Public Prosecutor v Hue An Li
[2014] SGHC 171

Case Number	:	Magistrate's Appeal No 287 of 2013
Decision Date	:	02 September 2014
Tribunal/Court	:	High Court
Coram	:	Sundaresh Menon CJ; Chao Hick Tin JA; Tan Siong Thye JC (as he then was)
Counsel Name(s)	:	Tai Wei Shyong, Ng Yiwen and Daphne Lim (Attorney-General's Chambers) for the appellant; Akramjeet Singh Khaira and Sonia Khoo Meng (Kelvin Chia Partnership) for the respondent; Zhuo Jiaxiang (Drew & Napier LLC) as amicus curiae.
Parties	:	Public Prosecutor — Hue An Li

2 September 2014

Sundaresh Menon CJ (delivering the grounds of decision of the court):

Introduction

1       This was an appeal against sentence brought by the Public Prosecutor (“the appellant”). The respondent, Hue An Li (“the respondent”), was involved in a tragic vehicular accident when she momentarily dozed off while driving and collided into a lorry. Among other consequences, this caused the death of a passenger in the lorry. The respondent pleaded guilty on 10 September 2013 to a charge of causing death by a negligent act, an offence under s 304A(b) of the Penal Code (Cap 224, 2008 Rev Ed) (“Penal Code”). Two other charges, one of causing grievous hurt by a negligent act (see s 338(b) of the Penal Code) and the other, of causing hurt by a negligent act (see s 337(b) of the Penal Code), were taken into consideration for sentencing purposes. The respondent was sentenced to a fine of $10,000 (in default, five weeks’ imprisonment) and was disqualified from driving for five years from the date of her conviction, ie, from 10 September 2013 (see Public Prosecutor v Hue An Li [2013] SGDC 370 (“the GD”)). The appellant appealed essentially on the basis that a custodial sentence ought to have been imposed on the respondent. Some important questions were raised, particularly in relation to the appropriate sentencing benchmarks and considerations that should guide sentencing decisions in cases of negligent driving which result in death and which are prosecuted under s 304A(b) of the Penal Code (“s 304A(b) traffic death cases”). We accordingly appointed Mr Zhuo Jiaxiang (“Mr Zhuo”) as amicus curiae.

2       At the end of the hearing, we allowed the appeal and varied the sentence to four weeks’ imprisonment. We upheld the five-year disqualification period imposed by the district judge (“the DJ”), save that we ordered it to take effect from the date of the respondent’s release from prison. We also ordered the fine of $10,000, which the respondent has already paid, to be returned to her. We now give the detailed reasons for our decision.

The facts

3       The respondent worked in the surveillance department of Marina Bay Sands Casino. On 14 March 2013, she ended her 12-hour shift at 7.00pm. She took a short nap in her car, a Hyundai Avante, before meeting her friends later that night at East Coast Park. The respondent left East Coast Park at about 6.30am the next morning, dropped her friend off at Pasir Ris and was making her way back to her home at Farrer Park when the accident occurred at around 7.20am. According to the First Information Report, the police were notified of the accident at 7.22am on 15 March 2013.

4       At the material time, the respondent was travelling westwards in the middle lane of the three-lane Pan-Island Expressway. She spotted a slow-moving lorry in the leftmost lane and decided to overtake the lorry, which was travelling at about 60–65 km/h. Video footage provided by a member of the public showed the respondent’s car gradually veering left before its front left collided with considerable force into the right rear of the lorry. The brake lights of the Avante only came on upon impact with the lorry. The collision caused the lorry to rotate in an anti-clockwise direction, hit the leftmost barricade of the expressway and flip. The lorry came to rest on its starboard side in a position between the emergency lane and the leftmost lane. There were nine foreign workers being transported in the rear cabin of the lorry at that time, all of whom were thrown out of the vehicle as a result of the collision. Of the nine, eight were injured, while one was pronounced dead at the scene. The lorry driver and his front passenger were also injured. At the time of the accident, the weather was fine, the road surface was dry, visibility was clear and traffic flow was light.

5       In the respondent’s mitigation plea, her counsel, Mr Akramjeet Singh Khaira (“Mr Khaira”), submitted that the respondent was unable to recall how the collision happened; the only explanation put forward was that she “in all probability, blanked out due to her tired mental state”.^{[note: 1]} It should also be noted that the respondent, in two cautioned statements, said that she had just bought the Avante and was still getting used to it. Records indicated that she had purchased the vehicle two weeks prior to the accident.

The decision below

6       In the proceedings below, the DJ began with the premise that a fine was not necessarily the starting point when sentencing an offender for the offence under s 304A(b) of the Penal Code, and that it was not necessary to establish a “most unusual case” (see [4] of the GD) before the imposition of a custodial sentence might be called for. Whether a custodial sentence ought to be imposed would depend on the nature and extent of the offender’s culpability: the more serious the negligence, the more a custodial sentence would be warranted (see likewise [4] of the GD). The DJ quoted extensively from the decision of the High Court in Public Prosecutor v Ng Jui Chuan [2011] SGHC 90 (“Ng Jui Chuan”), which laid down (at [7]) the following propositions:

(a)     Driving while feeling sleepy was not an offence, let alone an offence of rashness. It might, however, become so if the driver knew that he was in all likelihood going to fall asleep.

(b)     The length of time without sleep that a driver could safely endure was a subjective factor. Some drivers might fall asleep after just ten hours without sleep, while others might be able to drive without posing any danger even after 24 hours without sleep.

(c)     The point at which a person fell asleep was, ironically, a point which he would never be aware of.

7       The DJ agreed that there was no merit in the Defence’s submission that imposing a custodial sentence on the respondent would send the wrong message that employers of foreign workers, who commonly transported them in the rear cabin of lorries, bore no responsibility for the safely of their employees. The DJ noted that in Singapore, employers were allowed, subject to certain requirements, to transport their workers in the rear cabin of lorries (at [7] of the GD).

8       The issue in this case, the DJ stated, centred on the culpability of the respondent. In his view, the aggravating factors were that: (a) the respondent had momentarily blanked out while driving; and (b) the collision had resulted in enormous and tragic consequences (at [8] of the GD).

9       The DJ considered, on the one hand, that the thin skull rule did not apply in criminal cases (citing Public Prosecutor v AFR [2011] 3 SLR 653 (“AFR”)), and thus, a person could not be imputed to intend all the consequences, no matter how remote, of an act done by him on another. However, the DJ also said that “[u]ndeniably, the extent of harm and loss must be taken into consideration by a sentencing court” (at [9] of the GD).

10     In all the circumstances, the DJ concluded that a custodial sentence was not warranted having regard, in particular, to the following considerations:

(a)     The respondent worked in the surveillance department of a casino. This required her to be “mentally alert for long periods of time” (at [10(iii)] of the GD). Due to the nature of her work, which required her to work 12-hour shifts, the respondent “could only meet up with friends during the night and in the wee hours of the morning” (at [10(iii)] of the GD).

(b)     The respondent had just bought the Avante shortly before the accident and was still getting used to it (at [10(iii)] of the GD).

(c)     What happened on the day of the accident was unfortunate as the respondent had “blanked out for a moment due to her tired mental state” (see likewise [10(iii)] of the GD).

(d)     Although the respondent had not had proper sleep for the 24-hour period preceding the accident, she had taken some, albeit insufficient, precautions. In particular, she had taken a short rest in her car after finishing work on 14 March 2013 before meeting her friends later that night (at [10(iv)] of the GD).

(e)     The length of time without sleep that a driver could endure without experiencing adverse effects that might affect his ability to drive was a subjective factor. The evidence did not show that the respondent had made a conscious decision to drive despite knowing that she was very tired and sleepy. In particular, it had not been proved that she knew that she would, in all likelihood, fall asleep at the wheel at the time she decided to drive (see likewise [10(iv)] of the GD).

(f)     The respondent did not break any traffic rules at the material time and was travelling within the speed limit (at [10(v)] of the GD).

(g)     In the aftermath of the accident, the respondent had wanted to call for an ambulance, but was informed that that had already been done. She had also taken the initiative of directing traffic until she was sent to the hospital (at [10(vii)] of the GD).

(h)     The respondent was sincerely remorseful and had not driven since the accident even though her driving licence had not been revoked. She had even made several trips to Hindu temples to pray for the soul of the deceased victim as well as for those who had been injured and all their family members (at [10(viii)] of the GD).

(i)     General deterrence had a limited role in sentencing in s 304A(b) traffic death cases (as defined at [1] above) because “law-abiding persons are apt to be revolted by the prospect of injuring others by their driving and therefore do not need the added disincentive of a criminal penalty to keep them from offending” (at [10(ix)] of the GD, citing Public Prosecutor v Abdul Latiff bin Maideen Pillay [2006] SGDC 245 at [13]). Even if general deterrence was a relevant factor, a deterrent sentence need not always take the form of a custodial sentence because a severe fine might suffice (at [10(x)] of the GD).

(j)     Specific deterrence was of limited relevance in s 304A(b) traffic death cases except where the offender had a bad driving record (at [10(ix)] of the GD).

The submissions on appeal

The appellant’s submissions

11     There were four main strands to the appellant’s submissions in its appeal to this court. First, the appellant focused on the respondent’s level of culpability, which was said to be high. In the appellant’s view, the fact that the respondent had, prior to the accident, just spent 12 hours on work requiring intense concentration should not have been regarded as a mitigating factor; nor, for that matter, should the fact that the respondent’s friends were only available to meet her socially at night. The appellant highlighted that the respondent had driven at a time when she knew she had not slept for a substantial period. This was not a case of momentary inattention as the respondent’s vehicle had drifted into the adjacent lane over the course of a few seconds. The respondent had also decided to undertake the dangerous manoeuvre of overtaking the lorry when she was in a state of considerable fatigue.

12     The second main strand of the appellant’s case focused on the consequences of the accident. It was submitted that unsecured passengers in the rear cabin of a lorry were vulnerable. The appellant noted that it was not uncommon to come across such passengers in Singapore. The appellant then referred to the general principle that in criminal cases, “the accused persons must accept the victims and the consequences that [arise] as a result of the accused persons’ conduct as they occur”^{[note: 2]} [emphasis in original]. The appellant submitted that the enormous consequences which arose as a result of the respondent’s negligence were clearly a relevant consideration. Given that those consequences were reflected in the additional charges that were taken into consideration for sentencing purposes, the sentence imposed on the respondent for the charge proceeded with (viz, the charge under s 304A(b) of the Penal Code) should have been increased to recognise those consequences.

