The Redundancy of Originality Revisited
I published The Redundancy of Originality in 1994/1995. See earlier editions of Copyright Reasoning.
Gary Lea then published his response, In Defence of Originality, in 1996. See earlier editions of Copyright Reasoning.
Then, with a bit of delay, I wrote this response which was published (I think) in 2022.
The footnotes are at the end.
In 1995 the Entertainment Law Review published an article I had written entitled The Redundancy of Originality.[1] In that article, which was amongst other things a reaction to the US Supreme Court’s decision in Feist[2] to move away from sweat of the brow copyright, I argued that originality was redundant, that copyright would work better without it, and that compilations of facts should be afforded copyright protection in the same ways as the standard literary work.
In 1996, Gary Lea published a reply to my article. His article, entitled In Defence Of Originality, pushed back against the arguments I had made and, as its name suggests, made the case for originality.
Now, only twenty six years later, seems a good time to revisit the discussion, and that for a couple of reasons. Firstly, the arrival of AI and the ability of AI to create copyright works is challenging some of the established thinking behind copyright. Secondly, and more fundamentally, as we move towards an increasingly digital world and more and more of what we value is non-tangible in nature, it is important that the reasoning behind what we propertise, and how we propertise, is internally coherent and makes sense generally. If the intellectual property reasoning adopted by legislators and courts is wrong – and my contention is that it often is – then that does not bode well for our ability to get the most out of the digital world.
In In Defence Of Originality, Gary Lea summarised the arguments I had made as follows.
(1) the present system of copyright and author's rights only offers patchy, indirect protection of fact products through the selection or arrangement of facts in compilations and so on, and better protection of fact products could be achieved by removing the originality requirement from copyright and author's rights altogether;
(2) the abolition of originality would even go so far as to permit facts themselves to be protected but would not create a monopoly since one can distinguish between fact existence and fact availability - independent gathering of facts would still be possible in most cases - and, in those situations where monopolisation does occur, anti-trust/competition law can correct any imbalance;
(3) originality would not be missed - one may legitimately compare copyright with tangible property which has no equivalent threshold test for conferral of property rights ("propertisation');
(4) the standard argument that non-propertisation of facts leads to a greater flow of information is incorrect: refusing propertisation will diminish fact product production because producers' revenue-gathering ability will be worse off by either having to bear the costs of producers' efforts to find alternative forms of protection or, in the worst case scenario, finding that there are no new fact products since producers have gone out of business as a result of the unrestrained activity of copiers;
(5) the function that originality performs is not properly part of the concept of property in any event since property is a means to an end (in other words, a mechanism for appropriating economic rewards) and not an end in itself;
(6) on this basis, the traditional treatment of copyright as reward is incorrect - the reward is the value of property when realised in the market-place and not the property itself - and, therefore, the value judgment inherent in originality distorts the market by pre-empting market-based valuation;
(7) the distortion argument holds true whichever variant of the originality value judgment one adopts: even under the low threshold UK originality test, for example, the most efficient a producer is in producing its product, the less "skill, labour and judgment' will have been deemed to go into the fact product and the less likely that fact product is to gain copyright protection.
The key point is set out in paragraph 5 of Lea’s summary: “property is a means to an end (in other words, a mechanism for appropriating economic rewards) and not an end in itself”. My assertion was that the primary role of property[3] – both intellectual property and non-intellectual property – in a market economy is to provide a more efficient access to the market for the thing itself.
What do I mean my more efficient? Well, imagine that overnight property in apples has been abolished. If you are an apple farmer, the first thing you will do is to hire security guards to protect your orchards. You will then hire additional security to help you get your apples to market. Apple resellers will in turn adopt additional security to protect the apples they have bought for resale, and so on. The conferring of the status of property – propertisation, for want of a better word – reduces the producer’s costs of bringing the product to market by providing state-sponsored protection (ie. the system of property), and this is true of both intellectual property and non-intellectual property.
It is important to note (because often ignored in discussions about copyright) that access to the market, and the advantages of propertisation, are not limited to the moment of creation of the work. On the contrary, they apply throughout the downstream lifecycle of the work (distribution, resale, licensing), and the actual or anticipated success of the downstream exploitation has a direct impact on the valuation of the work.
