A The starting point: the Prussian Railway Operators Act

The first ever legislative act on strict liability was introduced by the Prussian state for railways in the year 1838.^[14] The Prussian Railway Operators Act (Preußisches Gesetz über die Eisenbahn-Unternehmungen) not only regulated strict liability for railway companies, but at the same time also provided the first stock corporation provisions.^[15] Notably, and contrary to the experience that legislators lag behind when it comes to technological change, the act came into force only four days after the first train ever went into operation between Berlin and Potsdam.^[16]

§ 25 of the Prussian Railway Operators Act imposed strict liability on railway companies for all damage to persons or property of operating trains: ‘The company is liable to pay damages arising during train transportation to transported goods and persons, as well as to other persons and goods, and the company can only exonerate itself from liability provided that the damage was caused through fault of the claimant or an unavoidable external chance. The dangerous nature of the undertaking as such does not constitute chance in this sense.’^[17]

The mastermind and father of the first provision was no less than Friedrich Carl von Savigny.^[18] This is interesting insofar as Savigny only very briefly mentions cases of strict liability in his ‘Law of Obligations’. He refers to them as obligations ex variis causarum figuris.^[19] Nevertheless, Savigny was not blind to the challenges of the industrial age that lay ahead.^[20] As member of the Council of State (Staatsrat), he convinced the other members to alter the initial draft of the Staatsministerium that was still fault based. In the original draft, the company would only have had to pay damages if an employee had been at fault and no compensation was obtainable from the latter.^[21] Savigny remained unconvinced by this approach, as it would often be impossible for the victim to single out the employee at fault.^[22] Furthermore, the company should also be liable for the ‘faulty condition of the track or the equipment’ or the ‘inevitable nature of the enterprise itself’^[23]. Thus, Savigny argues:

‘So far, it does not seem to have been possible to reduce the fire hazard of the fast-moving steam engines, which in England has frequently been the cause of considerable damage. In such cases no one can be blamed for anything. However, it would be unreasonable for the neighbouring landowners or travellers to let themselves be damaged in this way; rather, it seems just that the company would bear the damage resulting from the inevitable dangerousness of their trade, with the exception of any self-incurred damage, or such damage caused by external unavoidable causes.’^[24]

There are two striking aspects in Savigny’s argumentation: In the operation of railways, damage may occur although ‘nobody can be blamed for it’. However, the liability of the railway company can be ensured by justifying it on the basis of the ‘inevitable dangerousness of their trade’. This, in a nutshell, is the basic idea for installing strict liability, still influential today, for Austrian and German law.

B Enumeration, analogy and blanket clause

So, what has happened from the early days of continental European industrialisation through to the contemporary age of digitalisation? It would seem helpful for a comparative working tort lawyer to address this question by analysing the current status of regulation in the field of strict liability from a methodological standpoint. We must look at three aspects, namely, the almost total absence of any strict liability norms in the civil codes (i); the analogous application of provisions that order strict liability (ii); and the academic attempts to draft blanket clauses (iii).

The German Civil Code from 1900 (Bürgerliches Gesetzbuch, BGB) as well as the Austrian Civil Code dating from 1811 (Allgemeines Bürgerliches Gesetzbuch, ABGB) contain almost no practical relevant strict liability provisions.^[25] § 833 BGB of the German Civil Code is a good example. It deals with a rather specific case, namely, the strict liability for animals which serve no commercial or professional purpose. This is basically the result of the agricultural lobby that successfully pressurised the legislator to amend the BGB in 1908.^[26]

