Measuring Legislative Complexity

Data as at 12 Dec 2022

Legislative complexity is a growing challenge across the Commonwealth statute book. This page demonstrates how the complexity of Commonwealth Acts made since 1901 can be explored quantitatively.

The page prompts questions such as:

  • Why have cross-references between Acts become so common and why do particular Acts become nodes in complex networks of cross-references? Is there an alternative to such extensive cross-referencing, or is this the best approach to law design?
  • Why do certain areas of lawmaking, such as corporations and financial services, have so many definitions, while others areas of lawmaking include fewer defined terms?
  • What are the consequences of the rapid legislative change present in the modern statute book, including for people who must comply with the law and those who make it?
  • Why has the vocabulary of the average Commonwealth Act grown so much over the past 120 years? Is this a desirable increase in the precision of language or a conceptual explosion in which readers must comprehend more complex prose?
  • What explains the rise of indeterminate terms such as those related to ‘reasonableness’ and ‘good faith’? What does this mean for the roles of courts and administrative decision-makers?

This page is based on the ALRC’s analysis in Complexity and Legislative Design. The page principally uses the As made Acts – Complexity and linguistic data and In force Acts – Complexity and linguistic data data sets. Columns related to cross-references, definitions, amendments per year, vocabulary, conditional statements, and various indeterminate concepts form the basis of the analysis. To identify the relevant columns, please see the Explanatory Note for these data sets.

Cross-references and defined terms

Legislative provisions can be deeply interwoven with one another through cross-references and defined terms. This kind of interweaving, where few provisions of an Act stand on their own, is a common symptom of complexity. Such interconnections mean that in understanding even one section of an Act a person may need to refer to dozens of other sections, both within the Act and across the statute book. These interactions are not always drawn to the reader’s attention, and nor are connections always structured logically.

Cross-references to other Acts

Figure 1 highlights the scale of legislative cross-referencing between in force Commonwealth Acts, with each line indicating a cross-reference and its width indicating the number of references. The size of each point represents the length of the legislation. As is clear in the Figure and following Figures, some Acts are central nodes in the Commonwealth legislative universe, with hundreds or even thousands of cross-references to and from their text. Almost all Commonwealth Acts refer to another Act, and several refer to dozens or even hundreds of Acts. For example, the Income Tax Assessment Act 1997 refers to 160 other Commonwealth Acts, while the Criminal Code Act 1995 refers to 156 other such Acts.

Figure 1: Cross-references between in force Commonwealth Acts

The scale of these interconnections can be significant. Figure 1 shows the number of times the ten most cited Acts are cited by other Acts, as well as the number of times these ten Acts cite other Commonwealth Acts. The Income Tax Assessment Act 1997, for example, refers to another Commonwealth Act on 7,158 separate occasions. The Corporations Act 2001 refers to another Commonwealth Act more than 1,534 times, and is in turn referred to 4,658 times.

In total, Commonwealth Acts refer to other such Acts on 133,537 occasions.

Figure 1: Cross-references to and from the ten most referenced Acts

The approximate number of internal and external references appearing in as made principal Acts has fluctuated significantly from year to year, but has generally remained very high since the 1970s and particularly since the 1980s. Despite the narrow search criteria used to identify potential cross-references, the data generated to create Figure 2 still somewhat overestimates the number of internal and external cross-references. However, the trends and the changes in magnitude nonetheless provide clear evidence of increased use of cross-references. Moreover, the take-off in cross-references in the 1970s more broadly mirrors the rapid growth of Commonwealth lawmaking during and after this decade.

Figure 2: Internal and external cross-references in as made principal Acts

Legislative definitions

Approximately 49,039 definitions appear across all in force Commonwealth Acts, and approximately 13% of all words in these Acts are potentially affected by a definition. Figure 3 shows the number of definitions that appear in Commonwealth Acts, classified according to the subject-matter of the Act in which the definitions appear. The top ten subject areas in the Figure contain 72% of all definitions that appear in Commonwealth Acts.