Unfortunately, the prevalence of the reward myth has confused this picture. What is the reward myth? It is the theory that holds that copyright is a reward or incentive, and to receive the reward then the work itself must be deserving of the reward. This entails in turn that the work must have value or merit of some kind (such as utility, or to be the result of material investment). Failure to have value (or merit, or utility, or to be the result of material investment) means- according to the reward myth - that the work does not qualify as a copyright work and so is not propertised.
This is a deeply flawed approach. Assessments of value are best performed by the market, not the legal system.[4]and using the legal system to pre-empt the judgment of the market is hardly likely to lead to an efficient allocation of resources.[5]
One of the clearest examples of this is the pseudo-originality required under the Database Directive.[6] Under the Database Directive, a database must meet the requirement of a substantial investment if it is to qualify for the sui generis right.[7] Now, try out this thought experiment. Imagine two companies, Company A and Company B. By chance, both companies develop exactly the same dataset, Dataset X. Company A used a substantial investment to develop Dataset X: the database right applies to its version of Dataset X. Company B has used a less than substantial investment to develop its version: it does not get the database right. Which is the more efficient producer, Company A or Company B? Answer: Company B, because it used less than substantial investment to create its Dataset X. Therefore, the more efficient a producer you are, the less likely it is that your dataset will be protected by the database right.
It is an absurd result, but it is the kind of absurdity that results if you use the legal system, and not the market, to assess value. And it makes no difference to the absurdity whether value is assessed directly by the legal system, or via a proxy for value such as substantial investment.
Gary Lea argued against my analysis, and his counter-argument was made up of five different parts.
The nature of facts
The validity of the economic arguments advanced in ROO
Practical problems of removing originality
Property as a free market mechanism
Leaving the marketplace behind: copyright redux
The nature of facts
Lea points out that an implicit assumption in The Redundancy of Originality is that facts have an inherent utility which warrants propertisation whereas, in fact, a single datum has no inherent use without context. The key problem with this approach is “utility which warrants propertisation”: this is the reward myth again. It appoints the legal system as the arbiter of value (utility, in this case).
A better approach is to let the legal system be agnostic on value as a criterion for propertisation (provided it qualifies as a work, and subject to issues of public policy, see below): maybe thing X has utility, maybe it does not. Let the market decide.
The validity of the economic arguments advanced in ROO
In these sections, Lea reviewed the decision in Feist (and a number of fact-related decisions in other jurisdictions) and concluded that the lack of copyright protection for facts had not resulted in adverse effecta on the market for facts (partially as a result of more extensive use of contractual and technological access controls).
I am not proposing to counter or agree with the arguments put forward by Lea in this section. It seems to me, with the benefit of hindsight, that arguments based on the performance of specific economies or market sectors will always be too simplistic and, because always lacking the counterfactual, are not to be relied on.
I would also agree with Gary Lea that contractual and technological controls can be used to replicate the protection provided by copyright but I would add that these will (as a working assumption) generally increase the costs of making the product available.
Practical problems of removing originality
Lea puts forward three practical matters which militate against removing originality.
First, Lea argues, that that originality and other copyright concepts are firmly interlaced at root: if one removes originality, “then one rapidly reaches the point where what is left is not copyright law at all but an entirely new form of protection.” I think this is putting it too strongly (the UK already has copyright works that do not require originality), but whether removing originality results in a new system or just the modification of an existing system does not really matter. The important thing is to get to a system that makes sense.
Second, that there are hidden costs of propertisation. Lea gives as an example of a hidden cost of propertisation an expectation from consumers that the facts in the compilation are accurate, and this could lead to situations where publishers are liable for supplying incorrect information. In my view, whether a publisher is liable for supplying incorrect information is not something that depends on the propertisation, or lack of propertisation, of the incorrect information.
Thirdly, given that the creation of identical works of fact are possible (for example, two competing lists of dentists in London), giving copyright protection to such works would lead to almost irrebuttable presumptions of copying. Again, I do not see this as a supportable position. For the last 300 years we have had competing mapmakers, and for the last 20 years we have had Database Directive providing quasi-copyright protection under another name, both occurring without excessive litigation.
Property as a free market mechanism
Lea argues that I put forward the proposition that property was “a single, unified and absolute concept which exists as a free market component and which is refused or restricted only in extremis”.
This was not what I had proposed.
A single concept (as opposed to being a bundle of rights): I had not argued this, and I am quite happy to concede that property is, by its nature, a bundle of rights. It makes no difference to my analysis.