13     The third main strand of the appellant’s arguments concerned the relevant sentencing considerations at play in this case. It was submitted that the relevant interest was not specific deterrence, but rather, general deterrence. The appellant pointed out that drivers who took the wheel while sleep-deprived posed a serious risk both to themselves and to other road users. Sleep deprivation, like alcohol, impaired a person’s cognitive abilities. The appellant submitted that drivers who knowingly drove while sleep-deprived should be sentenced to imprisonment except in unusual cases; only then would other like-minded drivers be suitably deterred.

14     The fourth main strand of the appellant’s submissions concerned the decision of the High Court in Ng Jui Chuan. In the appellant’s view, that decision did not (or could not) stand for the proposition that the Prosecution must prove that the offender knew that he would in all likelihood fall asleep at the wheel before a custodial sentence would be warranted in s 304A(b) traffic death cases. The appellant submitted that Ng Jui Chuan was problematic because of its seeming emphasis of and reliance on a purely subjective test: the requirement to prove that the offender knew that he would in all likelihood fall asleep while driving was a high bar that would almost never be crossed except “on the very rare occasion that a driver admits to this”.^{[note: 3]} The appellant submitted that the inquiry should instead be directed at the offender’s actions prior to taking the wheel and his decision to continue driving despite being conscious that he was feeling or was likely to feel drowsy. References were made in this regard to English and Hong Kong authorities.

The respondent’s submissions

15     As against this, the first submission by Mr Khaira, who likewise acted for the respondent in this appeal, was that a custodial sentence for an offence under s 304A(b) of the Penal Code should be reserved for cases where it could be shown that the offender’s conduct was grossly negligent and bordering upon recklessness. The respondent, it was submitted, was not in this category.

16     Mr Khaira sought to distinguish all the cases cited by the appellant on the basis that they all involved victims who fell within the class of “vulnerable victims”. Mr Khaira submitted that while the sentence might be enhanced to encompass a term of imprisonment in cases involving vulnerable victims, there was no basis for making this a position of general application.

17     Mr Khaira further submitted that for sentencing purposes, it was not relevant that the driving conditions in this case had been ideal at the time of the accident and that the accident had taken place on an expressway. There was no special rule that was especially applicable in such circumstances. The question in each instance, Mr Khaira contended, remained whether the offender had been grossly negligent. In this regard, Mr Khaira submitted that the respondent could not have anticipated what in fact transpired and certainly did not expect to lose control of her car.

18     Mr Khaira sought to uphold Ng Jui Chuan and the propositions set out in that case. He submitted that the fact that the respondent knew that she had not had proper sleep for the 24-hour period preceding the accident did not show that she was unfit to drive, nor did it manifest a conscious decision on her part to drive despite knowing that she was unfit to do so. Those were matters falling within the appellant’s burden of proof, which, Mr Khaira contended, had not been discharged.

19     Mr Khaira also took issue with the assertion that the respondent must have been feeling drowsy before the collision. In this regard, he cited Criminal Liability of Drivers Who Fall Asleep Causing Motor Vehicle Crashes Resulting in Death or Other Serious Injury: Jiminez (Final Report No 13, October 2010), a report by the Tasmania Law Reform Institute, which discounted studies suggesting that sleep did not occur spontaneously without warning. The report contended, instead, that there was a high degree of variation between sleep-deprived individuals, and that people were generally poor at predicting that they were likely to fall asleep.

20     Mr Khaira further submitted that a distinction had to be drawn between cases of drink-driving and cases of driving while feeling sleepy or being sleep-deprived (“sleepy driving”). He pointed out that the courts had consistently take a stern view of the former, but not the latter. Drink-driving was an offence per se, whereas sleepy driving was not. Mr Khaira submitted that general and specific deterrence had limited roles in traffic offences that were prosecuted under s 304A(b) of the Penal Code, and that even if deterrence were a relevant factor, a heavy fine would usually be sufficient.

21     As to the consequences of the accident in the present case, Mr Khaira argued that the thin skull rule did not apply in criminal cases. Fortune or misfortune, he submitted, should not play an unduly significant role in determining the culpability of and the punishment befitting an offender in a s 304A(b) traffic death case.

22     Finally, Mr Khaira submitted that neither the statutory framework nor the case authorities mandated that the taking of offences into consideration for sentencing purposes must invariably lead to a more severe sentence for the offences in the charges proceeded with. In any event, the offences taken into consideration should hold little, if any, weight if they arose out of the same culpable act as the offences in the charges proceeded with.

The amicus curiae’s submissions

23     The amicus curiae commenced his submissions by distinguishing between acts of rashness and acts of negligence. He noted that “negligence is a unique offence … [that] is made out despite there being no intention to cause harm”.^{[note: 4]} He referred (among other cases) to Lim Poh Eng v Public Prosecutor [1999] 1 SLR(R) 428 (“Lim Poh Eng”), where the High Court declined to distinguish between civil and criminal cases of negligence on the basis that it would create an intermediate standard, and held that the distinguishing feature between criminal and civil negligence was instead to be found in the standard of proof.

24     The amicus curiae noted that the High Court’s approach in Lim Poh Eng had been criticised by an academic and had not been adopted by the Indian Supreme Court. The question of whether criminal liability should attach to what was in essence the equivalent of civil negligence had, the amicus curiae observed, “attracted serious academic debate”,^{[note: 5]} and there appeared to be no equivalent of s 304A of the Penal Code (at least in the way it had been understood and applied in Singapore) in much of the common law world. In this regard, the amicus curiae referred to Prof Ashworth’s comment that criminalising negligence moved away from advertence to the risk of harm ensuing upon one’s actions as the foundation of criminal responsibility (see A Ashworth & J Horder, Principles of Criminal Law (Oxford University Press, 7th Ed, 2013) at p 182).

25     The amicus curiae observed that premeditation and malice were, by definition, absent in s 304A cases, and therefore, considerations of prevention and specific deterrence were “naturally of limited relevance”^{[note: 6]} for sentencing purposes in these cases. The amicus curiae did accept, however, that there was a contrary academic view to the effect that general deterrence was a valid sentencing consideration. General deterrence, he noted, was certainly “a strong … consideration in sentencing”^{[note: 7]} for drink-driving offences. The amicus curiae concluded that general deterrence should therefore equally be a relevant sentencing consideration for offences arising out of sleepy driving. This, however, did not necessarily mean that a custodial sentence would be warranted.

26     The amicus curiae noted that sleepiness was a major cause of road accidents in many jurisdictions and had been the subject of study in many countries, including the US, the UK and Australia. He also pointed out that lack of sleep was but one of the leading causes of sleepiness, the others being the human circadian rhythm, monotony, medication, alcohol and sleep disorders.

27     The amicus curiae noted that prior to the amendments made to the Penal Code in 2008 by the Penal Code (Amendment) Act 2007 (Act 51 of 2007) (“the 2007 Penal Code Amendment Act”), the precedents were uniform in declaring that the mere fact that death had been caused by negligent driving which constituted an offence under the then version of s 304A (viz, s 304A of the Penal Code (Cap 224, 1985 Rev Ed) (“the 1985 revised edition of the Penal Code”)) did not, without more, justify the imposition of a custodial sentence; instead, a fine would be sufficient in most cases (citing Public Prosecutor v Gan Lim Soon [1993] 2 SLR(R) 67 (“Gan Lim Soon”)). At the very least, callousness had to be shown to warrant the imposition of a custodial sentence (citing Public Prosecutor v Teo Poh Leng [1991] 2 SLR(R) 541 (“Teo Poh Leng”)). The amicus curiae further noted that pursuant to the amendments made in 2008 by the 2007 Penal Code Amendment Act (“the 2008 Penal Code amendments”), s 304A of the 1985 revised edition of the Penal Code (“the old s 304A”) was bifurcated into two limbs, with the negligence limb being punishable with imprisonment of up to two years and the rashness limb, with imprisonment of up to five years. The amicus curiae submitted that although decisions post-dating the 2008 Penal Code amendments were ambivalent about the continued validity of the Gan Lim Soon position that a fine would be sufficient in most cases of death caused by negligent driving, in his view, a custodial sentence would not ordinarily be appropriate.

28     Canada and Australia, the amicus curiae noted, both required something more than negligence before a person could be found guilty of an offence similar to the offence under s 304A(b) of our Penal Code. In this regard, both jurisdictions required some sort of mens rea that went beyond the civil standard of negligence. Hong Kong and the UK took a more serious view of sleepy driving, with both jurisdictions holding that this was a significant aggravating factor. The amicus curiae also criticised Ng Jui Chuan, primarily on the basis that it was questionable whether a person would never be aware of the point at which he fell asleep.

The issues raised in this appeal

29     We considered the following issues in this appeal:

(a)     What is the distinction between rashness and negligence in the context of s 304A of the Penal Code?

(b)     What is the default punitive position for negligent driving which constitutes an offence under the negligence limb of s 304A (ie, under s 304A(b))?

(c)     On the facts of this case, what is the appropriate sentence to impose on the respondent? In particular, should a custodial sentence be imposed on her?

(d)     When is prospective overruling justified?

What is the distinction between rashness and negligence in the context of s 304A?

30     As regards the first of the above-mentioned issues, s 304A of the Penal Code, as it stands today, provides as follows:

Causing death by rash or negligent act

304A.    Whoever causes the death of any person by doing any rash or negligent act not amounting to culpable homicide, shall be punished —

(a)    in the case of a rash act, with imprisonment for a term which may extend to 5 years, or with fine, or with both; or

(b)    in the case of a negligent act, with imprisonment for a term which may extend to 2 years, or with fine, or with both.

31     Prior to the 2008 Penal Code amendments, this was how s 304A read:

Causing death by rash or negligent act

304A.    Whoever causes the death of any person by doing any rash or negligent act not amounting to culpable homicide, shall be punished with imprisonment for a term which may extend to 2 years, or with fine, or with both.

32     The present appeal does not, strictly speaking, raise the issue of the distinction between rashness and negligence. We nonetheless explore this issue for two reasons. First, given that the 2007 Penal Code Amendment Act expressly bifurcated the s 304A offence and separated the applicable penal consequences for offences involving rashness on the one hand and offences involving negligence on the other, it seems to us appropriate to consider the significance to be accorded to this change. We note also that the maximum imprisonment sentence for the rashness limb of s 304A was increased by that same Act from two to five years. Second, the interaction between s 304A(a) and s 304A(b) is an important question, and this case presents us with the opportunity to consider this more closely.

