prioritization re-search. Information and software technology, 56(6):568-585, 2014. [4] Kent Beck, Mike Beedle, Arie van Bennekum, Alistair Cockburn, Ward Cunningham, Martin Fowler, James Grenning, Jim Highsmith, Andrew 23	2104.14110v1.pdf
Hunt, Ron Jeﬀries, Jon Kern, Brian Marick, Robert C. Martin, Steve Mel-lor, Ken Schwaber, Jeﬀ Sutherland, and Dave Thomas. Manifesto for agile software development, 2001. [5] Patrik Berander and Anneliese Andrews. Requirements prioritization. In Engineering and managing software requirements, pages 69-94. Springer, 2005. [6] Barry Boehm, Prasanta Bose, Ellis Horowitz, and Ming June Lee. Software requirements negotiation and renegotiation aids: A theory-w based spiral approach. In Software Engineering, 1995. ICSE 1995. 17th International Conference on, pages 243-243. IEEE, 1995. [7] Barry W Boehm. Software engineering economics. Software Engineering, IEEE Transactions on, (1):4-21, 1984. [8] Barry W. Boehm. A spiral model of software development and enhance-ment. Computer, 21(5):61-72, 1988. [9] Barry W Boehm, John R Brown, and Myron Lipow. Quantitative evalua-tion of software quality. In Proceedings of the 2nd international conference on Software engineering, pages 592-605. IEEE Computer Society Press, 1976. [10] Barry W Boehm, Ray Madachy, Bert Steece, et al. Software Cost Estima-tion with Cocomo II. Prentice Hall PTR, 2000. [11] Jr. Brooks, F. P. No silver bullet: Essence and accidents of software engi-neering. Computer, 20(4):10-19, april 1987. [12] Yuriy Brun, Giovanna Di Marzo Serugendo, Cristina Gacek, Holger Giese, Holger Kienle, Marin Litoiu, Hausi M¨ uller, Mauro Pezz` e, and Mary Shaw. Engineering self-adaptive systems through feedback loops. In Software En-gineering for Self-Adaptive Systems, pages 48-70. Springer, 2009. [13] Jaelson Castro, Manuel Kolp, and John Mylopoulos. Towards requirements-driven information systems engineering: the tropos project. Information systems, 27(6):365-389, 2002. [14] Betty HC Cheng, Rog´ erio de Lemos, Holger Giese, Paola Inverardi, Jeﬀ Magee, Jesper Andersson, Basil Becker, Nelly Bencomo, Yuriy Brun, Bojan Cukic, et al. Software engineering for self-adaptive systems: A research roadmap. Springer, 2009. [15] Jane Cleland-Huang, Raﬀaella Settimi, Chuan Duan, and Xuchang Zou. Utilizing supporting evidence to improve dynamic requirements traceabil-ity. In Requirements Engineering, 2005. Proceedings. 13th IEEE Interna-tional Conference on, pages 135-144. IEEE, 2005. 24	2104.14110v1.pdf
[16] Anne Dardenne, Axel Van Lamsweerde, and Stephen Fickas. Goal-directed requirements acquisition. Science of computer programming, 20(1):3-50, 1993. [17] Robert Darimont and Axel Van Lamsweerde. Formal reﬁnement patterns for goal-driven requirements elaboration. ACM SIGSOFT Software Engi-neering Notes, 21(6):179-190, 1996. [18] Alan Davis, Oscar Dieste, Ann Hickey, Natalia Juristo, and Ana Mar´ ıa Moreno. Eﬀectiveness of requirements elicitation techniques: Empirical results derived from a systematic review. In Requirements Engineering, 14th IEEE International Conference, pages 179-188. IEEE, 2006. [19] Zoltan Dienes and Josef Perner. A theory of implicit and explicit knowledge. Behavioral and brain sciences, 22(5):735-808, 1999. [20] Neil A Ernst, Alexander Borgida, Ivan J Jureta, and John Mylopoulos. Agile requirements engineering via paraconsistent reasoning. Information Systems, 2013. [21] Anthony CW Finkelstein, Dov Gabbay, Anthony Hunter, Jeﬀ Kramer, and Bashar Nuseibeh. Inconsistency handling in multiperspective speciﬁcations. Software Engineering, IEEE Transactions on, 20(8):569-578, 1994. [22] Ariel Fuxman, Lin Liu, John Mylopoulos, Marco Pistore, Marco Roveri, and Paolo Traverso. Specifying and analyzing early requirements in tropos. Requirements Engineering, 9(2):132-150, 2004. [23] Paolo Giorgini, John Mylopoulos, Eleonora Nicchiarelli, and Roberto Se-bastiani. Reasoning with goal models. In Conceptual Modeling—ER 2002, pages 167-181. Springer, 2003. [24] Joseph A Goguen and Charlotte Linde. Techniques for requirements