A unified concept: on the contrary, I had argued that the property system in market economies is disunified and incoherent. I compared physical property to copyright: they performed very similar functions economically, but were otherwise very different. The latter had a propertisation requirement (originality), the former did not. The latter had a restricted duration, the former had a perpetual duration. More fundamentally, the point I was trying to make was that understanding how property in physical things works in the market would allow us to better understand how copyright works and, as consequence, design more intelligent laws to define and govern copyright.
An absolute concept: Lea makes the point that propertisation is not governed purely by economic considerations, and that morality and public policy are also relevant to the issue of whether to propertise or not. I agree with him: The Redundancy of Originality was explicitly limited to economic considerations and so arguments around morality and public policy did not get much of a look-in. Equally, The Redundancy of Originality did not look at rights-based arguments.
I touch on issues of public policy further below.
Leaving the marketplace behind: copyright redux
In the final section of his article, Gary Lea makes a number of points in relation the fundamentals of copyright.
Firstly, that copyright is a limited property right, with a number of cut-offs such as fair uses and limited duration. This is clearly correct, but dispensing with originality has no direct impact on these cut-offs. It is also worth bearing in mind that copyright’s fair use provisions have their counterparts in the physical world in the form of footpaths, bridlepaths and the right to roam. It is fair to say though, that the arguments put forward for removing originality (in particular market distortion) lead to questioning the nature of the limited term of copyright.
Secondly, that development in arts, education and entertainment is accretive, rather than revolutionary, and that (citing Feist) the ‘the primary objective of copyright is [to encourage] others to build freely upon the ideas and information contained in a work'.[8]
This argument proves too much: if copyright encourages others to build freely upon ideas and information, and if building freely upon ideas and information is a good thing, then a good approach would be to increase the amount of works that are copyright. Removing originality is a good way of doing this.
In any case, dispensing with originality is hardly likely to prevent the continued building on the ideas and information contained in previous works. Those works that were copyright when originality was a requirement would continue to be copyright: the only difference is that some additional works would become copyright too.
Does it make a difference if some of the works are fact-works? Lea submits that “facts are so vital to education that no monopoly or quasi-monopoly [by which Lea means copyright] should even be contemplated”. Except that a) facts have been historically protected by various kinds of intellectual property, as Lea himself points out, and b) with the enactment of the Database Directive we are well beyond the stage of contemplation and c) a right which allows independent creation can only very rarely, if ever, form the basis for monopoly or quasi-monopoly.[9]
Looking back
Looking back at the ROO, one of the things it was trying to do is dismantle the myths and rhetoric that surround copyright and in so doing lay the groundwork for a more coherent approach to copyright (and, by knock-on effect, a more coherent approach to property generally).
Clearly, I was not successful in changing the way the legislators and courts talk about copyright. However, the endeavour is more important than ever as the intellectual economy grows steadily more significant. If the foundations are not clear and sound, it is unlikely that the rest of the edifice will be solid.
Two examples will illustrate the poor reasoning that bedevils legislators and courts alike. In 2021 in Google v. Oracle,[10] the US Supreme Court re-hashed the old “copyright is a monopoly right” myth and as a result came to the wrong conclusion on the facts. According to the US Supreme Court, the basic purpose of copyright fair use is “providing a context-based check that can help to keep a copyright monopoly within its lawful bounds”. [emphasis added]. Having started down this mistaken path, it concluded that Oracle’s copyright was “a lock limiting the future creativity of new programs [and] Oracle alone would hold the key”, neatly ignoring the fact that both Microsoft and Apple had managed to develop their own smartphone operating systems without using the Java API (the work in questions).[11]
More recently the UK Government has consulted on whether the arrival of AI, and the potential for works created solely by AI, should affect the copyright protection afforded to such works. Unusually, the UK has since 1988 provided copyright protection for computer generated works, but in the paper calling for views[12] the UK Government indicated that it was considering removing or reducing the protection received by computer-generated works:[13]
The duration of protection of works would be chosen to reflect the effort or investment put into their creation…. [emphasis added] [14]
This is the reward myth again, this time being applied (bizarrely, and surely mistakenly) as part of a national strategy to increase the use of AI in the economy.[15] Put in plain English, the argument being made is this: AI can produce works more cheaply, therefore AI works should not have copyright protection.
What made the UK Government’s reasoning even stranger was it applied only those works which, in the UK, have a requirement of originality. If a work did not require originality (films, broadcasts, sound recordings) then, according to the UK Government, there was no objection to works created solely by AI.