33     The starting point, of course, is case law prior to the 2008 Penal Code amendments. In Public Prosecutor v Poh Teck Huat [2003] 2 SLR(R) 299 (“Poh Teck Huat”), Yong Pung How CJ cited (at [17]) Teo Poh Leng, which in turn cited Balchandra Waman Pathe v The State of Maharashtra (1967) 71 Bombay LR 684 for the following propositions:

(a)     Culpable rashness was “acting with the consciousness that the mischievous and illegal consequences may follow, but with the hope that they will not, and often with the belief that the actor has taken sufficient precautions to prevent their happening” [emphasis added] (see Teo Poh Leng at [7]).

(b)     Culpable negligence was “acting without the consciousness … that the illegal and mischievous effect will follow, but in circumstances which show that the actor has not exercised the caution incumbent upon him and if he had[,] he would have had the consciousness” [emphasis added] (see likewise [7] of Teo Poh Leng).

34     We note that this distinction between rashness and negligence does not inexorably follow from a plain reading of the old s 304A. That provision is susceptible to at least two plausible interpretations:

(a)     First, that the distinction between rashness and negligence is a dichotomous difference in kind, with the word “or” in the provision being purely disjunctive. For ease of reference, this is referred to as “the dichotomous interpretation”.

(b)     Second (and alternatively), that the distinction between rashness and negligence is a non-dichotomous difference of degree, with the word “or” in the provision being at least partially conjunctive. Under this interpretation, “rashness” and “negligence” shade into each other, and, for a certain subset of cases, are in fact synonyms. This is referred to as “the non-dichotomous interpretation”.

35     The two approaches may be represented graphically as follows:

(a)     The dichotomous interpretation

(b)     The non-dichotomous interpretation

36     The two characterisations have important implications for sentencing. Under the dichotomous interpretation, “negligence” and “rashness” demarcate separate offences with different starting points for sentencing. In contrast, under the non-dichotomous interpretation, “negligence” and “rashness” demarcate one and the same offence, and are merely labels for different levels of culpability.

37     It seems to us that Poh Teck Huat interpreted the old s 304A dichotomously. In that case, Yong CJ made the following observations:

(a)     Advertence to the potential risks that might arise from one’s conduct was the dividing line between negligence and rashness (at [17] of Poh Teck Huat; see also [33] above).

(b)     “Negligence” and “rashness” were instances of “disjunctive language” (at [18] of Poh Teck Huat). This necessarily meant that “negligence” and “rashness” could never be synonymous.

(c)     The distinction between the two was of particular importance when the trial judge examined the facts to determine if the charge had been made out (at [19] of Poh Teck Huat).

38     There are, however, also some comments by Yong CJ in Poh Teck Huat which ostensibly suggest that he did not fully subscribe to the dichotomous interpretation. These comments were as follows:

19    … It [ie, the distinction between negligence and rashness] however loses some of its significance at the sentencing stage. At this stage, the concern is to ensure that the sentence reflects and befits the seriousness of the crime. To do so, the court must look to the moral culpability of the offender.

20    In examining the moral culpability of an offender, the scale would start with mere negligence and end with gross recklessness. However negligence does not end nicely where rashness begins and there is a certain measure of overlap. As such, it is possible for the moral culpability of an offender who has committed a rash act to be akin to that of a negligent act.

39     In our view, what is significant about the above comments by Yong CJ, which refer to the moral culpability of the offender, is that they are directed at an inquiry undertaken at the sentencing stage. This does not detract from negligence and rashness being dichotomous jurisprudential concepts at the liability stage. In our judgment, Yong CJ was in fact fully committed to the dichotomous interpretation, at least for the purposes of determining liability.

40     In any case, any room for doubt was removed by the 2007 Penal Code Amendment Act. This bifurcated the old s 304A into s 304A(a) (the rashness limb) and s 304A(b) (the negligence limb), with the former carrying a maximum imprisonment term of five years and the latter, a maximum imprisonment term of two years. The disparate sentencing regimes, with different maximum imprisonment sentences, clearly indicate that rashness and negligence are dichotomous concepts. By definition, there cannot be two disparate sentencing regimes for one and the same offence; this necessarily rules out the non-dichotomous interpretation.

41     There remains the question of whether the line drawn in Poh Teck Huat between negligence and rashness, which entails advertence to the potential risks that might arise from one’s conduct, remains good law after the 2008 Penal Code amendments.

42     In this regard, one observation may safely be made. Neither rashness nor negligence is a strict liability offence; for both, it must be shown that the offender’s conduct fell below a certain standard. There are two ways of drawing a line between rashness and negligence:

(a)     having different objective standards for rashness and negligence, with the former requiring a greater degree of culpability on the offender’s part than the latter (“the ‘two standards’ approach”); or alternatively

(b)     having the same objective standard for rashness and negligence, but requiring mens rea for the former.

43     We consider the “two standards” approach to be untenable. There are already challenges in applying a single standard, and these difficulties will only be compounded if two standards are applied. The most commonly-invoked standard for negligence is that of the reasonable person: ie, negligence is regarded as the omission to do something which a reasonable person would do, or the doing of something which a reasonable person would not do (see Blyth v The Co of Proprietors of the Birmingham Water Works (1856) 11 Ex 781 at 784). The “reasonable person” standard has been the subject of trenchant criticism: see, for instance, M Moran, Rethinking the Reasonable Person: An Egalitarian Reconstruction of the Objective Standard (Oxford University Press, 2003), where the writer opines (at p 5) that the utility of an idealised person is doubtful where there are divergences between the default characteristics of the reasonable person and the characteristics of the actual person concerned. Notwithstanding these criticisms, the “reasonable person” standard remains widely used. Our point here in noting the criticisms that have been directed at this standard despite its being a well-established and widely-used one is to highlight the danger of compounding the difficulties inherent in using one such standard by adding a second one.

44     It follows that mens rea should be the dividing line between negligence and rashness, and this is in fact reflective of the law that has hitherto been applied. But, what is the requisite mens rea in this context? At the outset, intention can be ruled out: if intention were the dividing line, there would be no difference between the provisions criminalising rashness and those criminalising the intentional infliction of harm (eg, s 321 of the Penal Code).

45     In our judgment, awareness of the potential risks that might arise from one’s conduct ought, in general, to be the dividing line between negligence and rashness. For both negligence and rashness, the offender would have fallen below the requisite objective standard of the reasonable person. The harsher sentencing regime for rashness is justified on the basis that the offender was actually advertent to the potential risks which might arise from his conduct, but proceeded anyway despite such advertence. This is essentially a restatement of the definitions of “rashness” and “negligence” enunciated in Poh Teck Huat, which, in our judgment, remain good law. In short, advertence to risk will generally be an essential element of rashness. We have qualified this statement of principle as one that would “generally” apply because there remains a class of cases where the risks may be said to be so obvious had the offender paused to consider them that it would artificial to ignore this fact (see also our comments at [49]–[55] below). We leave it open as to whether advertence to risk must actually be proved before a finding of rashness can be made in this class of cases (“‘blatantly obvious risk’ cases”). In our judgment, it would be best to develop this issue by case law rather than by a pre-emptive statement of principle.

46     We note that advertence to risk is also the dividing line that is used in a similar context in England. In Regina v G and another [2004] 1 AC 1034 (“R v G”), which involved the offence under s 1 of the Criminal Damage Act 1971 (c 48) (UK) (“the 1971 UK Criminal Damage Act”) of damaging property “being reckless as to whether any such property would be damaged”, the House of Lords held that a person was reckless if he chose to act despite being aware of the risk of harm (at [41]). In so holding, the House of Lords reaffirmed the decision of the English Court of Criminal Appeal in Regina v Cunningham [1957] 2 QB 396 (“Cunningham”), and expressly overruled its earlier decision in Commissioner of Police of the Metropolis v Caldwell [1982] 1 AC 341 (“Caldwell”). Cunningham defined “recklessness” as the state of mind where “the accused has foreseen that the particular type of harm [that eventuates] might be done and yet has gone on to take the risk of it” (at 399). This was not followed in Caldwell, where Lord Diplock opted for an objective approach. His Lordship opined that “recklessness” included “failing to give any thought to whether or not there is any such risk [of harmful consequences resulting from one’s acts] in circumstances where, if any thought were given to the matter, it would be obvious that there was” (at 354), and further, that this was to be measured against the mind of the “ordinary prudent individual” (likewise at 354).

47     In R v G, Lord Bingham of Cornhill gave four reasons for overruling Caldwell:

(a)     It was a salutary principle that any conviction of a serious crime should depend on proof of a culpable state of mind (at [32]).

(b)      Caldwell was capable of leading to obvious unfairness. On the facts of R v G, it was neither moral nor just to convict the offenders, two children aged 11 and 12 respectively, on the strength of what someone else would have apprehended if the children themselves had no such apprehension (at [33]).

(c)      Caldwell had come under criticism from academics, judges and practitioners (at [34]).

(d)     Lord Diplock’s majority judgment in Caldwell (which was endorsed by two other members of the House of Lords) was a misinterpretation of the word “reckless” in s 1 of the 1971 UK Criminal Damage Act. Parliament had not intended to alter the definition of “reckless” when it passed that Act (at [29] and [35]).

48     As noted above (at [45]), we too conclude that advertence to risk is the touchstone that, in general, will distinguish rashness from negligence.

49     Before we conclude this part of our analysis, we refer briefly to the concurring speech of Lord Rodger of Earlsferry in R v G, where his Lordship noted that some offences might call for a broader conception of recklessness. Reckless driving was cited as one instance where the law “may properly treat as reckless the man who acts without even troubling to give his mind to a risk that would have been obvious to him if he had thought about it” (at [69]). This observation brings us back to the reservation which we made at [45] above about the “blatantly obvious risk” cases, which might not be neatly dealt with by drawing a rigid dividing line between rashness and negligence at the point of subjective advertence to the risk in question.

50     In this regard, we note that ss 64 and 66 of the Road Traffic Act (Cap 276, 2004 Rev Ed) (“RTA”) use “dangerous” in conjunction with “recklessly”:

Reckless or dangerous driving

64.—(1) If any person drives a motor vehicle on a road recklessly, or at a speed or in a manner which is dangerous to the public, having regard to all the circumstances of the case, including the nature, condition and use of the road, and the amount of traffic which is actually at the time, or which might reasonably be expected to be, on the road, he shall be guilty of an offence and shall be liable on conviction to a fine not exceeding $3,000 or to imprisonment for a term not exceeding 12 months or to both and, in the case of a second or subsequent conviction, to a fine not exceeding $5,000 or to imprisonment for a term not exceeding 2 years or to both.

