1 The appeals before me, HC/RA 185/2023 (“RA 185”) and HC/RA 186/2023 (“RA 186”), were against the decision of the learned Assistant Registrar (“the AR”) dismissing HC/SUM 235/2023 (“SUM 235”) and HC/SUM 1427/2023 (“SUM 1427”) (collectively, “the Stay Applications”), whereby the defendants applied to stay this action (“the Suit”). The first defendant (“Terraform”) applied on the basis that there was an arbitration agreement between it and the claimants, while the other defendants applied on case management grounds.

2 I dismissed the appeals, issuing brief grounds on 28 September 2023. On 1 November 2023, I dismissed Terraform’s further arguments and affirmed my earlier decision.

3 The appeals engage interesting questions, in the context of a representative action, involving the incorporation of an arbitration agreement in contracts formed online; the application of the relevant legal test for the purposes of granting a stay in favour of arbitration; and the circumstances under which a litigant may be said to have taken a “step in the proceedings” for the purposes of s 6(1) of the International Arbitration Act 1994 (2020 Rev Ed) (“IAA”) in the context of the Rules of Court 2021 (“ROC 2021”). These are my full grounds of decision.

4 The claimants are individuals who purchased algorithmic stable cryptocurrency tokens named TerraUSD (“UST”) issued by Terraform. Terraform is a Singapore-incorporated company in the business of developing software and applications which run the Terra blockchain. ¹ Terraform also operates the Terra ecosystem (“the Terra Ecosystem”), which provides a platform for Terraform’s development and sale of decentralised financial products and services. ² The central feature of the Terra Ecosystem is UST. ³ The Terra Ecosystem includes a number of other projects, platforms and applications that build atop the Terra blockchain. One of these is the Anchor Protocol, a lending and borrowing platform where users can stake their UST in consideration for promised returns calculated on an annualised yield basis. ⁴

5  The second defendant, Kwon Do Hyeong (“Kwon”), is the co-founder, shareholder, director and Chief Executive Officer of Terraform. ⁵ The third defendant, Nikolaos Alexandros Platias (“Platias”), has been described as a co‑founder and Head of Research of Terraform, and one of the co-founders or creators of the Anchor Protocol. ⁶ The fourth defendant, Luna Foundation Guard Ltd (“Luna”), is said to be an organisation supporting the growth of the Terra Ecosystem by building reserves to buttress the stability of UST. ⁷

6 The Suit is a representative action filed on 7 September 2022 by Julian Moreno Beltran (“Beltran”) and Douglas Gan Yi Dong (“Gan”) (collectively, “the Representative Claimants”), on behalf of themselves and 375 other individuals. ⁸

7 The claimants seek relief against the defendants for, inter alia, making several misrepresentations that they claimed induced them to purchase UST, stake UST on the Anchor Protocol, and hold UST while its value plummeted. As a result, the claimants suffered significant losses. ⁹ The relevant representations (“the Representations”) were: ¹⁰

(a) first, that UST was stable by design, as it was pegged to a fiat currency;

(b) second, that the Terra protocol and its underlying token economics would be able to maintain the price stability of UST regardless of market size, volatility, or demand through an algorithm that would allow an arbitrage process to happen with UST’s sister token, LUNA, thereby guaranteeing that UST would always return to the US$1 peg;

(c) third, that UST holders would be able to protect the value of their UST holdings, given that UST holders would always be able to exchange 1 UST for US$1 worth of LUNA on the Terra protocol;

(d) fourth, that the Anchor Protocol was a principal-guaranteed stablecoin savings product where UST holders could enjoy the stability of holding a stablecoin while earning passive income;

(e) fifth, that the purchasers of UST will earn up to 20% Annualised Percentage Yields if they staked their UST on the Anchor Protocol; and

(f) sixth, that Luna would be the final backstop in protecting the UST-USD price peg as it had built up a substantial reserve of funds, denominated in Bitcoin.

8 I shall refer to the first to third Representations collectively as the “Terra Representations”. The claimants allege that the first to second Representations were made in a white paper published in April 2019 (“the Terra White Paper”) by, inter alia, Kwon and Platias. ¹¹ The Terra Representations (including the third Representation) allegedly also appeared on Terraform’s website (“the Terra Website”). ¹²

9 On the claimants’ pleaded case, they were induced by the Terra Representations to purchase UST, ¹³ and this constituted a contract between the claimants and Terraform on the terms of the Terra Representations. ¹⁴ Alternatively, the claimants plead that the Terra Representations amounted to a unilateral contractual offer to the world at large, which they accepted when they purchased UST. ¹⁵

10 As for the fourth and fifth Representations (“the Anchor Representations”), the claimants refer to the website of the Anchor Protocol (“the Anchor Website”) and another white paper published in July 2020 (“the Anchor White Paper”) by, inter alia, Platias. ¹⁶ The claimants also reference a social media post made by Kwon, in which he promoted the Anchor Protocol’s yields on UST. ¹⁷

11 In respect of the sixth Representation (“the Luna Representation”), the claimants refer to announcements published on various websites (but emanating from the defendants), promoting Luna and its role and actions in building up reserves to ensure the stability of UST. ¹⁸

12 On the claimants’ pleaded case: eight of the claimants plead that they were only induced by the Terra Representations; 195 claimants were induced by the Terra Representations and the Anchor Representations; and 174 claimants (including Beltran and Gan) were induced by all the Representations. ¹⁹ It is pleaded that the claimants in the second and third groups (ie, other than the eight claimants in the first group) entered into a second contractual agreement with Terraform on the terms of the Anchor Representations. ²⁰

13 In May 2022, the value of UST plummeted. ²¹ The claimants allege that in reliance on the Representations, they held onto their UST. ²² The claimants claim that they suffered loss and damage in the sum of US$65,646,750.29, being the sum equivalent to the diminution in value between the US$1 peg for UST and the value of UST held by the claimants and/or the price at which the claimants sold their UST below the US$1 peg. ²³

14 For the purposes of the present appeals, it was undisputed that both the Terra and Anchor Websites (collectively, “the Websites”) contained terms requiring disputes to be resolved by arbitration.