elici-tation. In Requirements Engineering, 1993., Proceedings of IEEE Interna-tional Symposium on, pages 152-164. IEEE, 1993. [25] Orlena CZ Gotel and CW Finkelstein. An analysis of the requirements traceability problem. In Requirements Engineering, 1994., Proceedings of the First International Conference on, pages 94-101. IEEE, 1994. [26] Sol Greenspan, John Mylopoulos, and Alex Borgida. On formal require-ments modeling languages: Rml revisited. In Proceedings of the 16th inter-national conference on Software engineering, pages 135-147. IEEE Com-puter Society Press, 1994. [27] Carl A Gunter, Elsa L Gunter, Michael Jackson, and Pamela Zave. A refer-ence model for requirements and speciﬁcations. IEEE Software, 17(3):37-43, 2000. 25	2104.14110v1.pdf
[28] Constance L Heitmeyer, Ralph D Jeﬀords, and Bruce G Labaw. Automated consistency checking of requirements speciﬁcations. ACM Transactions on Software Engineering and Methodology (TOSEM), 5(3):231-261, 1996. [29] Andrea Herrmann and Maya Daneva. Requirements prioritization based on beneﬁt and cost prediction: An agenda for future research. In International Requirements Engineering, 2008. RE'08. 16th IEEE, pages 125-134. IEEE, 2008. [30] Ann M Hickey and Alan M Davis. A uniﬁed model of requirements elici-tation. Journal of Management Information Systems, 20(4):65-84, 2004. [31] Pei Hsia, Alan M Davis, and David Chenho Kung. Status report: require-ments engineering. IEEE software, 10(6):75-79, 1993. [32] Anthony Hunter and Bashar Nuseibeh. Managing inconsistent speciﬁca-tions: reasoning, analysis, and action. ACM Transactions on Software Engineering and Methodology (TOSEM), 7(4):335-367, 1998. [33] Michael Jackson. Problems and requirements [software development]. In Proceedings of 1995 IEEE International Symposium on Requirements En-gineering (RE'95), pages 2-8. IEEE, 1995. [34] Michael Jackson. The meaning of requirements. Annals of Software Engi-neering, 3(1):5-21, 1997. [35] Ivan Jureta, John Mylopoulos, and Stephane Faulkner. Revisiting the core ontology and problem in requirements engineering. In 2008 16th IEEE International Requirements Engineering Conference, pages 71-80. IEEE, 2008. [36] Ivan Jureta, John Mylopoulos, and St´ ephane Faulkner. Analysis of multi-party agreement in requirements validation. In Requirements Engineering Conference, 2009. RE'09. 17th IEEE International, pages 57-66. IEEE, 2009. [37] Ivan J. Jureta. What happens to intentional concepts in requirements engi-neering if intentional states cannot be known? In Heinrich C. Mayr, Gian-carlo Guizzardi, Hui Ma, and Oscar Pastor, editors, Conceptual Modeling-36th International Conference, ER 2017, Valencia, Spain, November 6-9, 2017, Proceedings, volume 10650 of Lecture Notes in Computer Science, pages 209-222. Springer, 2017. [38] Ivan J Jureta, Alex Borgida, Neil A Ernst, and John Mylopoulos. Techne: Towards a new generation of requirements modeling languages with goals, preferences, and inconsistency handling. In 2010 18th IEEE International Requirements Engineering Conference, pages 115-124. IEEE, 2010. 26	2104.14110v1.pdf
[39] Ivan J Jureta, Alexander Borgida, Neil A Ernst, and John Mylopoulos. The requirements problem for adaptive systems. ACM Transactions on Management Information Systems (TMIS), 5(3):1-33, 2014. [40] Ivan J Jureta and St´ ephane Faulkner. Clarifying goal models. In Tutori-als, posters, panels and industrial contributions at the 26th international conference on Conceptual modeling-Volume 83, pages 139-144. Australian Computer Society, Inc., 2007. [41] Ivan J Jureta, John Mylopoulos, and St´ ephane Faulkner. A core ontology for requirements. Applied Ontology, 4(3-4):169-244, 2009. [42] Joachim Karlsson, Claes Wohlin, and Bj¨ orn Regnell. An evaluation of methods for prioritizing software requirements. Information and Software Technology, 39(14):939-947, 1998. [43] John Krogstie, Odd Ivar Lindland, and Guttorm Sindre. Towards a deeper understanding of quality in requirements engineering. In Advanced Infor-mation Systems Engineering, pages 82-95. Springer, 