…

Causing death by reckless or dangerous driving

66.—(1) Any person who causes the death of another person by the driving of a motor vehicle on a road recklessly, or at a speed or in a manner which is dangerous to the public, having regard to all the circumstances of the case, including the nature, condition and use of the road, and the amount of traffic which is actually at the time, or which might reasonably be expected to be, on the road, shall be guilty of an offence and shall be liable on conviction to imprisonment for a term not exceeding 5 years.

…

51     It is evident that the offence under s 66(1) of the RTA carries the same maximum imprisonment term of five years as the offence under s 304A(a) of the Penal Code, the only difference being that there is a power to impose a fine under the latter provision but not under the former provision. We make some brief observations, although these should be understood as tentative since we did not have the benefit of arguments on this. At one level, it might be thought that “recklessly” (in ss 64 and 66 of the RTA) and driving “without due care and attention” or “without reasonable consideration for other persons using the road” (in s 65 of the RTA) are synonyms for, respectively, rashness and negligence in the Penal Code. After all, if this were not the case, drivers might potentially be subject to four separate standards, which is prone to be confusing. On the other hand, we note that the terms used are different. We also note that “recklessly” in ss 64 and 66 of the RTA is immediately followed by “or at a speed or in a manner which is dangerous” [emphasis added]. Thus far, no Singapore case has squarely addressed the question of whether the “dangerous driving” limb of ss 64 and 66 is to be read conjunctively with the “recklessness” limb of these two sections, or whether the two limbs are to be read disjunctively.

52     It is also significant that “dangerous driving” in ss 64 and 66 of the RTA is framed in terms that suggest an objective standard directed at the facts surrounding the offender’s driving and its effect on other road users. The question seems to be whether the offender was driving at a speed or in a manner that was dangerous to the public. On this basis, drivers who might not, strictly speaking, be found to be reckless could nevertheless face the same maximum punishment as reckless drivers if they are found to have driven dangerously. “Dangerous driving” could then potentially encompass factual scenarios where the driver had not directed his mind at all to an obvious risk, and had thereby created a source of danger. Having said that, we find it difficult at present to imagine the circumstances in which a driver could drive in a manner that is in fact dangerous to road users either because of his speed or manner of driving and yet be wholly unaware of the risks posed by his doing so.

53     We note further that s 64(1) of our RTA is similar to s 11(1) of the Road Traffic Act 1930 (c 43) (UK). In Regina v Spurge [1961] 2 QB 205, a five-judge panel of the English Court of Criminal Appeal interpreted “dangerous” in that subsection literally to mean driving in a manner which endangered the public, and did not refer to any requirement of advertence to risk. The court did not, and could not, consider the issue of how that subsection would interact with a provision akin to s 304A of our Penal Code because there was at that time (and there still is) no English equivalent of our s 304A. The closest parallel offence in England then was (and still remains) the offence of manslaughter, which is made out based on a single standard (namely, that of gross negligence), and not on the negligence-rashness dichotomy in s 304A.

54     We also note that in England, the Road Traffic Act 1988 (c 52) (UK) dispenses with recklessness, with dangerous driving and careless and inconsiderate driving being the only two touchstones for criminal liability. Section 2A of that Act explicitly and exhaustively defines “dangerous driving” in the following manner:

2A    Meaning of dangerous driving

(1)    … [A] person is to be regarded as driving dangerously if (and, subject to subsection (2) below, only if) —

(a)    the way he drives falls far below what would be expected of a competent and careful driver, and

(b)    it would be obvious to a competent and careful driver that driving in that way would be dangerous.

(2)    A person is also to be regarded as driving dangerously … if it would be obvious to a competent and careful driver that driving the vehicle in its current state would be dangerous.

…

This seems to point to an objective consideration of the circumstances, regardless of any question of subjective advertence to the risks in question.

55     We do not think it would be helpful for us to say more without the benefit of arguments. This is a vexed question, and we leave the issue of non-advertence to risk and the requisite standard to be met under ss 64 and 66 of the RTA to be decided on another occasion when we have the benefit of full arguments.

What is the default punitive position for negligent driving which constitutes an offence under s 304A(b)?

Gan Lim Soon and the effect of the 2008 Penal Code amendments

56     We turn now to consider the default punitive position for negligent driving which constitutes an offence under s 304A(b) of the Penal Code. The starting point for our analysis is Gan Lim Soon ([27] supra), where Yong CJ drew a distinction between rashness and negligence: for the former, imprisonment would be warranted; while for the latter, “it would be sufficient in most cases to inflict a fine” (at [10]). In Teo Poh Leng ([27] supra), the High Court held that callousness would have to be shown before a custodial sentence would be imposed. These two cases were decided before the 2008 Penal Code amendments.

57     In Public Prosecutor v Lee Kao Chong Sylvester [2012] SGHC 96 (“Sylvester Lee”), a High Court case decided after the 2008 Penal Code amendments, the court cited Gan Lim Soon for the proposition that negligent driving which constituted an offence under s 304A(b) would normally attract a punishment of a fine (at [17] and [19]), but did not consider the effect of the said amendments. Subsequently, in Public Prosecutor v Wong Yew Foo [2013] 3 SLR 1198 (“Wong Yew Foo”), the High Court considered the effect of the 2008 Penal Code amendments and commented (at [27]) that “it is highly questionable if the starting point as regards sentencing for the s 304A(b) offence remains a fine and that only a ‘most unusual case’ would warrant the imposition of a custodial sentence”. The court did not, however, go on to offer definitive guidance for future cases.

58     In Public Prosecutor v Kwong Kok Hing [2008] 2 SLR(R) 684 (“Kwong Kok Hing”), which involved a charge of attempted culpable homicide under s 308 of the 1985 revised edition of the Penal Code, the Court of Appeal held that: (a) a sentence close to or at the statutory maximum would be imposed for conduct that was amongst the worst conceivable for that particular offence (at [44]); and (b) it was incumbent on a sentencing court to take note of the statutory maximum sentence and determine precisely where a particular offender’s conduct fell within the spectrum of punishment devised by Parliament (likewise at [44]).

59     We agree with the above propositions in Kwong Kok Hing. Extrapolating from them, the default punitive position for a particular offence must be determined with reference to the punishment at the two ends of the spectrum. For the offence under s 304A(b) of the Penal Code, the two ends are, respectively, two years’ imprisonment and a nominal fine. We recognise that a fine and a term of imprisonment are, for most intents and purposes, incommensurate: most offenders would rather pay a fine than spend a period of time in jail. It is therefore impossible to pinpoint a precise midpoint between a two-year imprisonment term and a nominal fine.

60     Given that a period of incarceration and a fine are incommensurate, there are, in our judgment, two possible approximate midpoints: a large fine, and a brief period of incarceration. It seems clear to us that following the 2008 Penal Code amendments, the position laid down in Gan Lim Soon is no longer tenable, given the bifurcation of the old s 304A into two limbs. Parliament could have chosen to retain the language of the old s 304A and to merely increase the statutory maximum term of imprisonment to five years. That would have left the Gan Lim Soon position untouched. Instead, Parliament chose to bifurcate the old s 304A and make it clear that causing death by negligence and causing death by rashness would each have its own sentencing range.

61     In the premises, we are satisfied that the starting point for sentencing in a s 304A(b) traffic death case is a brief period of incarceration for up to four weeks (see [133] below). This does not mean that a sentence of imprisonment will be imposed in every s 304A(b) traffic death case. This is because the court must examine all the circumstances of each individual case as well as any aggravating and/or mitigating factors to determine the gravity of the particular offender’s conduct before deciding what the appropriate sentence should be.

62     We pause at this juncture to make another observation. The dichotomous sentencing regimes for the negligence and the rashness limbs of s 304A entail the possibility of a conviction under the rashness limb carrying a more lenient sentence than a conviction (of a different person in different circumstances) under the negligence limb. This is because it is entirely plausible for a person to be advertent to the potential risks that might arise from his conduct, and yet be less culpable than another who is oblivious to such risks. We emphasise that it is the presence of mitigating and/or aggravating factors, and not merely the categorisation of an offender’s conduct as rash or negligent, that will be determinative of the actual penal consequences that follow upon the commission of a s 304A offence.

The propositions laid down in Ng Jui Chuan

63     Against that background, we turn to the High Court’s decision in Ng Jui Chuan, which involved a driver who fell asleep at the wheel after having gone for 22 hours without sleep. In that case, the offender collided into two pedestrians, killing one and injuring the other. It was said in that case:

7    … Driving when one is tired or sleepy is not an offence, let alone an offence of rashness. It may become so if it had been proved that the tired driver knew that he was in all likelihood to fall asleep at the wheel and yet he drove. … The DPP submitted that the respondent had been devoid of sleep for 22 hours and on the strength of that, he ought to have told himself that he was in no position to drive and should not have driven. The DPP further submitted that the fact that the respondent did continue to drive was a strong factor indicating the rashness of his conduct. What was overlooked in this argument was that the length of time without sleep is a subjective factor. Some people will fall asleep at the wheel if they are devoid of only 10 hours of sleep, some can drive with no danger even after 24 hours without sleep. In this case, when the respondent started off from Yishun, he was only feeling tired but there was nothing to indicate that he clearly ought not to drive. The point of importance therefore occurred at the junction between Upper Thomson Road and Sin Ming Avenue, where the respondent felt sleepy at the wheel and slapped himself on the neck to stay awake. However, it must be remembered that he was, at that time, only five minutes away from home and he thought he would be able to make it back home without incident. … The mental state in the circumstances of the respondent may indicate an element of negligence, but I agree with the trial judge that in the totality of the circumstances of the case, they fall short of rashness. …

8    The DPP submitted that in any event, even if the charges were rightly reduced, the trial judge ought to have imposed a custodial sentence. In my view, on the facts of the case, that would amount to saying that that there was no distinction between a rash act and a negligent one. …

64     It should be noted that in Ng Jui Chuan, the Prosecution proceeded with a charge under s 304A(a) of the Penal Code of causing death by a rash act, but the trial judge subsequently amended the charge to one of causing death by a negligent act under s 304A(b); that was also the basis on which the case was dealt with on appeal to the High Court. Ng Jui Chuan made two propositions, which may be paraphrased as follows. First, rashness was made out if the offender knew (presumably when he started driving) that he could – and in all likelihood would – fall asleep at the wheel. Second, a custodial sentence would only be imposed if rashness was made out.