15 The Terra Website contained terms and conditions accessible via a hyperlink entitled “Terms of Use” (“the Terra Terms of Use”). Clause 13 of the Terra Terms of Use provided that any dispute involving the Terra Website or the Terra Terms of Use shall be resolved by arbitration: ²⁴

16 Similarly, the Anchor Website contained terms and conditions accessible via a hyperlink entitled “Terms of Service” (“the Anchor Terms of Service”). Clause 18 of the Anchor Terms of Service provided that any dispute shall be resolved by arbitration: ²⁵

17 I shall refer to these clauses collectively as “the Arbitration Clauses”.

18 It was common ground that the Arbitration Clauses were wide enough to encompass the claims pleaded in the Suit. What was in issue was whether they applied to the claimants. I consider this in greater detail below at [120]–[161].

19 On 9 November 2022, Terraform filed a Pre–Case Conference Questionnaire which stated, inter alia, that it intended to challenge the jurisdiction of the court on the basis of arbitration agreements said to have been entered into between Terraform and the claimants, based on the Arbitration Clauses. ²⁶

20 On 24 November 2022, Terraform filed its Defence. Significantly, the Defence was not limited to Terraform’s jurisdictional challenge, but included its defence on the merits of the claims as well as a Counterclaim for various declarations. The Defence however, contained a reservation that it was filed “without prejudice” to Terraform’s contention that the court had no jurisdiction to hear the case, and its filing was “not to be construed as a submission to the jurisdiction of the Court”. ²⁷ A similar reservation was pleaded in respect of the Counterclaim. I shall refer to these reservations collectively as “the Reservations”. I set out the reservation in the Defence in full: ²⁸

21 Kwon and Luna (who were represented by the same solicitors as Terraform) filed similar defences, addressing both the jurisdictional challenge and the merits, with similar reservations. ²⁹ On 23 December 2022, Platias filed his defence, addressing only the jurisdictional challenge, without prejudice to his right to file a defence on the merits should the jurisdictional challenge fail. ³⁰

22 Parties engaged in a series of correspondence, spanning several case conferences, in which the timelines for (a) the claimants to give further and better particulars (“F&BPs”) of their Statement of Claim (“SOC”) (which had previously been requested by the defendants); and (b) the defendants to file and serve their jurisdictional challenges, were negotiated, and extended.

23 First, at a case conference on 9 December 2022, in view of an agreed extension of the time within which the claimants were to respond with their F&BPs, and a corresponding agreed extension of the timeline for the defendants to file and serve their jurisdictional challenges, the learned Senior Assistant Registrar (“the SAR”) gave directions for the defendants to file and serve their jurisdictional challenges by 19 January 2023. ³¹

24 On 27 December 2022, following a further agreed extension of time for the claimants’ F&BPs, the court issued directions extending the deadline for the filing and serving of jurisdictional challenges to 30 January 2023. ³²

25 Only Platias complied with the directions, filing SUM 235 on 30 January 2023, which asked for a case management stay on the basis of, inter alia, the existence of a valid arbitration agreement between the claimants and Terraform. ³³ On the same day, the solicitors representing Terraform, Kwon and Luna wrote to the SAR seeking a further extension of time to file their jurisdictional challenges by 27 February 2023, citing the complexity and length of the claimants’ F&BPs (which had been issued on 12 January 2023) as the reason for their request. ³⁴ Following a number of letters from parties to the court on 30 January 2023 and the days after, the court granted, on 7 February 2023, a further extension of time for the filing of the defendants’ jurisdictional challenges, pending directions from the court at the next case conference scheduled for 8 February 2023 (“the 8 Feb Case Conference”). ³⁵

26 On 6 February 2023, instead of filing their application to challenge the court’s jurisdiction, Terraform, Kwon and Luna filed a “Request for Permission to File Application” to the Registrar of the Supreme Court (“the Request Application”), which sought the court’s leave to file an “omnibus application” seeking a number of reliefs, including F&BPs, the production of the claimants’ representative action agreements (“RAAs”), and the striking out and/or stay of the Suit on various grounds. ³⁶

27 Later the same day, Terraform, Kwon and Luna filed a summons seeking, inter alia, F&BPs, the production of the RAAs, and the stay or striking out of the Suit on various grounds (“the SAPT Summons”). ³⁷ The SAPT Summons is the “omnibus” application that Terraform sought to make in its Request Application. I deal with the Request Application and SAPT Summons in detail below at [79]–[116].

28 At the 8 Feb Case Conference, the learned SAR indicated that the Registry would reject the Request Application and SAPT Summons, and granted a further extension of time, until the date of the next case conference or as given in further directions of the court (whichever later), for Terraform, Kwon and Luna to file their jurisdictional challenges. ³⁸

29 On 1 March 2023, Terraform, Kwon and Luna filed SUM 1427 for a stay of the Suit. On 11 May 2023, SUM 1427 was amended to clarify that Terraform was seeking a stay in favour of arbitration whereas Kwon and Luna were seeking “case management” stays on the basis that the claims in the Suit against Kwon and Luna were closely related and/or ancillary to the claims against Terraform.

30 The AR dismissed Terraform’s application for a stay on the ground that it failed to make out a prima facie case that a valid arbitration agreement existed between it and the claimants. ³⁹

31 The AR held that the question was whether the Arbitration Clauses had been validly incorporated into the contracts formed between Terraform, on the one hand, and the claimants who visited the Websites on the other, so as to form a valid arbitration agreement between them. ⁴⁰ This would be the case if a user of the Websites would have actual or constructive notice of the Arbitration Clauses, which in turn depended on whether the Websites provided such user with a legitimate opportunity to learn that his or her use was subject to the Terms of Use/Service.