Nor does the UK (and EU’s) present definition of originality make much sense. The author’s own intellectual creation acts as gatekeeper, excluding from copyright two main categories of works[16]. Firstly, those created by non-humans (eg. works created by AI). Secondly, those works created by humans where, although the end-product is an intellectual artefact, the labour expended in the act of creation is not intellectual. A good example of this is a compilation of facts: for example, a list of all the dentists in London. Producing the list of dentists is not intellectual labour in the pure sense, but the resulting listing clearly is an intellectual artefact.[17]
So we now have, with the author’s own intellectual creation, a definition of originality which is essentially focussed on the means of production. This means that we have a two-tier regime which allocates different forms of legal protection based on how the thing in question is produced. To give some context on this, equivalents for physical products would be: different forms of property for a table, depending on whether the table was made by machine or made by hand; or different forms of property for a loaf of bread, depending on whether the load was baked or steamed. This two-tier regime will apply even if it is impossible to tell whether the work in question is produced by man or machine, and will even be the case if, as may happen, the works are identical.[18]
One of the key sources of this problem is the enduring myth of copyright and the exceptional work. Copyright works may have been rare once upon a time, but they are hardly so anymore. To give a practical perspective, 320 billion emails were sent every day in 2021.[19] Assume that 90% of those were duplicates, and of the remaining 10%, half are long enough to qualify as copyright works. That means that, in email alone, 5840 billion[20] new copyright works are created every year: add in LinkedIn posts, Facebook posts, etc and the number probably doubles. These are all copyright works and they meet the test of author’s own intellectual creation. Are they exceptional? Hardly. Do they “promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries.” [21] Unlikely.
It should not be a surprise that the vast majority of copyright works are trivial and of no consequence because this is exactly how the property system works for physical things. Witness the huge amount of property that ends up in landfill every year. There is no contradiction in having a property system that treats equally things of great value and things of no value (however you define value). The nature of property (at least in relation to physical property and copyright) is that it is immanent in the things that are propertised, and only manifests itself when there is a dispute about the thing in question.
So, taking a step back, if we had to build a property system for the future, on what grounds should we choose to propertise or not propertise? I can think of three primary ones.[22]
Firstly, rivalry.[23] The less rivalrous the thing being propertised, the quicker we should be to propertise, and the greater the extent of property to be granted. This means that copyright, the least rivalrous of all property rights, should have the least restrictions on propertisation.
Secondly, to the extent practical, the duration of similar types of property should be equal. Different terms of protection affect investment with (other things being equal) investment being drawn to those forms of protection that offer longer terms of protection.
The issue here is to ensure optimal allocation of resources and avoid distorting the market unnecessarily. If the comparison is only between physical property (indefinite duration and copyright (generally 70 years pma) then, given a) the non-rivalrous nature of copyright as compared to physical property, b) the difficulty of introducing a shorter term of property for physical property, it would make sense to extend the duration of copyright to match that of physical property.[24]
Thirdly, public policy. As Lea points out, propertisation has always been subject to public policy and consideration of morality. For copyright, this has historically consisted of post-propertisation restrictions such as the fair use provisions which recognise that some copyright works are not just objects of individual property but also of a shared commonwealth. The same can be said of property in real estate, where footpaths, bridlepaths, the right to roam and laws controlling planning commonly trump individual property rights.
Public policy arguments are rarely raised in relation to the propertisation of copyright works, except in relation to the duration of the copyright and the extent to which copyright in facts increase or decrease access to facts and other information.[25] It is hard to object, on public policy grounds, to a form of property which is non-rivalrous and which has little negative impact on the environment (unlike blockchain, for example), but in its consultation on the impact of AI on IP the UK government made an implicit policy argument in relation to copyright: copyright in AI works should be restricted so as to protect the employment of human authors. This is a strangely Luddite argument, particularly as part of a programme designed to maximise the uptake of AI in the UK.[26] Hopefully it will not succeed.
Where does originality fit into this?
It doesn’t.
[1] Sherwood-Edwards, The Redundancy of Originality [1995] 3 ENT. LR 94. Also published a year earlier in IIC: International review of industrial property and copyright law 0018-98551994(25):5, s.658-689.
[2] Feist Publications Inc v Rural Telephone Service Co Inc (1991) 111 S. Ct. 1282.
[3] Primary, at least, for trading purposes. As we will be seen later, the vast majority of copyright works are not intended to be traded .