65     The first proposition attempts to draw a line between rashness and negligence. We have already stated that advertence to the potential risks that might arise from one’s conduct will in general be the key ingredient for a finding of rashness. Ng Jui Chuan added the gloss (at [7]) that such advertence must extend to an appreciation that the risk would “in all likelihood” materialise. With respect, we disagree with this. In our judgment, advertence to a real (as opposed to a merely theoretical or fanciful) risk of adverse consequences arising from one’s conduct is sufficient for rashness to be made out. Indeed, it seems to us that if a person is aware that an adverse outcome would in all likelihood occur as a result of his conduct, such conduct might even be viewed as being intentional in nature. The second proposition in Ng Jui Chuan has already been dealt with in the preceding discussion: we have held that the rashness and negligence limbs of s 304A have different ranges of punishment, and rashness does not need to be made out before a custodial sentence can be meted out.

66     We therefore consider that Ng Jui Chuan was wrong on both propositions and should not be followed. We now turn to consider some of the factors that ought to be taken into account by a sentencing court when determining the appropriate sentence for a s 304A(b) traffic death case such as the present.

Sentencing considerations in s 304A(b) traffic death cases

The amount of harm caused

67     In his decision, the DJ cited [34] of AFR ([9] supra) for the proposition that the thin skull rule did not apply in criminal law, and thus, “a person cannot be imputed to intend all [the] consequences, no matter how remote, of an act done by him on another” (see [9] of the GD).

68     The question of whether a sentencing court can take into account the full extent of the harm caused by a particular criminal act can be a difficult philosophical issue. It is a cardinal principle of criminal law that the punishment must be proportionate to the crime (see, eg, Mohamed Shouffee bin Adam v Public Prosecutor [2014] 2 SLR 998 at [47] and Muhammad Saiful bin Ismail v Public Prosecutor [2014] 2 SLR 1028 at [21]). Two irreconcilable fundamental principles underlie the notion of proportionality. The first is the principle that no man should be held accountable for that which is beyond his control (“the control principle”); the second is the brute principle that moral and legal assessments often depend on factors that are beyond the actor’s control (“the outcome materiality principle”).

69     The control principle is but a restatement of the intuitive moral sense that people should not be morally assessed for what is not their fault. Common is the refrain that one cannot be blamed for being late for work because of an unforeseen traffic jam. Specific illustrations of the control principle are legion. Chapter IV of the Penal Code lists certain general exceptions which act as complete defences – for instance, unsound mind (see s 84) and, in certain circumstances, duress (see s 94). The common thread between these two general exceptions is a lack of control on the part of the offender. Where an offender acts in a particular way because of the unsoundness of his mind, the law takes cognisance of the fact that he cannot help but be of unsound mind; similarly, when he acts under duress, the law recognises that he is not acting of his own free will. We absolve such offenders of criminal responsibility, either wholly or partially, because they were not in control of their actions at the material time.

70     This must be juxtaposed against the intuitive moral sense that outcomes do matter. There are many examples of outcomes featuring significantly in criminal law. For instance, the line between attempted murder and murder is a fine one, and details like whether the victim was wearing a bullet-proof vest at the material time or whether a bird flew into the path of the bullet can result in dramatically different outcomes.

71     In our judgment, there are three reasons why the outcome materiality principle should trump the control principle in the context of criminal negligence. First, we note that those provisions of the Penal Code which criminalise negligent conduct are predicated on outcome materiality in two readily-observable aspects. First, despite the requisite standard of care for civil negligence being identical to that for criminal negligence (see Lim Poh Eng at [19]–[28]), civil negligence is not co-extensive with criminal negligence: the Penal Code only criminalises conduct which imperils bodily safety. Second, the prescribed maximum punishment under the Penal Code frequently increases as the gravity of the resultant harm increases, as reflected in the following table setting out the maximum punishments for the negligence limbs of ss 336, 337, 338 and 304:

Section	Elements	Maximum punishment
336(b)	Endangering personal safety by a negligent act	3 months’ imprisonment $1,500 fine
337(b)	Causing hurt by a negligent act	6 months’ imprisonment $2,500 fine
338(b)	Causing grievous hurt by a negligent act	2 years’ imprisonment $5,000 fine
304(b)	Causing death by a negligent act	2 years’ imprisonment Fine without any stipulated maximum amount

72     It is well settled that the power to prescribe the type and range of permissible punishments (within which the Judiciary exercises a sentencing discretion) is something that is within the purview of Parliament (see Mohammad Faizal bin Sabtu v Public Prosecutor [2012] 4 SLR 94 at [43]–[45]). The different tiers of defined negligence-based offences in the Penal Code clearly evince an intention on Parliament’s part to lean in favour of the outcome materiality principle for sentencing purposes in relation to these offences.

73     The second, and perhaps more fundamental, reason why we are of the view that the outcome materiality principle should trump the control principle where criminal negligence is concerned is that there is no exact correspondence between legal and moral assessment. The law does take into account considerations that go beyond moral assessment. It is well settled that the four principles of deterrence, retribution, prevention and rehabilitation underlie sentencing (see, eg, Public Prosecutor v Law Aik Meng [2007] 2 SLR(R) 814 at [17]). In particular, general deterrence, prevention and rehabilitation do not quite equate with a moral assessment of the offender. General deterrence has less to do with the moral condemnation of individual offenders, and more to do with advancing the public interest of reducing crime by deterring the general public from similarly offending. Prevention is concerned with incapacitating offenders who pose a danger to society at large. Rehabilitation, where it is a dominant consideration, is aimed at turning offenders away from a life of crime by altering their values.

74     The third and last reason why we are of the view that the outcome materiality principle should prevail over the control principle in the context of criminal negligence is that a countervailing species of legal luck can operate in favour of a putative offender. Take, for instance, two drivers who briefly fall asleep while driving straight at the same speed along the same stretch of road. One driver wakes up before any harm is caused. The other driver collides into and kills a jaywalking pedestrian. It could be said that as a matter of moral assessment, both drivers are equally culpable. However, as a matter of practical fact, the former will not suffer any legal repercussions because no detectable harm has occurred. Putative offenders take the benefit of legal luck operating in their favour if adverse consequences do not eventuate; it is only fair that an offender should not be heard to raise the control principle as a shield when a harmful outcome does eventuate.

75     It follows that the thin skull rule cannot be ignored in the context of criminal negligence. The outcome materiality principle trumps the control principle, at least in the context of cases involving criminal negligence. How moral luck should be resolved in other contexts is something that we leave to be explored if and when that issue arises.

76     This is not to say, however, that the extent of harm caused will be fully determinative of the sentence meted out. Instead, it is but one factor that is to be considered in determining the appropriate sentence.

Special classes of vulnerable victims?

77     In the GD, the DJ cited Public Prosecutor v Wong Siow Kam [2005] SGDC 125 and Public Prosecutor v Singarayar Cleetruse Geethan Magistrate’s Arrest Case No 11239 of 2012 (unreported) (at [6(ii)]) for the proposition that offenders in s 304A(b) traffic death cases should be punished more harshly if they collided into a particularly vulnerable class of road users, such as pedestrians. In the submissions before us, there was some dispute over whether foreign workers sitting unsecured in the rear cabin of a lorry would constitute a vulnerable class. In our view, this issue was a red herring.

78     It is undeniably true that pedestrians, amongst other classes of road users, are vulnerable. But, offenders in s 304A(b) traffic death cases should not, as a rule, be punished more harshly simply because they have collided into a vulnerable class of road users. There are two objections to such a rule.

79     First, singling out vulnerable classes could result in double-counting where the Prosecution has framed the charge(s) by reference to the harm actually caused. Vulnerable classes are, by definition, those who are more susceptible to injury. A collision with a member of a vulnerable class is inherently more likely to result in hurt, grievous hurt or death. In this regard, the Penal Code, as a matter of law, provides for harsher punishment if greater harm is caused (see, eg, the table set out at [71] above). An offender who collides into a member of a vulnerable class would thus already have a higher chance of falling into a harsher band of punishment, and there is no reason to doubly punish such an offender by enhancing his sentence on the grounds that his victim is a vulnerable victim.

80     Second, singling out classes as vulnerable might well result in wasteful litigation over whether a particular class ought to be recognised as vulnerable. The law is ill-equipped to draw invidious lines between road users. The facts of this case are a pertinent example: the DJ declined to hold that workers sitting unsecured in the rear cabin of lorries were a vulnerable class of road users. Taking this line of logic to its conclusion, it would seem that any driver of a four-wheeled vehicle ought to be punished more severely for colliding into a road user not enclosed within another four-wheeled vehicle, as compared to colliding into a road user enclosed within such a vehicle. It seems untenable to make such assessments when all lives should be held to be equally precious in the eyes of the law.

81     This is not to say that an offender who collides into a member of a class hitherto recognised as “vulnerable” will never be punished more severely. Much will depend on the precise facts of the case. For instance, if a driver spots a pedestrian from afar and sounds the horn to warn the pedestrian not to cross but does not slow down in anticipation of the pedestrian not heeding the warning, and if he tragically knocks down the pedestrian in such circumstances, he can and should be punished more severely. However, this is because of the particular facts of the case and not because of a rule that singles out pedestrians as a vulnerable class.

Speeding

82     Travelling above the speed limit is an aggravating factor in s 304A(b) traffic death cases. It is almost axiomatic to say that speeding increases the risk of harm occurring. One reason for this is that more force is required to stop an object that is travelling at a higher speed: given that brakes can only apply a certain maximum amount of frictional force, this results in a longer braking distance for a car travelling at a higher speed as compared to one travelling at a lower speed.

83     Another reason for the increase in risk inherent in speeding is perception-reaction time. As noted in Marc Green, “‘How Long Does It Take to Stop?’ Methodological Analysis of Driver Perception-Brake Times” (2000) 2(3) Transportation Human Factors 195 (at p 213), “the most ecologically valid driver response time for surprise intrusions is about 1.5 sec”. When a vehicle is travelling at a higher speed, it will travel a greater distance in those critical 1.5 seconds.

Impaired judgment: drink-driving

84     Drink-driving is another relevant aggravating factor in s 304A(b) traffic death cases. Alcohol is a drug that has depressant effects. As noted by Henri Begleiter and Arthur Platz in their chapter on “The Effects of Alcohol on the Central Nervous System in Humans” in The Biology of Alcoholism, Volume 2: Physiology and Behavior (Benjamin Kissin & Henri Begleiter eds) (Plenum Publishing Corporation, 1972) ch 10 (at p 338):

Acute intoxication produced by increasing concentrations of alcohol in the blood produces impairment of psychological functions such as perception, discrimination, association and voluntary response. …

85     Section 67 of the RTA criminalises driving while under the influence of alcohol or drugs. It is, quite rightly, a crime for a person to drink alcohol (or consume drugs) beyond the permitted limit and then drive, even if no harmful consequences ensue. There is no requirement to prove actual harm because the risk of harm inherent in such conduct is so great.