32 In that regard, the AR found that:

(a) the relevant hyperlink for the Terra Terms of Use was “tucked away at the bottom of the website such that it lacked prominence”, and a reasonably prudent user would not have had notice thereof, owing to the “relative obscurity” of that hyperlink; ⁴¹

(b) similarly, the Anchor Terms of Service could not be found on the Anchor Website’s homepage, and could only be found on the “dashboard” of the Anchor Website (“the Dashboard”), which was a separate tab within the Website; ⁴²

(c) further, it did not matter that users of the Websites may arguably have been sophisticated users looking for information on the Terra Ecosystem, as opposed to simple e-commerce customers looking to buy everyday items. In contrast to the argument that users of the Websites should have expected there to be terms and conditions, a countervailing argument could be made that given the nature of the product promoted by the Websites, the Terms of Use/Service should have been prominently displayed to users; ⁴³ and

(d) ultimately, arbitration clauses must be expressly brought to the attention of the other contracting party. ⁴⁴ In the circumstances, the facts did not show that a reasonably prudent user would have actual or constructive notice of the Terms of Use/Service. ⁴⁵ Hence, the Arbitration Clauses were not properly and adequately incorporated and there was no prima facie valid arbitration agreement between the claimants and Terraform. ⁴⁶

33 In the alternative, the learned AR held that even if a valid arbitration agreement could be shown to exist prima facie, Terraform had taken multiple steps in the proceedings and so had submitted to the jurisdiction of the court. ⁴⁷

34 Finally, she dismissed the “case management stays” of the other defendants as they were premised on Terraform obtaining a stay in favour of arbitration. ⁴⁸

35 The issues on appeal were:

36 Under s 6(1) of the IAA, any party to an arbitration agreement “may, at any time after filing and serving a notice of intention to contest or not contest and before delivering any pleading (other than a pleading asserting that the court does not have jurisdiction in the proceedings) or taking any other step in the proceedings, apply to that court to stay the proceedings so far as the proceedings relate to that matter” [emphasis added].

37 What constitutes a “step in the proceedings” has been considered in several authorities. The starting and basic point is that an act which indicates an intention that the court proceedings should proceed instead of arbitration is a “step in the proceedings”: Carona Holdings Pte Ltd and others v Go Go Delicacy Pte Ltd [2008] 4 SLR(R) 460 (“Carona Holdings”) at [47] and [50], citing Austin and Whiteley Limited v S Bowley and Son (1913) 108 LT 921 at 921 and Eagle Star Insurance Co Ltd v Yuval Insurance Co Ltd [1978] 1 Lloyd’s Rep 357 at 361 respectively.

38 Such intention may be shown where the acts of the defendant, or his solicitors, is objectively inconsistent with the making and maintaining of a challenge to the court’s jurisdiction: Carona Holdings at [60], citing Global Multimedia International Ltd v Ara Media Services [2007] 1 All ER (Comm) 1160 at [27]. In other words, an act will be taken to be a “step in the proceedings” where it cannot be explained, except on the assumption that the defendant accepts that the court should be given jurisdiction, or that any objection to the court’s jurisdiction has been waived or has never been entertained: Reputation Administration Service Pte Ltd v Spamhaus Technology Ltd [2021] 2 SLR 342 (“Reputation Administration”) at [20], citing Shanghai Turbo Enterprises Ltd v Liu Ming [2019] 1 SLR 779 (“Shanghai Turbo”) at [44] and Zoom Communications Ltd v Broadcast Solutions Pte Ltd [2014] 4 SLR 500 (“Zoom Communications”) at [43].

39 Thus, an act which goes toward the advancement of a jurisdictional challenge will not constitute a “step in the proceedings”. For example, a request for F&BPs or production of documents for the purpose of deciding whether to bring a jurisdictional challenge will not be a “step”: Carona Holdings at [91], citing Fathers of Confederation Buildings Trust v Pigott Construction Co Ltd (1974) 44 DLR (3d) 265; Reputation Administration at [31]. However, in so far as the requested particulars or documents go beyond that purpose, and indicate that the defendant intends to defend the court proceedings, such request would be considered a “step in the proceedings”: see Carona Holdings at [87], citing The Dufferin Paving Co Ltd v The George A Fuller Co of Canada Ltd [1935] OR 21 (“Dufferin”) at 24.

40 Certain acts may appear to engage the court’s powers, but are nevertheless not inconsistent with a jurisdictional challenge. For example, an act which merely parries a blow by the plaintiff does not constitute a “step in the proceedings” (see Carona Holdings at [49], citing Roussel-Uclaf v G D Searle & Co Ltd and G D Searle & Co [1978] 1 Lloyd’s Rep 225 at 231) – this consideration is particularly important in the context of a defendant resisting an injunction. Such an act is not inconsistent with a jurisdictional challenge as it is meant to safeguard the defendant’s position pending the determination of the jurisdictional challenge.

41 Similarly, an application relating to the propriety of the proceedings in limine, or at the threshold, may appear to engage the court’s powers but is not in fact inconsistent with a jurisdictional challenge. For example, in Maniach Pte Ltd v L Capital Jones Ltd and another [2016] 3 SLR 801 (“Maniach”), the court held that a striking-out application – which was filed on the basis that the claimant had failed to obtain leave under the Companies Act (Cap 50, 2006 Rev Ed) before commencing proceedings against a company which was the subject of a judicial management application – was not a “step in the proceedings” (Maniach at [81]). This decision may be explained on the basis that the issue of leave had to be resolved before the question of a stay could be addressed. Without leave to commence the proceedings, the company could not rightfully be before the court.

42 I note that in Maniach, the court found (at [81]) that “the Company’s prayer to strike out these proceedings on grounds of a procedural defect cannot amount to a step … [m]aking that application did not put in play the merits of [the] claim” [emphasis added]. The court held further (at [93]) that the striking‑out application “did not cross the line which separates a procedural act which is not a step in the proceedings from one which is” [emphasis added].