[4] Markets establish economic value, and that may or may not correspond to aesthetic or utility value. However, even for assessments of aesthetic or utility value the market is a better mechanism than the legal system.
[5] How do patents fit into this – is the requirement for an inventive step not a requirement for value to be established before propertisation is granted? There are two distinctions worth making here. Firstly, it is doubtful whether the assessment of inventive step is a value assessment in the same ways as the assessment of copyright value has historically been made. Secondly, and more importantly, patents impose a restriction on the public domain. Unlike copyright, the fact that you created the same thing independently and without copying is not a defence to an infringement claim. Copyright has no such impact: subsequent creators are entitled to create exactly the same work provided that they do not copy the first work.
[6] Directive 96/9/EC of the European Parliament and of the Council of 11 March 1996 on the legal protection of databases.
[7] The sui generis right being itself a quasi-copyright.
[8] Feist, 369 to 370.
[9] Magill being a rare exception. Radio Telefis Eireann (RTE) and Independent Television Publications Ltd (ITP) v Commission of the European Communities. European Court Reports 1995 I-00743. Magill, a magazine publisher, wanted to publish the BBC and ITV's television programme schedules in its weekly magazines, but under the UK copyright rules the programme schedules were compilations and therefore copyright. The BBC and ITV refused to license the TV schedules to Magill, mainly because to do so would prejudice the sales of the BBC's and ITV's weekly magazines. The decision in Magill relied on the television broadcasting being a restricted activity and the BBC and ITV having a duopoly over the listings for upcoming TV shows. It is hard to see the same conclusion being reached in a world of Prime, Netflix and 100+ TV channels.
[10] Google LLC v. Oracle America, Inc., 593 U.S. 18-956 (2021).
[11] The monopoly myth: that copyright (and particularly copyright in facts) is a monopoly or quasi-monopoly. Monopolies are not bad in themselves. Companies compete and try to increase their market share: we can hardly penalise them when they win enough market share to have a monopoly, and therefore anti-trust rules do not penalise monopolies per se, they penalise abuses of monopoly power. A right, like copyright, which allows for independent creation can only very rarely if ever form the basis for a monopoly or quasi-monopoly, and Monopolies in facts are very rare and generally only sustainable where the copyright owner has privileged access to the relevant facts.
[12] Government response to call for views on artificial intelligence and intellectual property:https://www.gov.uk/government/consultations/artificial-intelligence-and-intellectual-property-call-for-views/government-response-to-call-for-views-on-artificial-intelligence-and-intellectual-property.
[13] For more information on this, see Mark Sherwood-Edwards, 'Consultation on AI and IP legislation: the UK government loses the plot in relation to copyright', (2022) 44 E.I.P.R. Issue 5.
[14] Government response to call for views on artificial intelligence and intellectual property: https://www.gov.uk/government/consultations/artificial-intelligence-and-intellectual-property-call-for-views/government-response-to-call-for-views-on-artificial-intelligence-and-intellectual-property.
[15] See the UK government’s strategy on AI: https://www.gov.uk/government/publications/national-ai-strategy.
[16] There is arguably a third: copyright works which are functional works but not compilations of facts.
[17] In relation to compilations of fact, the EU’s Database Directive has addressed the issue by creating what is effectively copyright in databases under a different label, though the Database Directive reliance on the reward myth makes its present incarnation unsustainable from an economic perspective.
[18] At the same time, but independently, a human author and an AI system each create a poem. The two poems are identical to each other. According to the government’s Proposals 1 and 2, the two poems, though identical and indistinguishable, would attract separate forms of property.
[19] See data at this site: https://www.statista.com/statistics/456500/daily-number-of-e-mails-worldwide/
[20] 10% of 320 is 32. 32/2 = 16. 16 x 365 = 5840.
[21] US Constitution, Art. I, §8, cl. 8, quoted by the US Supreme Court in Google v Oracle.
[22] In The Redundancy of Originality I listed four ingredients for a working property system (ie. once the propertisation threshold had been crossed). They are: the definition of the property (ie. for copyright, is it a work or not?), a method of appropriation, a definition of misappropriation, and a set of antitrust rules.
[23] For a background on rivalry, see https://en.wikipedia.org/wiki/Rivalry_(economics).
[24] This is likely to be easier than reducing the term of physical property.
[25] Addressed in some detail in The Redundancy of Originality.
[26] See the UK government’s strategy on AI: https://www.gov.uk/government/publications/national-ai-strategy.