1995. [44] Nupul Kukreja, Barry Boehm, Sheetal Swaroop Payyavula, and Srinivas Padmanabhuni. Selecting an appropriate framework for value-based re-quirements prioritization. In 2012 20th IEEE International Requirements Engineering Conference (RE), pages 303-308. IEEE, 2012. [45] Julio Cesar Sampaio do Prado Leite and Peter A Freeman. Require-ments validation through viewpoint resolution. Software Engineering, IEEE Transactions on, 17(12):1253-1269, 1991. [46] Emmanuel Letier and Axel Van Lamsweerde. Reasoning about partial goal satisfaction for requirements and design engineering. In ACM SIGSOFT Software Engineering Notes, volume 29, pages 53-62. ACM, 2004. [47] Sotirios Liaskos, Sheila A Mc Ilraith, Shirin Sohrabi, and John Mylopou-los. Integrating preferences into goal models for requirements engineering. In Requirements Engineering Conference (RE), 2010 18th IEEE Interna-tional, pages 135-144. IEEE, 2010. [48] Daniel Markovits. Theories of the Common Law of Contracts. In Edward N. Zalta, editor, The Stanford Encyclopedia of Philosophy. Metaphysics Re-search Lab, Stanford University, winter 2019 edition, 2019. [49] John Mylopoulos, Lawrence Chung, and Brian Nixon. Representing and using nonfunctional requirements: A process-oriented approach. Software Engineering, IEEE Transactions on, 18(6):483-497, 1992. [50] Bashar Nuseibeh, Jeﬀ Kramer, and Anthony Finkelstein. A framework for expressing the relationships between multiple views in requirements speciﬁcation. Software Engineering, IEEE Transactions on, 20(10):760-773, 1994. 27	2104.14110v1.pdf
[51] Balasubramaniam Ramesh and Matthias Jarke. Toward reference models for requirements traceability. Software Engineering, IEEE Transactions on, 27(1):58-93, 2001. [52] Norman Riegel and Joerg Doerr. A systematic literature review of require-ments prioritization criteria. In International Working Conference on Re-quirements Engineering: Foundation for Software Quality, pages 300-317. Springer, 2015. [53] William N Robinson, Suzanne D Pawlowski, and Vecheslav Volkov. Re-quirements interaction management. ACM Computing Surveys (CSUR), 35(2):132-190, 2003. [54] Douglas T. Ross and Kenneth E Schoman. Structured analysis for require-ments deﬁnition. IEEE transactions on Software Engineering, (1):6-15, 1977. [55] Guttorm Sindre and Andreas L Opdahl. Eliciting security requirements with misuse cases. Requirements Engineering, 10(1):34-44, 2005. [56] Siv Sivzattian and Bashar Nuseibeh. Linking the selection of requirements to market value: A portfolio-based approach. In Proceedings of 7th Inter-national Workshop on Requirements Engineering: Foundation for Software Quality (REFSQ 2001), 2001. [57] Stephen A Smith. Contract theory. OUP Oxford, 2004. [58] Ian Sommerville. Integrated requirements engineering: A tutorial. IEEE software, 22(1):16-23, 2005. [59] Axel Van Lamsweerde. Goal-oriented requirements engineering: A guided tour. In Requirements Engineering, 2001. Proceedings. Fifth IEEE Inter-national Symposium on, pages 249-262. IEEE, 2001. [60] Axel Van Lamsweerde, Robert Darimont, and Emmanuel Letier. Managing conﬂicts in goal-driven requirements engineering. IEEE Transactions on Software Engineering, 24(11):908-926, 1998. [61] Axel van Lamsweerde and Emmanuel Letier. Handling obstacles in goal-oriented requirements engineering. Software Engineering, IEEE Transac-tions on, 26(10):978-1005, 2000. [62] Jon Whittle, Peter Sawyer, Nelly Bencomo, Betty HC Cheng, and J-M Bruel. Relax: Incorporating uncertainty into the speciﬁcation of self-adaptive systems. In Requirements Engineering Conference, 2009. RE'09. 17th IEEE International, pages 79-88. IEEE, 2009. [63] Pamela Zave. Classiﬁcation of research eﬀorts in requirements engineering. In Proceedings of 1995 IEEE International Symposium on Requirements Engineering (RE'95), pages 214-216. IEEE, 1995. 28	2104.14110v1.pdf
[64] Pamela Zave and Michael Jackson. Four dark corners of requirements en-gineering. ACM transactions on Software Engineering and Methodology (TOSEM), 6(1):1-30, 1997. 29	2104.14110v1.pdf