86     It is not surprising then that the courts have treated driving while under the influence of intoxicants as a significant aggravating factor in s 304A(b) traffic death cases. For instance, in Wong Yew Foo ([57] supra), Chan Seng Onn J commented that drink-driving was an act “of complete selfish disregard for the safety of other fellow road users” (at [32]) and sentenced the offender to a total of four months’ imprisonment.

Impaired judgment: sleepy driving

87     In our judgment, just as drink-driving is a significant aggravating factor in s 304A(b) traffic death cases, so too is sleepy driving.

88     Sleep deprivation has discernible effects on a person’s cognitive abilities. There is considerable literature on the subject, and for illustration purposes, we refer to a small selection below:

(a)     M Thomas et al, “Neural Basis of Alertness and Cognitive Performance during Sleepiness. I. Effects of 24 h of Sleep Deprivation on Waking Human Regional Brain Activity” (2000) 9 J Sleep Res 335, which concluded that alertness and cognitive performance declined in association with certain physiological brain deactivations;

(b)     J A Horne & L A Reyner, “Sleep related vehicle accidents” (1995) 310 BMJ 565, which reported that: (i) sleep-related vehicular accidents comprised 16% of all accidents on major roads in southwest England; and (ii) such accidents peaked at certain times of the day, with the most vulnerable times being around 2.00am to 7.00am and 3.00pm to 6.00pm; and

(c)     S Blazejewski et al, “Factors Associated With Serious Traffic Crashes: A Prospective Study in Southwest France” (2012) 172 Arch Intern Med 1039, which suggested that, at least for vehicular accidents, alcohol consumption and sleepiness were risk factors of almost equal magnitude.

89     Notwithstanding the adverse and discernible effects which sleep deprivation has on a person’s cognitive abilities, there is, at the same time, a good and simple reason why sleepy driving in and of itself does not constitute an offence – there is no simple physiological test which can be administered to detect how sleepy a person is. As technology currently stands, numeric biochemical indicators cannot stand as a proxy for a person’s own level or sense of sleepiness: unlike a numerical limit that can be applied in cases of alcohol consumption, there is no easily testable objective bright line that a legislature can draw to curtail sleepy driving. A legislative fiat that it is illegal to drive without having had eight hours of sleep within the preceding 24 hours would be impractical, likely overbroad and easily evaded; it would also bring the law into disrepute.

90     In contrast, a breathalyser or blood test is able to reveal the amount of alcohol in a person’s body. This makes it easy for the law to draw a bright line in relation to alcohol consumption, beyond which it is illegal for a person to drive (see s 72 of the RTA). Even so, numeric biochemical indicators are not perfect, simply because people react to alcohol in different ways: an amount of alcohol which is sufficient to cause one person to be drunk may barely have an effect on another. Notwithstanding that, the law can and does, in the interests of safety, prescribe a common standard of “permissible” alcohol consumption, beyond which it is illegal for a person to drive regardless of the particular effect of that level of alcohol consumption on him.

91     In the case of sleep deprivation, for the reasons set out above, the law cannot set out a meaningful common standard. But, where an accident has occurred and investigations reveal that the offender went through a prolonged period of time without sleep prior to the accident, this is likely to be an aggravating factor that calls for enhanced punishment.

92     As to what would constitute sufficient sleep, we recognise that this might well be highly dependent on a person’s subjective characteristics. We are unable and unwilling to draw any bright lines in this regard. It suffices for us to reiterate that weight will be placed on the fact that an offender in a s 304A(b) traffic death case was sleep-deprived at the material time if that was indeed the factual situation.

Our decision on the appropriate sentence to impose on the respondent

93     Before coming to the grounds for our sentencing decision in this appeal, two aspects of this case merit a brief mention. First, the respondent pleaded guilty and admitted to the Statement of Facts, which did not state that she was speeding. We studied the video footage provided by a member of the public, and calculated her speed at the time of the accident on the basis of the distance traversed between lamp posts. It appeared from this that the respondent was travelling at an average speed some way above the speed limit prior to the collision with the lorry. Nonetheless, we took no heed of this because both the Prosecution and the Defence conducted their respective cases in the court below as well as in this appeal on the basis that the respondent was not speeding at the material time.

94     Second, the respondent pleaded guilty to a charge under s 304A(b) of the Penal Code, the negligence limb of s 304(A). The respondent’s advertence or non-advertence to the risks she was running, while not relevant to a finding of liability under s 304A(b) (see [45] and [48] above), is nevertheless still relevant as an aggravating factor for sentencing purposes (see [38]–[39] above).

95     In our view, the DJ erred in not placing any, or in not placing sufficient weight on the following six aggravating factors:

(a)     The respondent had gone for more than 24 hours without proper sleep prior to the accident. She ended her 12-hour shift on 14 March 2013 at 7.00pm, which meant that she started work at 7.00am that day and must have been awake for some time before that. As mentioned earlier (see [3] above), the accident occurred at about 7.20am on 15 March 2013.

(b)     The respondent worked in the surveillance department of Marina Bay Sands Casino. Her job entailed, in her counsel’s words, “mentally grueling [sic]”^{[note: 8]} 12-hour shifts. The intense concentration required would have drained her mentally, and this was a factor she must have appreciated.

(c)     The respondent was sufficiently alive to the risk of being overcome by fatigue that she thought to get some rest between the end of her shift on 14 March 2013 and the time she met her friends.

(d)     The respondent admitted in her cautioned statements that she was still in the midst of getting used to her new car, which she had bought shortly before the accident.

(e)     The respondent drove on an expressway during the build-up to the morning rush hour. She must have known that the expressway was likely to be increasingly crowded with relatively fast-moving traffic, which would in turn call for a heightened sense of alertness on her part.

(f)     The collision caused one death and injuries to ten others, seven of whom suffered grievous hurt and one of whom is now paralysed from the waist down.

96     In our view, the DJ also erred in taking into account one consideration which he ought not to have taken into account. The DJ referred to the respondent, owing to the nature of her work, being able to meet her friends only at night and in the early hours of the morning (at [10(iii)] of the GD). With respect, we were unable to see how this could possibly be regarded as a mitigating factor.

97     We thus allowed the appeal. The factors listed at [95(a)]–[95(e)] above increased the risk that the respondent would end up being overcome by fatigue and, as a result, drive in a state of unconsciousness with disastrous consequences. That is precisely what ended up happening. We were amply satisfied that the threshold for imposing a custodial sentence had been crossed. We further observe in passing that the fact that the respondent thought it necessary to have a brief rest after she ended her shift on 14 March 2013 before venturing out to meet her friends suggested advertence to the aforesaid risk. Moreover, all the factors listed at [95] above, save for the last, were matters within her knowledge.

98     That said, notwithstanding our observations above, we felt obliged to have regard to the past precedents. We shall elaborate on this in the next section. It was because of these precedents that we imposed a term of imprisonment of only four weeks. If we had considered the question of the appropriate sentence to impose in this case without regard to the past precedents, a term of imprisonment extending to months, rather than weeks, would, in our judgment, have been warranted.

A coda on prospective overruling: when is prospective overruling justified?

The declaratory theory of law

99     Common law systems are by and large inductive in nature. Cases are the atomistic building blocks; inductively-derived principles of law are validated when they are “recognised to govern particular factual matrices and are actually applied” in subsequent cases involving similar questions (see, eg, See Toh Siew Kee v Ho Ah Lam Ferrocement (Pte) Ltd and others [2013] 3 SLR 284 at [35]–[36]).

100    Ordinarily, when a common law court pronounces on the law, the pronouncement is unbound by time and operates both retrospectively and prospectively. The manufacturer of the ginger beer in M’Alister (or Donoghue) (Pauper) v Stevenson [1932] AC 562 (more commonly cited as “Donoghue v Stevenson”) was liable under Lord Atkin’s neighbour principle, even though the principle was first articulated in the judgment of the House of Lords well after the alleged negligence had taken place.

101    The retroactivity of the common law was initially normatively justified by the declaratory theory of judicial decisions (“the declaratory theory of law”). Under this conception of the law, the common law was immanent and unchanging. Blackstone stated that “decifions [sic] of courts of juftice [sic] are the evidence of what is common law” (see Sir W Blackstone, Commentaries on the Law of England (Clarendon Press, 1765–1769) at p 71). In like manner, Hale wrote that judicial decisions “do not make a law, properly so called … yet they have a great weight and authority in expounding, declaring and publishing what the law of this kingdom is” (see Sir M Hale, The History of the Common Law of England (H Butterworth, 6th Ed, 1820) at p 90). If judicial decisions are merely evidence of a Platonic ideal, it naturally follows that any declaration of what the law is, is also a declaration of what the law was and what the law will be going forward into the future. The inevitable corollary of the declaratory theory of law is that there can never be a change in the law. There are, no doubt, changes in the “evidence” or one’s “understanding” of the law, but these changes necessarily mean that the prior “evidence” or “understanding” was misconceived or wrong.

102    The first cracks in the declaratory theory of law appeared as early as 1880 in In re Hallett’s Estate; Knatchbull v Hallett (1880) 13 Ch D 696, where Jessel MR held that the rules of equity, unlike the rules of the common law, were not established from time immemorial, but were instead altered, improved, refined and invented from time to time (at 710). It is, of course, questionable whether equitable adjudication is so radically different from common law adjudication that the declaratory theory of law applies to the latter but not the former; after all, equity is also atomistically built upon case law.

103    To modern eyes, the declaratory theory of law is nothing more than an elaborate fiction. Lord Reid opined, albeit extra-curially, that the theory was a fairy tale that society no longer believed in (see Lord Reid, “The Judge as Law Maker” (1972–1973) 12 J Soc’y Pub Tchrs L 22 at p 22). The death knell for this theory finally came in the House of Lords decision of Kleinwort Benson Ltd v Lincoln City Council [1999] 2 AC 349 (“Kleinwort Benson”). The appellant bank in that case had entered into various interest rate swap agreements with four respondent local authorities (interest rate swap agreements are derivatives which allow two parties to exchange interest rate cash flows and hedge against or profit from the differential between fixed and floating interest rates). An earlier House of Lords decision, Hazell v Hammersmith and Fulham London Borough Council [1992] 2 AC 1, had declared that such interest rate swap agreements with local authorities were ultra vires and void. The appellant bank in Kleinwort Benson sought to recover monies paid over to the respondent local authorities on the basis of unjust enrichment and, more specifically, a mistake of law. The House of Lords (by a majority) allowed the appellant’s appeal, and in so doing, four of the law lords rejected the declaratory theory of law. The common thread between these four law lords was the view that the common law, being a system of judge-made law, did change from time to time in order to keep abreast with the times.