Figure 3: Number of definitions in Commonwealth Acts by subject-matter

As Figure 4 shows, definitions remain a persistent and significant feature of Commonwealth lawmaking, with many definitions being contained in new principal and amending Acts annually. However, the number of definitions being added to the statute book annually since 2015 is significantly less than in the previous decade.

Figure 4: Number of definitions in as made Commonwealth Acts annually

Legislative change

Legislative change is an inherent and desirable feature of legislation. But change also has consequences for the complexity of legislation. The statute book is ‘an ever-evolving network of complex information that expands organically and is extremely difficult to map’.1 This difficulty is exacerbated when legislation is amended with great frequency, and is particularly problematic when change occurs in a way that does not make clear how existing and new legislation fit together.2 Frequent amendment also results in changes to the existing ‘scheme structures’ which makes both the ‘legislative and procedural arrangements’ associated with the law more complex.3.

Frequent amendment arguably increases the risk that changes will be undertaken ‘without a review of existing provisions, or the legislation as a whole’, which the Commonwealth Attorney-General’s Department suggests can be a cause of complexity.4 Godwin, Brand and Langford note that patchwork legislative amendments result in an ‘inherent risk of incoherence’ that is only ‘exacerbated by legislation that is the product of continuing evolution over a long period of time’.5

As Figure 5 shows, the proportion of Commonwealth Act pages that appear in amending Acts has increased significantly over the past 120 years. More attention is paid to amending existing Acts than to establishing new principal Acts. The first amending Act was passed in 1903, reflecting the fact that the Parliament was occupied with making new principal Acts during the first two years of Federation. The pace of change to existing Acts has accelerated as amending Acts have become a focus of Commonwealth lawmaking.

Figure 5: Proportion of as made Commonwealth Act pages that are amending (1903–2022)

However, amendments to Commonwealth Acts are highly concentrated. Figure 6 plots every Commonwealth Act according to the the number of Acts amending it and the number of years it was in force. Blue points indicate Acts that are still in force, while yellow points represent repealed Acts. All Acts that have been amended by more than 200 Acts are labelled. As is clear in the Figure, amendments are focused on a selected number of Acts. Across all principal Acts, there is no correlation between the age of an Act and the number of amendments made to it. This suggests that amendments may be driven instead by other factors, such as the importance of selected legislation or subject-matter.

Among in force principal Acts, there is a moderate correlation between the age of an Act and the number of amendments made to it (0.27). But this could represent a form of ‘survivor bias’, in which the important Acts last a long time and accrue many amendments because they are important (rather than simply because they are old). The Acts that experience rapid and ongoing change through amendments are at greatest risk of incoherence, and the challenge of managing their complexity remains particularly important compared to more stable legislative schemes.

Figure 6: Number of Acts amending principal Commonwealth Acts based on their age

The true scale of reform can also be explored by analysing the endnotes that appear in legislative compilations of Acts prepared by the Office of Parliamentary Counsel (Cth). These endnotes show the number of distinct amendments made to a text, providing a more granular picture than that provided by Act or page counts. Figure 7 plots each in force principal Act based on the number of distinct amendments made to it, including repeals and insertion of new provisions, and the number of years it has been in force. The Acts most frequently amended per year on average are the Social Security Act 1991 (592 amendments per year), the Income Tax Assessment Act 1997 (559 amendments per year), the Customs Tariff Act 1995 (469 amendments per year), and the Corporations Act 2001 (303 amendments per year). All Acts with more than 5,000 amendments are labelled.