The current legal situation in both Austria and Germany is to be seen against this background: strict liability is regulated in an ever growing number of special statutes outside of the civil law codification, including areas such as rail transportation, cars, airplanes, nuclear power, and environmental pollution in general, etc.^[27] It is on this basis that the German judicatory formulated what is known in German jurisprudence as the principle of enumeration (Enumerationsprinzip).^[28] Behind this somewhat highfalutin formulation lies a simple rule: courts should not create new cases of strict liability by way of analogy.^[29] Proponents of this approach believe in the distinct legislative prerogative of the Bundestag in the field of strict liability. Some German scholars have argued that the cases of strict liability must be predictable, and that the criterion of dangerousness is too vague and uncertain to justify analogy.^[30] These arguments are hardly convincing. I would even go so far as to say that it is a source of embarrassment that German lawyers point to the enumeration principle while at the same time courts were, and still are, eager to develop a vast number of duties of care (Verkehrssicherungspflichten) in fault liability.^[31] The tenacity with which both German mainstream tort law scholars and courts embrace the principle of enumeration is most likely not a matter of adhering to good arguments, but a matter of path-dependency on an all too often repeated standpoint.

The Austrian judicatory took a much more convincing approach, as it accepted that strict liability laws can be applied analogously.^[32] Examples of case law from the Austrian Supreme Court of Justice (Oberster Gerichtshof, OGH) range from the development of a liability of factory owners for emitting dangerous flue gas^[33] or, in cases of a racehorse^[34] in action. The decisive factor for the Austrian courts seems to be the evaluation of the dangerousness and the comparability of an existing specific, strict liability norm in the case at hand.^[35] This is not the place to discuss the subtleties of how to construct and justify new legal rules by way of analogy. It is nevertheless important to mention two general points. Firstly, Austrian judges do not recklessly create new strict liability provisions out of thin air.^[36] They rather cautiously develop new case law. Secondly, creating strict liability by way of analogy has its limitations.^[37] In a world in which legislators tend to create highly detailed regulations, the margin for arguing in favour of an analogy shrinks. One example would be that, according to the German Environmental Liability Act (Umwelthaftungsgesetz), farms with 1,700 fattening pigs are subject to strict liability, whereas those with only 1,699 pigs are not.^[38] By introducing the number 1,700, the legislator used its discretion to shape the scope of the Act. The reasons for choosing this number might be arbitrary – something which may be of concern to constitutional lawyers, but not to civil lawyers. All in all, creating new fields of strict liability by way of analogy can help recalibrate and fine-tune the system. This tool, however, does have limitations.

This leads me to the final methodological observation on how to regulate strict liability: blanket clauses.^[39] The idea behind this approach is to introduce a general clause of strict liability into the civil code. Proposals for fixing the patchwork of the existing special statutes were formulated in the last fifty years for German law,^[40] Swiss law,^[41] Austrian law,^[42] and, in recent years, on a broad basis, by European tort scholars.^[43] None have so far succeeded. In Austria, in particular, an extremely controversial debate ensued.^[44] Without tracing the lines of this conflict here, the two conflicting poles are easy to identify. The opponents of a blanket clause lament the lack of legal certainty, and the vagueness of the concept of dangerousness. The proponents of a blanket clause seek to remove the patchwork, and to establish no less than a just system of risk distribution.

C What about computers?

The enumeration principle, on the one hand, and the idea of a blanket clause, on the other, represent two different concepts as to how the legislator may address the principle question of liability without fault. In a legal system, strict liability is an essential instrument for implementing a powerful risk distribution mechanism. Its importance is likely to increase in the future as it provides a simple and efficient tool for distributing risks for damage without thereby blaming any human agent. Conceptionally, there is no difference to the trains in the 19th century. And Savigny’s characterisation of railroad accidents, namely, that ‘no one can be blamed for anything’ is no less relevant in our times. As I see it, however, one question still remains open for discussion. How adaptable is the principle of ‘dangerousness’ in the age of digitalisation? Can we address and encompass within dangerousness the risks associated with automatisation? According to a statement by Gunter Teubner in a recent conference lecture, risks associated with digital automatisation do not have the same attributes as those cases we refer to when seeking to understand dangerousness in strict liability.^[45] For Teubner, when autonomous systems act, there is no inevitability of damage, a high certainty of incidents as in road accidents, or enormous damage caused by nuclear incident. But how does he know? As yet, the question as to how our perception of risk and the way in which dangerousness of autonomous systems will evolve remains unanswered.