104    In the Singapore context, Review Publishing Co Ltd and another v Lee Hsien Loong and another appeal [2010] 1 SLR 52 sounded the death knell for the declaratory theory of law. The issue in that case was whether the defendant in a defamation suit could avail herself of the Reynolds privilege (named after the eponymous Reynolds v Times Newspapers Ltd [2001] 2 AC 127). One of the arguments advanced on the defendant’s behalf was that the Reynolds privilege had always been part of Singapore law. Chan Sek Keong CJ rejected this argument, holding (among other reasons) that the declaratory theory of law was no longer part of the prevailing orthodoxy (at [241]) and could not explain evolutionary changes in the common law in response to changing policy considerations (at [243]).

Retroactive and prospective overruling compared

105    The rejection of the declaratory theory of law in both its normative and descriptive guises does not, and should not, entail a wholesale rejection of its myriad consequences, some of which are deeply entrenched in our system of justice. As has already been mentioned, one important consequence is that judicial decisions are unbound by time.

106    Perhaps, the most important reason for this is that the entire mechanism of justice is premised on litigants, embroiled in real disputes, resorting to courts of law. Parties are incentivised to engage in the system of justice and put their best cases forward because they stand to benefit if they manage to persuade the courts to rule in their favour. This incentive will by and large be absent if the default position is that judicial decisions are only prospective in nature. In such a setting, persons similarly situated to the winning litigant will benefit from any change in the law after the decision concerning the winning litigant is handed down, but the winning litigant himself will garner no benefit from his win. This is not tenable.

107    Abandoning retroactivity of judicial decisions would also arbitrarily draw a line between similarly-situated litigants. If a judicial decision changes the law, there must be good reason for that change, and it is difficult to justify not applying the change to a class of persons simply because they fall on the wrong side of an arbitrary date. In this regard, we agree with Lord Goff of Chieveley’s retort in Kleinwort Benson that he “cannot imagine how a common law system, or indeed any legal system, can operate otherwise [than by giving judicial decisions retroactive effect] if the law is to be applied equally to all and yet be capable of organic change” (at 379).

108    That said, there are at the same time compelling arguments in favour of prospective overruling. The most compelling pertains to the rule of law. Friedrich von Hayek succinctly summarised the main thrust of the rule of law thus (see F A Hayek, The Road to Serfdom (Routledge & Sons, 1944) at p 54):

… [S]tripped of all technicalities [the rule of law] means that government in all its actions is bound by rules fixed and announced beforehand-rules which make it possible to foresee with fair certainty how the authority will use its coercive powers in given circumstances and to plan one’s individual affairs on the basis of this knowledge.

109    One of Joseph Raz’s eight principles on the rule of law was the principle that all laws should be prospective, open and clear in order to be able to guide conduct. In Raz’s view, retrospective laws conflicted with such a conception of the rule of law and were not able to give such guidance (see J Raz, “The Rule of Law and its Virtue” (1977) 93 LQR 195 at pp 198–99). In a similar vein, Lon Fuller, in expounding on the inner morality of the law, contended that a retrospective law was a monstrosity: “[t]o speak of governing or directing conduct today by rules that will be enacted tomorrow is to talk in blank prose” (see L Fuller, The Morality of Law (Yale University Press, 1964) at p 53). The premise underlying the position taken by jurists such as Raz and Fuller is that because people conduct their affairs on the basis of what they understand the law to be, a retrospective change in the law can frustrate legitimate expectations.

110    The final complication is that special considerations must come into play in the criminal context, especially where a person’s physical liberty is at stake. Article 11(1) of the Constitution of the Republic of Singapore (1985 Rev Ed, 1999 Reprint) (“the Singapore Constitution”) prohibits punishment on the basis of a retroactive criminal law. Article 2 defines “law” to include the common law. In Public Prosecutor v Manogaran s/o R Ramu [1996] 3 SLR(R) 390 (“Manogaran”), the Court of Appeal confirmed that Art 11(1) applied not just to Acts passed by the Legislature, but also to judicial pronouncements (at [66]). This is but an embodiment of the Latin maxim “nullum crimen nulla poena sine lege” (“the nullum crimen sine lege maxim”), which means, as set out at [61] of Manogaran, that “conduct cannot be punished as criminal unless some rule of law has already declared conduct of that kind to be criminal and punishable as such”. Some trace this maxim to the Magna Carta and the writings of Locke and Blackstone (see, eg, S Glaser, “Nullum Crimen Sine Lege” (1942) 24 J Comp Leg 29 at p 29). On the other hand, if a system of pure prospective overruling were to be adopted, an appellant-accused who successfully appeals against his conviction might find himself languishing in prison despite winning his appeal.

111    Amidst this patchwork of competing considerations, different systems of law have come to different positions on prospective overruling. We recount below (at [112]–[119]) five common law systems that have, at some point in time, considered some form of prospective overruling.

The position in other common law jurisdictions

England

112    In England, the House of Lords in In re Spectrum Plus Ltd (in liquidation) [2005] 2 AC 680 (“Spectrum Plus”) examined Kleinwort Benson, among other decisions, and unanimously came to the conclusion that there could be exceptional cases which called for prospective overruling. Spectrum Plus involved a debenture which “by way of specific charge” created a charge over a company’s book debts in favour of a bank, and obligated the company not to sell, factor, discount or otherwise charge or assign any book debt in favour of any other person without the consent of the bank. According to the first-instance decision of Siebe Gorman & Co Ltd v Barclays Bank Ltd [1979] 2 Lloyd’s Rep 142 (“Siebe Gorman”), this created a fixed charge over the company’s book debts. The House of Lords overruled Siebe Gorman and held that in fact, a floating charge had been created because the company could freely deal with book debts collected and paid into its current account. A seven-judge bench of the House of Lords unanimously held that it was open to the House of Lords to declare that a decision would only have prospective effect. To Lord Scott of Foscote, prospective overruling would be justified where a decision would have gravely unfair and disruptive consequences for past transactions or happenings (at [40]). On the facts of Spectrum Plus, this requirement was not met because, amongst other reasons, a first-instance decision could not be said to have definitively settled the law and lulled sophisticated operators into a false sense of security.

113    In the criminal sphere, England is bound by Art 7(1) of the European Convention of Human Rights, which prohibits, among other things, punishment under a retroactive law. SW v United Kingdom (1996) 21 EHRR 363 (“SW v UK”), a case before the European Court of Human Rights (“the European Court”), involved an applicant-accused who had unsuccessfully invoked the defence of marital immunity at his trial for the rape of his wife. As English law stood then, the general proposition, albeit subject to a number of exceptions, was that a man could not be guilty of raping his wife. The European Court found that Art 7(1) had not been violated because by the time of the accused’s conviction, considerable doubt had been cast on the doctrine of marital immunity for rape (“the doctrine of marital immunity”), and there were strong indications that the English courts would increase the number and the width of the exceptions to that doctrine.

The US

114    In the US, the high-water mark for prospective overruling was the US Supreme Court’s decision in Chevron Oil Company v Gaines Ted Huson (1971) 404 US 97 (“Chevron v Huson”). That case held that three factors were to be considered for both criminal and civil cases in determining whether prospective overruling was justified, namely: (a) the decision to be applied non-retroactively must establish a new principle of law by overruling a past decision or deciding an issue for the first time; (b) the history, purpose and effect of the rule in question must be analysed to determine if retroactive operation would further or retard its operation; and (c) the inequity imposed by retroactive application must be weighed (at 106–107).

115    The subsequent US Supreme Court decision of Randall Lamont Griffith v Kentucky (1987) 479 US 314 overruled Chevron v Huson, and concluded that prospective overruling was inapposite in criminal cases for two reasons, viz: (a) the integrity of judicial review (the US Supreme Court equivalent of appellate review in the Commonwealth context) required that a declared rule be applied to all similar pending cases (at 322–323); and (b) selective application of new rules violated the principle of treating similarly-situated offenders in the same way (at 323). James B Beam Distilling Company v Georgia (1991) 501 US 529, also a US Supreme Court decision, cited much the same reasons in holding that prospective overruling was also inapposite in the civil context.

India

116    In I C Golaknath & Ors v State of Punjab & Anrs [1967] 2 SCR 762, an 11-judge bench of the Supreme Court of India laid down three propositions in relation to prospective overruling as follows: (a) it could only be invoked in matters arising under the Indian Constitution; (b) it could only be applied by the Indian Supreme Court; and (c) “the scope of the retrospective operation of the law declared by the [Indian] [S]upreme Court superseding its earlier decisions is left to its discretion to be moulded in accordance with … the justice of the cause or matter before it” (at 766).

New Zealand

117    It is uncertain if prospective overruling is part of New Zealand law. The decision of the Supreme Court of New Zealand in Lai v Chamberlains [2007] 2 NZLR 7, which followed the lead of Arthur J S Hall & Co (a Firm) v Simons [2002] 1 AC 615 and abolished barrister immunity, is equivocal. A majority of three judges considered that it was not necessary to consider whether the courts had the power to make prospective changes to the law because there was no evidence that the removal of barrister immunity would upset expectations to such an extent that such removal should only apply prospectively (at [95]). A minority of two judges held that the court did have the power (at [147] and [205]).

Canada

118    Canada rejects prospective overruling as a matter of principle, but seems to have in fact applied it on one occasion. In Edward v Edward Estate (1987) 39 DLR (4th) 654, the Saskatchewan Court of Appeal rejected prospective overruling as being a dramatic deviation from the norm in both Canada and England (at [31]) that would affect the court’s independent, neutral and non-legislative role (at [30]).

119    The Saskatchewan Court of Appeal appeared not to have taken cognisance of the earlier decision of the Supreme Court of Canada in Reference re Language Rights Under s. 23 of Manitoba Act, 1870 and s. 133 of Constitution Act, 1867 (1985) 19 DLR (4th) 1. In that case, two constitutional statutory instruments made it mandatory for statutes to be enacted, printed and published in both English and French. The Canadian Supreme Court ruled that statutes enacted, printed and published only in English were invalid, but restricted the retrospective effect of the ruling. The impugned statutes were deemed temporarily valid for the minimum period necessary for their translation, re-enactment, printing and publication in both English and French. The Canadian Supreme Court held that it would recognise unconstitutional enactments as valid where a failure to do so would lead to legal chaos and thus violate the rule of law (at [109]). It should be noted, however, that the Canadian Supreme Court did not invoke the language of prospective overruling.