Figure 7: Number of amendments to principal Commonwealth Acts based on their age

Vocabulary and entropy

Greater diversity of words in an Act increases the number of concepts with which a person must engage. These words comprise an Act’s vocabulary, and it is possible to count the number of distinct words that appear in an Act’s text. Figure 8 shows the increase over the past 70 years in the average number of words that comprise an Act’s vocabulary. This increasing linguistic diversity may be driving ‘the inhomogeneity of texts and is an indicator for the number of distinct topics covered by a law’.6

Figure 8: Vocabulary of as made Commonwealth Acts (1950–2022)

Figure 9: Vocabulary of in force Commonwealth Acts

Entropy is a measure of conceptual diversity in a text. Proposed by Shannon in the context of information theory in 1948,7 the ALRC has adapted it as a numerical score that seeks to reflect the diversity of language and concepts used within a legislative text. When an individual is reading an Act with greater entropy, the person will encounter a greater range of concepts than they would encounter if reading a lower entropy text. Entropy can serve as a proxy for the cognitive difficulty which a person will experience in reading a text, given ‘it is more difficult for an individual to understand a set of concepts with high variance than one comprised of homogeneous material’.8 The average entropy of principal Commonwealth Acts has increased over the past 70 years.

For comparison, scholars have noted that the ‘original English texts of Shakespeare have entropies spanning from 9.01 (Julius Caesar) to 9.42 (King Lear); the translated works of Goethe have entropies that range between 9.02 (Wilhelm Meister’s Apprenticeship and Travels, book 6) and 9.42 (Iphigenia in Tauris)’.9

Figure 10: Entropy of as made principal Commonwealth Acts

In force Acts generally have higher entropy than as made Acts because many in force Acts are compilations and include amendments that appear in amending as made Acts. The ten Acts with the highest entropy are listed below.

Figure 11: Top ten most entropic in force Acts

Indeterminate concepts

Words or concepts are potentially indeterminate if their application to a particular set of facts may be subject to multiple interpretations. Consequently, their interpretation by courts may be less predictable than if a more determinate word or concept were used.

As examined by the ALRC in its Background Paper on Complexity and Legislative Design, potentially indeterminate terms are an inevitable and in many ways desirable legislative feature. These concepts are used to ensure that a rule or standard applies appropriately in a variety of cases. Likewise, indeterminate terms may be used in principles-based legislation to reduce other types of complexity, such as prescriptiveness, length, linguistic diversity, and density of obligations and prohibitions.

However, the use of indeterminate concepts can increase the difficulty of interpretation. Although measuring indeterminacy requires a degree of subjective judgment in selecting ‘indeterminate’ words to analyse, it does provide one way of identifying and measuring legislative complexity.

This sections examines the use of ‘reasonableness’ and ‘good faith’ concepts in Commonwealth Acts, as case studies for how indeterminacy can be analysed more systematically across the statute book.

Reasonableness

There are currently 19,858 reasonableness standards across all in force Commonwealth Acts. Reasonableness has emerged as an increasingly important concept in Commonwealth law, as illustrated in Figure 12. The proportion of Acts containing reasonableness standards has regularly been close to or more than 40% over the past 40 years, and hundreds of new reasonableness standards are added annually.

Examples of reasonableness standards include:

  • Any officer may, at any reasonable hour in the day time… (Immigration Restriction Act 1910)
  • …with the degree of care and diligence that a reasonable person would exercise… (Corporations Act 2001)
  • …is reasonably capable of being considered appropriate and adapted… (Biosecurity Act 2015)
  • …satisfy the reasonable requirements of the public… (Copyright Act 1968)
  • …the police officer has reasonable grounds for suspecting… (International Criminal Court Act 2002)

Figure 12: The rise of reasonableness in as made Acts

Reasonableness standards are particularly prominent in several Acts. Figure 13 shows the ten in force Acts that contain the most reasonableness standards. These ten Acts contain 25% of all reasonableness standards currently in force in Acts. They contain 20% of all words.

Figure 13: Top ten in force Acts for reasonableness standards

Good faith

There are currently 974 good faith standards across all in force Commonwealth Acts, making it a much less frequent concept than those related to reasonableness. Figure 14 shows that a small minority of as made Acts include good faith standards, though its use has risen significantly since the 1980s.