43 I do not understand these passages as suggesting that the key distinction for the purpose of identifying a “step” is whether an act is procedural rather than one which addresses the merits of the claim. In some situations, it may be a useful guide – an act which engages the merits of the action surely constitutes a “step in the proceedings”. However, it is not always the case that a procedural act will not count as a “step in the proceedings”. Some “procedural” acts indicate that the defendant intends to engage and advance the court proceedings – for example, an application for security for costs (Carona Holdings at [73] and [83]); requiring disclosure of documents (Carona Holdings at [55], citing Parker, Gaines & Co, Limited v Turpin [1918] 1 KB 358); or attending a summons for directions (Carona Holdings at [55], citing The County Theatres and Hotels, Limited v Knowles [1902] 1 KB 480 and Richardson v Le Maitre [1903] 2 Ch 222).

44 Thus, a truer compass may be found in the statement in Carona Holdings (at [55]) that a “step in the proceedings” will be deemed to have been taken if the defendant “employs court procedures to enable him to defeat or defend [the] proceedings on their merits” [emphasis in original retained; emphasis added]. The focus is not on whether the act itself goes toward the merits of the action, but whether it enables, or advances, a future engagement of the merits of the action.

45 Consistent with the above principles, an equivocal act will not be considered a “step in the proceedings”: Carona Holdings at [60]. This refers to an act which may be consistent with either acceptance of the court’s jurisdiction or the maintenance of a jurisdictional challenge. Having said that, although a party may assert that its actions were intended to support its jurisdictional challenge, the court must look beyond such claims and examine the substance of the action. Parties should not be allowed to equivocate or hedge their actions – they should be decisive in whether they are insisting on arbitration in preference to litigation, and disingenuous reservations will be disregarded: Carona Holdings at [93].

46 Overall, the court’s assessment of whether an act constitutes a “step in the proceedings” should be conducted in a practical and commonsensical way which considers the circumstances surrounding the defendant’s act: Carona Holdings at [52].

47 All the authorities cited above were decided before the enactment of the ROC 2021. The ROC 2021 introduces a new regime on the filing of defences on jurisdiction. For clarity, the applicable legal principles relating to a “step in the proceedings” remain the same, but the application of these principles to the new statutory context under the ROC 2021 may produce different results.

48 Prior to the ROC 2021, a defendant seeking to challenge the jurisdiction of the court had no option to file a defence objecting to jurisdiction alone (as opposed to a defence on the merits). The effect was to place a defendant in a difficult position: a failure to file a defence within the stipulated deadline exposed him to the risk of judgment in default being entered: see Australian Timber Products Pte Ltd v Koh Brothers Building & Civil Engineering Contractor (Pte) Ltd [2005] 1 SLR(R) 168 (“Australian Timber”) at [16]. The onus was therefore on the defendant to be proactive by applying for a stay quickly and seeking an extension of time to file the defence, an urgent hearing of the stay application (Australian Timber at [16]), or a subsidiary stay of the proceedings (including the running of the timeline to file and serve the defence) until the main stay application could be disposed of (Carona Holdings at [38]). Such applications would invariably be allowed and were therefore a waste of costs and the court’s resources.

49 Further delays and costs would be incurred when a defendant failed to file a defence or a stay application on a timely basis, and then sought to oppose the entering of judgment in default or applied to set aside the judgment in default.

50 The ROC 2021 addresses these issues by allowing a defendant to file a defence contesting jurisdiction only (“a Defence (Jurisdiction)”): see O 2 r 5(2) and O 6 r 7(4). With that filing, no judgment in default of a defence can be entered: O 6 r 7(7). Further, filing a Defence (Jurisdiction) is expressly provided to not amount to a submission to the court’s jurisdiction: see O 2 r 5(4) and O 6 r 7(6). Section 6(1) of the IAA was also amended to state that delivering “a pleading asserting that the court does not have jurisdiction in the proceedings” will not preclude parties to an arbitration agreement from applying to stay the court proceedings.

51 It is therefore clear that a defendant seeking to challenge the jurisdiction of the court on the basis that the dispute should be arbitrated only needs to file a Defence (Jurisdiction): see also Civil Justice Commission, Civil Justice Commission Report (29 December 2017) (Chairperson: Justice Tay Yong Kwang) (“Civil Justice Commission Report”) at p 14, which states that for a defendant “challenging the jurisdiction of the Court … [O 6 r 7] requires the defendant to file and serve a bare defence, stating the ground of challenge on jurisdiction” [emphasis added]. Building on this, the ROC 2021 was designed to ensure that jurisdictional challenges are dealt with expeditiously, with a view to prevent delays, ensure the efficient use of court resources and save costs. This is in line with the Ideals set out in O 3 of the ROC 2021. The procedure for dealing with a jurisdictional challenge is as follows:

(a) first, a defendant who is served must file and serve a defence to the originating claim after the SOC is served on it: see O 6 r 7(1). The defence must be in Form 13: see O 6 r 7(3). Form 13 sets out a menu of options for the defendant depending on the nature of the challenge it intends to bring;

(b) second, the court is required to fix a case conference after the filing of an originating claim or an originating application: see O 2 r 6(1);

(c) third, at the case conference, if a Defence (Jurisdiction) has been filed, the court must first deal with the objection to its jurisdiction: see O 9 r 7(1). In that regard, short timelines are imposed with respect to the challenge: the court must direct the defendant to file and serve the necessary application for the jurisdictional challenge with the supporting affidavit within 14 days after the date of the case conference, direct the claimant to file and serve any affidavit in reply within 14 days thereafter, and (unless the court otherwise orders) fix the hearing of the application on a date no later than 14 days after all affidavits have been filed and served: see O 9 r 7(2) and O 2 r 6(7). The words in italics emphasise the intention for jurisdictional challenges to be dealt with and disposed of quickly; and

(d) where the jurisdictional challenge fails, the court will give directions to file a defence contesting the merits of the claim (“a Defence (Merits)”): see O 2 r 7(1). As such, the defendant will have the opportunity to deal with the merits of the claimants’ pleaded case and bring any counterclaim it intends to institute, after its jurisdictional objection has been dealt with.