A Systematising bases of defence

Austrian and German law have at least five chief reasons used as a defence in strict liability scenarios.^[46] Historically, the oldest one is höhere Gewalt, vis major, Act of God or the King’s or Queen’s enemies. To juxtapose these terms is obviously highly problematic since doing so frequently triggers very different images, concepts and ideas in the heads of jurists throughout the world.^[47] However, essentially two different ‘forces’ are at stake: God and man. When talking about God in the field of damages, we refer not to the mighty hand of our ‘Lord’, but to the forces of nature, or simply natural hazards: hurricanes, floods, earthquakes, volcanic eruption, etc. By contrast, there are man-made events, such as wars, strikes, crimes, etc. Of course, in times of climate change, some would argue that even natural hazards are triggered by human acts or environmental abuses. Leaving aside these considerations, evidently the idea of what remains within the sphere of the controllable has changed as a result of technological advances. The Austrian jurist, Adolf Exner, illustrated this point in the following way: ‘The old man considered lightning a form of vis major; we domesticated it and let it work for us like a pet.’^[48]

Further reasons for a (total) release or distribution of liability are so-called unavoidable events, fault of the injured or third party, and consent. Not all reasons for defence are available in all strict liability scenarios. One can clearly identify a graduation among the cases based on the weightiness of the reasons for imposing strict liability. Strict liability is excluded by vis major in German law in cases involving trains, cars, water pollution, and environmental liability.^[49] In Austria, for example, strict liability is excluded on the basis of vis major in cases of damage caused by electricity and gas. Furthermore, liability for vehicles is excluded where persons are able to prove that the injury was caused by an unavoidable event. Vis major, or the excuse of an ‘unavoidable event’, is not accepted in Austria and Germany in cases relating to nuclear energy^[50] – this would have been possible under art 9 of the Treaty of Paris^[51] – or for aircraft. Essentially, companies running airplanes and atomic installations cannot excuse themselves from the effects of war, unrest and terrorist attacks. This makes sense, as there must be incentives for such companies to take effective precautions.

By contrast, as a rule of thumb, fault as pertaining to the injured party is accepted in all strict liability scenarios in Austria and Germany.^[52] Contributory negligence can lead to a reduction in the amount of damages awarded or even exclude compensation altogether.^[53] Each of the statutes refers to the General Rule § 254 BGB of the German Civil Code or, in the case of Austria, § 1304 ABGB. Neither does fault of a third party exclude liability; rather, all contributors are jointly liable. Finally, consent may exclude liability.^[54] However, the scope for exemption differs according to statute. This is not the place for a more detailed discussion.

B Force majeure (höhere Gewalt) and inevitable events (unvermeidbare Ereignisse)

It is still necessary to elaborate a little further on the legal construction of force majeure and inevitable events. Contrary to Austrian law, in 2002, the German legislation abandoned inevitable events as a mechanism for excluding strict liability in all relevant constellations, such as the German Road Traffic Act, the German Liability Act or the Environmental Liability Act. I address these reasons in greater detail further on.

But allow me to first discuss höhere Gewalt. Today, höhere Gewalt or force majeure represents the sole exclusionary factor of strict liability in Germany. None of the statutes, however, define what is meant by force majeure. § 7 subsec 2 of the German Road Traffic Act, for example, simply formulates: ‘The duty to compensate is excluded if the accident is caused by force majeure.’ According to well-established German case law, force majeure can be defined as an exceptional external event, and cannot be avoided even with the utmost care.^[55] In eras of technical advance, it is hard to find convincing examples. Naturally, snowstorms, landslides, falling rocks can be dangerous, but in most cases, they are not exceptional. Hurricane Kyrill in 2009 was not considered an act of God by the higher regional court of Celle, because damage could have been avoided by rail closure.^[56] Furthermore, terrorist attacks are frequently not considered as instances of force majeure.^[57] Ostensibly, this standpoint seems counter-intuitive, since terrorist attacks normally occur as an exceptional event in a given society. However, this clearly depends on the circumstances. With a situation of heightened security following terrorist threats, the latter may also (sadly) become a frequent occurrence and thus no longer appear exceptional. By contrast, incidents caused in railway transportation due to suicide are considered as instances of force majeure – despite the fact that approximately 1,000 incidents occur each year in Germany.^[58] German courts argue that such incidents are unavoidable for economic reasons, because the German Bahn is unable to counteract these tragic cases with reasonable care.^[59]