The prevailing position in Singapore

120    The Singapore Court of Appeal has recognised the doctrine of prospective overruling, and applied it on two occasions.

121    The first occasion was in Manogaran ([110] supra). The respondent in that case was charged with trafficking “cannabis mixture” because the Prosecution’s expert had testified that the substance in question could not be certified as cannabis. The earlier Court of Appeal decision of Abdul Raman bin Yusof v Public Prosecutor [1996] 2 SLR(R) 538 (“Abdul Raman”) had defined “cannabis mixture” exhaustively to mean a mixture of the cannabis plant and another species of plant. No other species of plant was, however, present in the “cannabis mixture” in Manogaran. The Court of Appeal overruled its earlier decision in Abdul Raman prospectively, and held that “cannabis mixture” could also constitute a mixture of different parts of the cannabis plant. This extension of the ambit of criminal liability, if not done prospectively, would have violated Art 11(1) of the Singapore Constitution and the nullum crimen sine lege maxim (at [75]); it would also have violated the legitimate expectations of persons who had expected their actions to be legal (at [81]).

122    The second occasion on which prospective overruling was applied was in Abdul Nasir bin Amer Hamsah v Public Prosecutor [1997] 2 SLR(R) 842 (“Abdul Nasir”). In that case, the appellant had been sentenced to life imprisonment and 12 strokes of the cane for kidnapping. The prevailing practice then was to treat a sentence of life imprisonment as equivalent to one of 20 years’ imprisonment. In a judgment delivered by Yong CJ, the Court of Appeal held that life imprisonment should be understood as imprisonment for the whole of the remaining period of the convicted person’s natural life. However, this holding was only to take prospective effect from the date of the decision. Article 11(1) of the Singapore Constitution and the nullum crimen sine lege maxim were again invoked: those who had conducted their affairs by relying on a reasonable and legitimate interpretation of the law should not be penalised if a later judicial pronouncement established that that interpretation was wrong (at [51]).

The framework for prospective overruling

123    Singapore case law has thus far analysed prospective overruling within the rubric of the nullum crimen sine lege maxim. We consider that the arguments in favour of prospective overruling are not peculiar to any one milieu of law and cannot be restricted solely to criminal law.

124    As has already been established, legal systems at either end of the spectrum (that is, those which adopt purely prospective overruling and those which adopt purely retroactive overruling) are apt to produce their own brands of injustice. The tension between retroactivity and prospectivity, in our judgment, is best resolved by a framework in which judicial pronouncements are, by default, fully retroactive in nature. Our appellate courts (that is, our High Court sitting in its appellate capacity and our Court of Appeal) nevertheless have the discretion, in exceptional circumstances, to restrict the retroactive effect of their pronouncements. This discretion is to be guided by the following factors:

(a)      The extent to which the law or legal principle concerned is entrenched: The more entrenched a law or legal principle is, the greater the need for any overruling of that law or legal principle to be prospective. This will be measured by, amongst other things, the position of the courts in the hierarchy that have adopted the law or legal principle that is to be overruled and the number of cases which have followed it. A pronouncement by our Court of Appeal which exhaustively analyses several disparate positions before coming to a single position on a point of law will be more entrenched than a passing pronouncement on that same point of law by a first-instance court. Similarly, a law or legal principle cited in a long line of cases is more entrenched than one cited in a smaller number of cases.

(b)      The extent of the change to the law: The greater the change to the law, the greater the need for prospective overruling. A wholesale revolutionary abandonment of a legal position (as was done in, for instance, Manogaran ([110] supra)) is a greater change than an evolutionary reframing of the law (see, for instance, Sembcorp Marine Ltd v PPL Holdings Pte Ltd and another and another appeal [2013] 4 SLR 193, which re-examined the distinction between interpretation and implication in contract law, but by and large built on the foundations laid down by prior cases).

(c)      The extent to which the change to the law is foreseeable: The less foreseeable the change to the law, the greater the need for prospective overruling. In SW v UK ([113] supra), for example, the abolition of the doctrine of marital immunity was eminently foreseeable because of past judicial pronouncements which had expressed distaste for the doctrine and progressively expanded the exceptions to it. There was therefore no need to curtail the retroactive application of the change in the legal position.

(d)      The extent of reliance on the law or legal principle concerned: The greater the reliance on the law or legal principle being overruled, the greater the need for prospective overruling. This factor is particularly compelling in the criminal law context, where a person’s physical liberty is potentially at stake. Quite apart from Art 11(1) of the Singapore Constitution, a person who conducts his affairs in reliance on the ostensible legality of his actions would be unfairly taken by surprise if a retrospective change to the law were to expose him to criminal liability.

125    We stress that this framework lays down a factors-based test; as such, no one factor is preponderant over any other, and no one factor is necessary before prospective overruling can be adopted in a particular case. Indeed, a first-time judicial pronouncement, despite generally not fulfilling the first three factors listed at [124] above, could conceivably warrant prospective overruling. We refer to the exceptional facts of Abdul Nasir ([122] supra) as an analogous example. Prior to that decision, there was no Singapore case which had pronounced on the meaning of life imprisonment, but the Executive had consistently taken it to mean 20 years’ imprisonment. Offenders had pleaded guilty or conducted their defences on the basis that life imprisonment was understood to mean imprisonment for 20 years, and it would have been grossly unfair if the rug had been pulled from under their feet, especially as this concerned their physical liberty.

Application of the framework to the present facts

126    Prior to the 2008 Penal Code amendments, the Gan Lim Soon position on sentencing for negligent driving which constituted an offence under the old s 304A was well entrenched in the law. It was a pronouncement by the then sitting Chief Justice in the High Court (which is ordinarily the apex appellate court for magistrate’s appeals), and it definitively settled the default punitive position for that scenario. A long string of cases followed the lead of Gan Lim Soon in only imposing a fine. Even after the 2008 Penal Code amendments, reliance continued to be placed on Gan Lim Soon in at least one High Court decision (namely, Sylvester Lee ([57] supra)). The lone reservation was expressed in the High Court decision of Wong Yew Foo ([57] supra), which doubted that the starting point for sentencing in a s 304A(b) traffic death case remained a fine, but declined to give guidance for future cases. By this metric, Ng Jui Chuan was not well-entrenched in the law because it was the only High Court decision to opine on the punitive position for s 304A(b) offences arising from sleepy driving.

127    We have moved away from the Gan Lim Soon position, and have also concluded that the court in Ng Jui Chuan erred in concluding that driving while seriously sleep-deprived would not warrant the imposition of a custodial sentence. The shift from a default sentence of a fine to a default sentence of a term of imprisonment is a significant change in the law.

128    The change to the law in relation to Gan Lim Soon was not eminently foreseeable. As we have just noted, the only case which questioned the continuing validity of the sentencing guidelines stated there was Wong Yew Foo. This stands in sharp contrast with the situation obtaining in SW v UK, where English law had progressively expanded the exceptions to the doctrine of marital immunity and had inexorably marched in the direction of its abolition.

129    It must be taken that offenders in s 304A(b) traffic death cases have placed reliance on Gan Lim Soon. While we do not have empirical evidence in this regard, numerous offenders would have pleaded guilty to or conducted their defences on the basis of advice that the starting point for sentencing in such cases would likely be only a fine. Indeed, we note that in the court below, Mr Khaira, counsel for the respondent, admitted to the respondent having momentarily “blanked out due to her tired mental state”^{[note: 9]} and placed considerable reliance on Gan Lim Soon and Ng Jui Chuan to press for a sentence of only a fine.

130    Having regard to the above-mentioned factors, it would, in our judgment, have been unfair to the respondent if we had approached sentencing in this case without regard to Gan Lim Soon. However, we did not face any such restraint in relation to departing from Ng Jui Chuan.

131    It was clear to us that a fine would have been inapposite on the facts of this case, which demonstrated an egregious level of negligence. Even on the basis of Gan Lim Soon, a period of imprisonment would be appropriate. In the premises, the four-week sentence of imprisonment which we ordered was a compromise: we gave retroactive effect to the departure from Ng Jui Chuan (which, as we have noted, did not lay down any well-established principle), but gave only prospective effect to our decision to depart from Gan Lim Soon.

Conclusion

132    In the premises, we allowed the appeal and ordered the respondent to serve a four-week term of imprisonment. We did not disturb the five-year disqualification period ordered by the DJ, save to direct that it should take effect only after the respondent has served her term of imprisonment. We also ordered the fine of $10,000 imposed by the DJ, which the respondent has already paid, to be returned to her.

133    To reiterate, the default starting position in terms of the punishment to be meted out in a s 304A(b) traffic death case is a short custodial sentence of up to four weeks’ imprisonment (see [61] above). This is liable to be adjusted up or down by reference to the extent of negligence involved as well as the presence of aggravating and/or mitigating factors.

134    In particular, the presence of any of the aggravating factors analysed in these grounds of decision (viz, speeding, drink-driving and sleepy driving) would call for a starting point of between two and four months’ imprisonment. The amount of harm caused would also have to be taken into account for the purposes of sentencing. As mentioned earlier (see [98] above), were prospective overruling not warranted in this case, we would have sentenced the respondent – who had driven while being sleep-deprived; whose actions resulted in one person being killed and several others being seriously injured; and in respect of whom there were several other aggravating factors as noted above – to a much longer term of imprisonment.

135    We would like to take this opportunity to signal to drivers the consequences of the tremendous risks that they take on, not only to themselves but also to other innocent road users, when they drive despite not being in a fit condition to do so.

136    We close by expressing our appreciation to Mr Zhuo, the amicus curiae, for his assistance in putting forward a number of significant cases and arguments for our consideration.

---

^{[note: 1]}Para 22(b) of the respondent’s plea in mitigation.

^{[note: 2]}Para 22 of the appellant’s submissions.

^{[note: 3]}Para 35 of the appellant’s submissions.

^{[note: 4]}Para 23 of the amicus curiae’s submissions.

^{[note: 5]}Para 38 of the amicus curiae’s submissions.

^{[note: 6]}Para 39 of the amicus curiae’s submissions.

^{[note: 7]}Para 42 of the amicus curiae’s submissions.

^{[note: 8]}Para 12 of the respondent’s plea in mitigation.

^{[note: 9]}Para 22(b) of the respondent’s plea in mitigation.