52 Referring to step (a) of this procedure, the plain wording of O 6 r 7(4) draws a clear distinction between a Defence (Jurisdiction) and a Defence (Merits). Order 6 r 7(4) states:

53 This is reinforced by O 6 r 7(6), which states that “[a] defence filed under [O 6 r 7(4)] is not treated as a submission to jurisdiction”. This can only be a reference to a Defence (Jurisdiction).

54 Terraform argued that the phrase “need not” in O 6 r 7(4) is permissive, in that a defendant can choose to file a Defence (Merits) together with a Defence (Jurisdiction). ⁴⁹ I do not accept that argument. The phrase “need not” must be understood in the context of the whole of O 6 r 7. Order 6 rr 7(1) and 7(2) provide that a defendant “must” file and serve a defence after the statement of claim is served on it. In the circumstances, the phrase “need not” in O 6 r 7(4) simply clarifies that a defendant challenging jurisdiction only needs to file a Defence (Jurisdiction) and not a Defence (Merits) – it does not give an option to do both.

55 Terraform’s argument also ignores the issues with the predecessor to the ROC 2021 highlighted at [48]–[49] above and undermines what O 6 r 7(4) was plainly intended to do. It should also be interpreted and understood in the context of O 3, which I elaborate on at [63] below.

56 Order 6 r 7(3) provides that the defence must be in Form 13. Terraform argued that Form 13 required the defendants to plead a “paragraph-by-paragraph response to the [SOC]” and that it was only complying with what Form 13 mandates. ⁵⁰ I do not accept that argument. Form 13 is only a basic template, intended to serve as a guide on how a defence and counterclaim should be structured and what it should or should not contain. This is evident by the use of square brackets (“[ ]”) in Form 13, which requires a defendant to make modifications as necessary – it is simply a “menu”. Form 13 must be understood in light of O 6 r 7 as explained above.

57 Terraform argued that O 3 r 2(4) of the ROC 2021 puts a party at risk of having its pleadings rejected by the court if they are not prepared in accordance with the applicable rules or practice directions. ⁵¹ But that is not a legitimate concern as O 3 r 6(1) expressly provides that the “[f]orms as set out in the practice directions must be used with such variations as the circumstances require” [emphasis added]. More importantly, a Defence (Jurisdiction) complies with O 6 r 7 and Form 13 and will not be rejected.

58 Indeed, at the hearing before the AR, counsel for Terraform conceded that Terraform “may have been wrong” in filing a paragraph-by-paragraph response pursuant to Form 13. ⁵² In so far as Terraform had made an error in its interpretation of ROC 2021 and Form 13, that does not assist it. The “transitional learning phase” for the ROC 2021, during which the courts would be generally more sympathetic when dealing with non-compliance occasioned by lack of familiarity with the ROC 2021, ended on 30 June 2022: see Newspaper Seng Logistics Pte Ltd v Chiap Seng Productions Pte Ltd [2023] SGHC(A) 5 at [11]. Further, as elaborated above at [51], not only is it clear that a Defence (Jurisdiction) need not be filed with a Defence (Merits), it is also expressly provided that a Defence (Merits) may be filed after the jurisdictional challenge is dealt with: see O 2 r 7(1). There is therefore no ambiguity nor jeopardy for defendants. Platias correctly filed only a Defence (Jurisdiction) (see above at [21]) – there is no reason Terraform could not have similarly done so. In any event, whether Terraform or its solicitors may have misunderstood ROC 2021 is irrelevant as the court must assess whether a party had taken a “step in the proceedings” on an objective basis: Carona Holdings at [60].

59 The authorities decided pre-ROC 2021 make clear that a reservation is a relevant factor which should be taken into consideration in assessing whether a step taken by a party can be objectively construed as a “step in the proceedings”. Thus, it was held in Australian Timber (at [22], citing Chong Long Hak Kee Construction Trading Co v IEC Global Pte Ltd [2003] 4 SLR(R) 499 (“Chong Long”) at [13]) that “an act, which would otherwise be regarded as a step in the proceedings, will not be treated as such if the applicant has specifically stated that he intends to seek a stay or expressly reserves his right to do so”.

60 However, even in the pre-ROC 2021 context, the making of a reservation was not by itself determinative of the question of whether an act constituted a step in the proceedings. The court still had to consider whether such reservation was disingenuous (Carona Holdings at [93]), and whether notwithstanding the reservation, a party’s conduct clearly and unequivocally signified a submission to jurisdiction (Shanghai Turbo at [38]). The court in Shanghai Turbo (at [36]) was careful to note that the remarks in Australian Timber “did not lay down any blanket rule that no conduct would ever amount to a submission to jurisdiction if it was accompanied by a reservation of that party’s right to challenge jurisdiction” [emphasis in original].

61 Importantly, Australian Timber and Chong Long were decided in the pre-ROC 2021 context. As noted above at [48], there was no option pre‑ROC 2021 to file a Defence (Jurisdiction). In that context, the filing of a Defence (Merits) within the stipulated deadline, coupled with an express reservation, could be fairly construed as a neutral step performed to protect the defendant’s position by preventing the entering of judgment in default. Such an act preserved the status quo, similar to an application for an extension of time to file a defence to avoid judgment in default being entered (Carona Holdings at [100]–[101]), and could be construed as a pre-emptive act of self‑defence, done in the interest of “parrying a blow from the plaintiff”: see International SOS Pte Ltd v Overton Mark Harold George [2001] 2 SLR(R) 777 at [6] and Shanghai Turbo at [41]).