It is especially in this specific field of law that the Roman paroemia omnis definitio in jure civili periculosa est (Dig 50. 17. 202 [Javolenus]) – every definition in civil law is dangerous – is of particular relevance. However, referring briefly to the basic definition of force majeure in German law, it would appear that one aspect may be particularly disturbing, namely, why do German lawyers include elements of liability for misconduct in the definition, and introduce a normative standard of care? Permit me to reiterate: according to well-established German case law, force majeure may be defined as an external event which is both exceptional and cannot be avoided in spite of utmost care being taken. The combination of an objective element in an exceptional event on the one hand, and a subjective element of care on the other, would appear surprising. How can an event be exceptional, when one may contemplate avoiding it? The reverse is far more likely: precisely because it is exceptional, one never considers avoiding it. The aforementioned Viennese scholar, Adolf Exner, published a brilliant monograph on this specific question in 1888.^[60] He opposed the predominant subjective approach to defining vis major as established by the famous commercial lawyer, Levin Goldschmidt.^[61] For Exner, force majeure is an event which, firstly, occurs outside of the sphere of an act causing personal injury or damage to property, and which, secondly, by the nature and force of its occurrence, obviously exceeds the coincidences expected in the ordinary course of life.^[62] Court proceedings should be relieved from unanswerable questions of standard of care, and the injured party should be granted an easy and efficient way to claim damages. Exner’s book, which contains many visionary and thought-provoking passages, ought to be read and discussed again.

The concepts of excluding strict liability on grounds of force majeure and ‘inevitable events’ are intimately connected. In fact, as most Austrian scholars point out, the concept of inevitable events contains the constellation of force majeure.^[63] This is a conceptually sound assessment, as long as one adheres to the subjective element defining force majeure. Historically, at least in Germany, the excuse of ‘inevitable incidents’ was introduced in 1908, partly owing to successful lobbying by the automotive industry.^[64] Industry representatives feared harsh consequences for car owners and consequently – as the argument goes – obstacles to the development of the industry itself.^[65] The concept of inevitable events was developed as a solution to delimit the scope of force majeure. Regulation came in the form of the German Road Traffic Act, and can be found still today in § 9 of the Austrian Act on Liability for Railways and Motor Vehicles (Eisenbahn- und Kraftfahrzeughaftpflichtgesetz, EKHG). § 9 subsec 2 EKHG reads as follows:

‘An event is in particular considered unavoidable if its cause is attributable to the conduct of the person harmed, a third party not employed in the operation or an animal, provided that the operator or keeper, as well as the persons employed with consent over the course of the operation, exerted all due care in accordance with the circumstances of the case; and, furthermore, that the accident was not directly caused by an exceptional operational danger as induced by a third party not employed in the operation, or by an animal.’^[66]

The person operating a car or railway has to prove utmost care in respect to the situation in question.^[67] One acknowledged example of an inevitable event is a child who suddenly runs onto a road full of traffic.

In 2002, the German legislation abandoned the concept of ‘unavoidable events’ in the German Road Traffic Act and Liability Act on the grounds of two arguments.^[68]Firstly, the legislator argued that it is dogmatically unsound to have recourse to an excuse within the system of strict liability based on considerations of standard of care.^[69]Secondly, the position of children and persons in need of help should be strengthened. In view of children’s irrational behaviour, or their inability to act safely in traffic situations, recourse to this excuse should no longer be available. I personally find the reasons given for the recalibration of strict liability convincing. Naturally, one should similarly further reflect on the conceptual basis of force majeure and objectify its elements.