Examples of good faith standards include:

  • …not liable to an action or other proceeding for damages for or in relation to an act done or omitted in good faith in performance of the duty… (Telecommunications Act 1997)
  • …make the judgment in good faith for a proper purpose… (Corporations Act 2001)
  • An officer is not liable to any civil or criminal action for doing in good faith, or failing in good faith to do… (Migration Act 1958)
  • the trade mark had been entered in good faith in the Register of Trade Marks… (Wine Australia Act 2013)
  • …negotiate in good faith in relation to the request… (Native Title Act 1993)

Figure 14: The rise of good faith

Good faith standards are highly concentrated in a relatively small number of Acts. Just 25 Acts contain 54% of all good faith standards, despite there being more than 1,225 Acts currently in force. Figure 15 shows the ten in force Acts that contain the most reasonableness standards. These ten Acts contain 35% of all reasonable standards currently in force in Acts, despite having just 10% of all words.

Figure 15: Top ten in force Acts for good faith standards

Conditional statements

Conditional statements have been identified by Li et al as a source of complexity. They suggest that conditional statements create differing legal pathways, usually indicating the presence of exceptions and special cases.10

They argue that, while ‘[a]pplying different rules to different situations is not inherently bad… such “balancing tests” make it more challenging to fully appreciate the consequences of a given piece of legislation. Further, an excessive number of conditional statements might suggest that the underlying rule is faulty, requiring many special cases and exceptions.’

Conditional statements in language include “if”, “except”, “but”, “provided”, “when”, “where”, “whenever”, “unless“ and”notwithstanding“. Conditional statements indicate that a decision needs to be made – a fork in the road is presented to the reader of the legislation. For example, the ‘if’ statement in a particular subsection will either be satisfied or it will not. Many conditional statements actually include multiple conditions (e.g. ‘you must do X if:’, followed by a list of conditions that must be satisfied).

Figure 16: Number of conditional statements in as made Acts (1901–2021)

 

  1. Office of the Parliamentary Counsel (UK), When Laws Become Too Complex: A Review into the Causes of Complex Legislation (2013) 14↩︎

  2. Office of the Parliamentary Counsel (UK), When Laws Become Too Complex: A Review into the Causes of Complex Legislation (2013) 13↩︎

  3. Neville Harris, ‘Complexity in the Law and Administration of Social Security’ (2015) 37(2) Journal of Social Welfare and Family Law 209, 210–11↩︎

  4. Attorney-General’s Department (Cth), Causes of Complex Legislation and Strategies to Address These (2011).↩︎

  5. Andrew Godwin, Vivienne Brand and Rosemary Teele Langford, ‘Legislative Design – Clarifying the Legislative Porridge’ (2021) 38 Corporate and Securities Law Journal 280, 283.↩︎

  6. Bernhard Waltl and Florian Matthes, ‘Towards Measures of Complexity: Applying Structural and Linguistic Metrics to German Laws’ in Rinke Hoekstra (ed), Legal Knowledge and Information Systems (IOS Press, 2014) 153, 160.↩︎

  7. Claude Shannon, ‘A Mathematical Theory of Communication’ (1948) 27 The Bell System Technical Journal 379.↩︎

  8. Michael J Bommarito II and Daniel M Katz, ‘A Mathematical Approach to the Study of the United States Code’ (2010) 389 Physica A 4195, 4199.↩︎

  9. Patrick A McLaughlin et al, ‘Is Dodd-Frank the Biggest Law Ever?’ (2021) 7(1) Journal of Financial Regulation 149 172↩︎

  10. William Li, Pablo Azar, David Larochelle, Phil Hill and Andrew W. Lo, ‘Law Is Code: A Software Engineering Approach to Analyzing the United States Code’ (2015) 10 Journal of Business and Technology Law 297.↩︎