62 However, under the ROC 2021, there is little, if any, reason for a defendant challenging jurisdiction to file a Defence (Merits) with a reservation. Indeed, as seen above at [51], the ROC 2021 expressly mandates that the jurisdictional challenge be (quickly) disposed of first before the merits of the action are dealt with, if necessary.

63 There are also good reasons why defendants should not file a Defence (Merits) coupled with a reservation. First, this practice increases costs. Terraform pointed out that under the ROC 2021, the claimants do not have to file a Reply, and they did not do so here. ⁵³ That misses the point. The defendant will incur time and costs in preparing and filing the Defence (Merits) and the claimant and other parties will incur time and costs in having to review and consider that defence. Further, where a counterclaim is filed (as in this case), the claimant is compelled to file a defence: see O 6 rr 9(1) and 9(3). All these would be wasted if the jurisdictional challenge ultimately succeeds. This would be contrary to the Ideals in O 3 r 1 of the ROC 2021, especially those of achieving expeditious proceedings (O 3 r 1(2)(b)) and cost-effective work (O 3 r 1(2)(c)). Second, parties may use the Defence (Merits) as a backdoor to introduce facts relating to the merits of the action in their application for a stay or to muddy the waters on what is relevant to that application. This should not be allowed. Third, allowing the use of reservations encourages “gaming” – parties may take steps relevant to, or to enquire into, the merits of the claim and later rely on those reservations to claim that their conduct was not “unequivocal”. As will be seen below, that is precisely what Terraform did.

64 Thus, in relation to pleadings under the ROC 2021, reservations are no longer relevant or necessary. A party should simply file a Defence (Jurisdiction) without more. As for other contexts, eg, in affidavits, a reservation may be useful where an act is equivocal – ie, statements contained therein may be equivocal, and the defendant may include a reservation to make his position clear. However, it bears emphasising that a reservation is not a panacea – it will not save an act that is inconsistent with the bringing of a jurisdictional challenge from being a “step in the proceedings”.

65 Overall, under the ROC 2021 regime, defendants should focus their energies on the jurisdictional challenge and only take out applications that are necessary to advance it. Any other action inconsistent with the maintenance or advancement of a jurisdictional challenge should be considered a “step in the proceedings”. To hold otherwise would be to encourage the proliferation of other applications which contributed to the situation pre-ROC 2021, and which the ROC 2021 was introduced to solve: see Civil Justice Commission Report at p 14.

66 The claimants submitted that the following acts by Terraform amounted to steps in the proceedings:

(a) first, Terraform filed a defence on the merits and a counterclaim; ⁵⁴ and

(b) second, Terraform filed the SAPT Summons seeking substantive remedies relating to the merits of the action. These included (i) the request for F&BPs; (ii) the application for production of the RAAs; and (iii) the striking-out application. ⁵⁵

67 Terraform’s defence contained not only its jurisdictional challenge but also its defence on the merits of the Suit and a counterclaim for various declarations, coupled with the Reservations (see [20] above). ⁵⁶

68 As discussed above, Terraform argued that it was obliged under O 6 r 7(3) of the ROC 2021 to file its defence as a paragraph-by-paragraph response to the claimants’ SOC, in accordance with Form 13, ⁵⁷ and that it was also obliged under O 6 r 7(7) to file a defence or risk entry of judgment in default of defence. ⁵⁸

69 In respect of the counterclaim, Terraform similarly argued that it “submitted this along with its Defence (Jurisdiction) because it believed by reason of Order 6 Rule 8 of the ROC 2021 that it was obliged to submit any counterclaim it may have along with its Defence in Form 13” [emphasis added]. ⁵⁹

70 Terraform argued further that the defence and counterclaim was not a “step in the proceedings” as it was prefaced with the Reservations. ⁶⁰ Terraform pointed out that in light of the Reservations, it could not be said that it intended to ask the court to determine the dispute and to abandon its right to arbitrate. ⁶¹

71 For the reasons stated above at [48]–[58], Terraform’s filing of a Defence (Merits) was a “step in the proceedings”. No such defence was required under the ROC 2021 regime, and this act affirmed the correctness of the court proceedings and demonstrated Terraform’s willingness to accede to the court’s jurisdiction on the matter: see Australian Timbers at [19]. The Reservations included in Terraform’s defence do not change this conclusion, as there is no longer any reason to file a Defence (Merits) coupled with a reservation: see above at [59]–[64]. Any alleged misunderstanding of the ROC 2021 regime on Terraform’s part is irrelevant: see above at [58].

72 In its further arguments, Terraform argued that a statement that a defence is filed “without prejudice to the application … to stay all proceedings in Singapore” effectively preserved a party’s right to seek a stay. ⁶² It relied on Capital Trust Investments Ltd v Radio Design TJ AB [2002] 2 All ER 159, where the English Court of Appeal held (at [59]–[60]) that an application for summary judgment was not a step because the applicant had made it clear that the application was advanced only if its stay application was unsuccessful. But this decision only undermines Terraform’s position as it was a case where the defendant had filed an application dealing with the merits only as a “fallback”. As discussed below at [108]–[115], that was not the case here. In so far as Terraform is arguing that pleading a reservation is a licence to make all manner of applications without prejudicing its jurisdictional challenge, that is plainly misconceived.

73 In further arguments, Terraform argued that it was not certain of its position on jurisdiction since there was a lack of information in the SOC on the claimants and the alleged contracts formed with Terraform. ⁶³ Hence, when it filed its defence on 24 November 2022, it did so without the benefit of information needed to confirm its position on jurisdiction (since the claimants’ F&BPs were only served on 13 January 2023). ⁶⁴ At this stage, Terraform had yet to determine its ability to bring a jurisdictional challenge, and considered that it might be unable to apply for a stay even after receipt of the F&BPs. If that eventuality materialised, and Terraform’s defence did not address the merits, it would have filed no proper defence to the claimants’ claims and would be at risk of judgment in default and/or face difficulty defending the claims in court. ⁶⁵ Hence, Terraform’s filing of the defence and counterclaim was a “neutral procedural step” done merely to preserve the status quo and to protect its interests in court pending the receipt of the F&BPs. ⁶⁶

74 These arguments are misconceived. First, it is for a defendant to decide if it has a proper basis to challenge jurisdiction. It would be a rare case for the defendant to be unable to do so based on the pleaded cause of action. Second, even if a defendant requires better particulars or information from the claimant to enable it to make that decision, there is no reason why it cannot file a Defence (Jurisdiction) and seek leave or directions from the court to make applications strictly confined to that issue: see Reputation Administration at [31], where the court noted that a defendant faced with uncertainty as to its jurisdictional challenge should take steps to resolve such uncertainty, rather than filing a Defence (Merits). As will be seen below, that is not what Terraform did. Third, if the defendant later determines that there is no basis to challenge jurisdiction, there is nothing to stop it from informing the court that it is not proceeding with the challenge (or withdrawing it if already filed) and obtaining directions to file a Defence (Merits). The risk of judgment in default as argued by Terraform simply does not exist.

75 Even if I am wrong with respect to the defence, there is clearly no reason for Terraform to have filed a Counterclaim. That is plainly an acceptance of the court’s jurisdiction to decide on the merits of the dispute. In this regard, I reiterate that parties should be decisive in whether they are insisting on arbitration in preference to litigation, and disingenuous reservations will be disregarded: Carona Holdings at [93]. In the circumstances, the Reservations did not detract from Terraform’s contradictory act in filing a Counterclaim.

76 Terraform relied on O 6 r 8(1), which states that “[i]f the defendant intends to counterclaim against the claimant, the defendant must file and serve the counterclaim with the defence”. ⁶⁷ This cannot be read as an obligation on a defendant, seeking to challenge the jurisdiction of the court, to file their intended counterclaim along with their Defence (Jurisdiction). Such a construction would render O 6 r 7(4) and O 2 r 5(2) (which introduce the Defence (Jurisdiction)) redundant. Form 13, which Terraform relies on, also does not assist it as it only requires Terraform to plead the jurisdictional basis on which it is asking the court to try the counterclaim.

77 Terraform’s argument that it filed a Defence (Merits) and counterclaim because it was unsure, as a result of the claimants’ (poorly) pleaded case, if it could put forth a jurisdictional challenge, ⁶⁸ is also not supported by the facts. Pertinently, this argument appears to have only been raised in Terraform’s further arguments. It was not raised in Terraform’s (initial) submissions before me or the AR. Neither did Terraform’s defence and counterclaim include any statement to the effect that the Reservations were on account of Terraform’s uncertainty over its jurisdictional challenge. Rather, Terraform’s argument all along was that the Defence (Merits) was filed because Terraform believed that O 6 r 7(4) of the ROC 2021 allowed it to, and Form 13 obliged it to. ⁶⁹ Further, in an affidavit dated 28 February 2023 – well after Terraform’s Defence and Counterclaim was filed – Mr Arrash Christopher Amani (“Mr Amani”), Terraform’s Head of Operations and Community, emphasised that in the Pre‑Case Conference Questionnaire submitted on or around 9 November 2022, Terraform stated definitively that “[b]oth [Terraform’s] Terms of Use and the Anchor website’s Terms of Service contain inter alia an obligation to refer disputes to arbitration, as well as a prohibition against class or representative proceedings. The [claimants] (and/or any members of their purported groups) are therefore not entitled to bring or continue with representative proceedings in the Singapore court in relation to their claims for alleged misrepresentations found on [the Terra Ecosystem]” [emphasis added]. ⁷⁰ In the same affidavit, Mr Amani noted that Terraform filed its defence on 24 November 2022, in which (a) they challenged the court’s jurisdiction on, inter alia, the fact that the claimants were bound by an arbitration agreement; and (b) a reservation was included. ⁷¹ Tellingly, Mr Amani did not express any uncertainty over the validity of this jurisdictional challenge. Hence, Terraform’s (further) argument that it was unsure of its position appears to be an afterthought. In any event, even if Terraform was unsure, the answer was not to file a Defence (Merits) and a Counterclaim, but to take steps to address its uncertainty (see [74] above).

78 Even if I am wrong in relation to Terraform’s conduct in filing its defence on merits and counterclaim, the subsequent steps it took made plain that it had taken steps in the proceedings.

79 As noted above at [26]–[27], Terraform, Kwon and Luna filed the Request Application and the SAPT Summons, the latter of which sought various reliefs, including (a) F&BPs, (b) production of documents, and (c) a stay or striking out of the Suit. The claimants argued that these reliefs related to the merits of the Suit and showed a willingness to accept the jurisdiction of the court. ⁷²

80 The fact that the Request Application (which enclosed the SAPT Summons) was merely a request does not mean that it was not a “step in the proceedings”. The Request Application was filed in the afternoon of 6 February 2023, and was followed almost immediately by the filing of the SAPT Summons the same day, before the Request Application itself even came up for consideration. On 8 February 2023, the SAR indicated that the Registry would reject both the Request Application and the SAPT Summons, and the court later recorded both as having been rejected. ⁷³

81 The SAPT Summons was not a mere request – it was a proper application. As I discuss below, it involved matters that were clearly unrelated to Terraform’s jurisdictional challenge and would only have been necessary if Terraform accepted that the court had jurisdiction. In the circumstances, the Request Application could not be separated from the SAPT Summons and regarded as a mere request.

82 In any event, it is doubtful that the characterisation of an application as a request can save it from constituting a “step in the proceedings”. There is no principled reason why a request cannot be viewed as an affirmation of the correctness of the court proceedings, or an act which advances the hearing of the matter in court, which would result in it being considered a “step in the proceedings”: see Australian Timber at [19] and Carona Holdings at [93]. The determination will depend on the specific facts and circumstances involving the application.

83 Further, it does not matter that that the Request Application and the SAPT Summons were rejected. As noted in L Capital Jones Ltd and another v Maniach Pte Ltd [2017] 1 SLR 312 (“L Capital Jones”) at [83], even an application that is withdrawn, or one which the party no longer wishes to proceed with, cannot change the fact that a step has been taken in the proceedings. In the present case, the Request Application and the SAPT Summons were rejected by the court; it is therefore not the case that Terraform decided not to pursue the applications contained within the SAPT Summons. Thus, there was an even stronger argument here that the Request Application and the SAPT Summons demonstrated a submission to the court’s jurisdiction.

84 An application to the court for F&BPs may be considered a “step in the proceedings”: see Carona Holdings at [101], referencing S P Chua Pte Ltd v Lee Kim Tah (Pte) Ltd [1993] 1 SLR(R) 793 at [17]. However, an application for F&BPs will not constitute a “step in the proceedings” if it was sought to guide the defendant as to whether he should allow the action to proceed in court, or apply for a stay of proceedings: Carona Holdings at [91]. Applying these principles, a vital question was whether the F&BPs sought by Terraform went beyond what was necessary to ascertain the applicability of the Arbitration Clauses to the claims in the Suit.

85 In the SAPT Summons, Terraform applied for two sets of particulars of the SOC – (a) in “Annex A”, F&BPs in response to some of the F&BP requests made by Terraform, Kwon and Luna on 10 November 2022 (Terraform had complained that the claimants omitted to respond to some of these requests in their F&BPs response served on 13 January 2023); and (b) in “Annex B”, a new set of F&BPs comprising seven instances of particulars of the SOC. ⁷⁴ A number of these requests were clearly not relevant to the issue of jurisdiction.

86 First, a considerable number of the Annex A requests were plainly not relevant to Terraform’s jurisdictional challenge on the grounds of a prima facie arbitration agreement. For example:

(a) para 5 of Annex A requested for the claimants to “state with proper particularity the facts and circumstances by which it is alleged that the contents of the [Terra Website] and the [Terra White Paper] … constituted an intention on the part of [Terraform] to create legal relations with the world”. ⁷⁵ This related to the issue of whether a contract was concluded between Terraform and the claimants such that the claimants could claim for breach of contract. This had nothing to do with whether the Arbitration Clauses applied – Terraform argued that an arbitration agreement “may be in the form of a separate agreement”; ⁷⁶

(b) similarly, para 24 of Annex A requested for the claimants to “state with proper particularity the facts and circumstances by which it is said that [Terraform] had an intention to create legal relations with [the claimants]” [emphasis added]; ⁷⁷

(c) para 10 of Annex A requested for the claimants to “state with proper particularity how Anchor is regarded in fact or in law to be subsidized by [Terraform and Luna]”. ⁷⁸ This appeared relevant to the merits of the claimants’ claims under the tort of deceit, and in particular, the pleading that the Anchor Representations were false because “[i]n respect of the 5th Representation: Anchor was artificially subsidized by [Terraform and Luna], and in fact had no actual way to keep up the 20% APY [ie, Annualised Percentage Yields] in the long run”; ⁷⁹

(d) para 12 of Annex A requested for the claimants to state “[t]he precise manner in which Basis Cash may be said to be analogous to UST” and to particularise the “same reasons” why Basis Cash is said to have “failed for the same reasons as” UST did. ⁸⁰ This appeared relevant to the claimants’ claims under the tort of deceit against Terraform and Kwon, on the basis of their pleaded case that Terraform and Kwon “knew, or ought to have known, that the algorithm could not ensure UST’s price peg given that an analogous algorithmic stablecoin scheme, ‘Basis Cash’, was attempted by [Kwon] unsuccessfully. In fact, ‘Basis Cash’ had failed for the same reasons as UST”. ⁸¹ This allegation pertained to both Terraform and Kwon, since the claimants’ pleaded case was that “[a]s [Kwon] was the director and therefore the controlling mind of [Terraform], [his] knowledge can be attributed to [Terraform]”; ⁸²

(e) para 26 of Annex A requested for the claimants to “state with proper particularity which of the matters pleaded at paragraphs 12 to 25 [of the SOC] are said to constitute the alleged conspiracy”. ⁸³ This request pertained to the claims of tortious conspiracy to injure and the pleaded case of the claimants that “the [defendants] (or any two or more together) wrongfully and with intent to injure the Claimants by unlawful means, conspired and combined together to commit fraudulent misrepresentations on the Claimants causing injury to the Claimants”; ⁸⁴ and

(f) similarly, para 27 of Annex A requested for the claimants to “state with proper particularity which of the matters pleaded at paragraphs 12 to 27 [of the SOC] are said to constitute the alleged conspiracy”. ⁸⁵ This also pertained to the claims of tortious conspiracy to injure in the Suit, and the claimants’ pleaded case that “the 1st to 3rd Defendants (or any two or more together) wrongfully and with intent to injure the Claimants by unlawful means, conspired and combined together to defraud the Claimants and to conceal such fraud and the proceeds of such fraud from the Claimants”. ⁸⁶

87 These requests clearly did not concern Terraform’s jurisdictional challenge. The requests in (a)–(b) related to the merits of the claim for breach of a contract formed between Terraform and the claimants; the requests in (c)–(d) related to the merits of the claim in deceit; and the requests in (e)–(f) related to the merits of the claim in conspiracy. Such requests would only be necessary on the assumption that Terraform accepted the court’s jurisdiction and intended to engage it over the merits of the Suit.

88 Terraform argued, in relation to Annex A, that the court “should not take an unnecessarily narrow view of what can and cannot be asked in [F&BPs] for a jurisdictional challenge”. ⁸⁷ They submitted that even if a request may not clearly relate to jurisdiction, it may have relevance to the overall case which a party wishes to bring in its jurisdictional challenge, or may be necessary for the party to determine if the asserted claims fall within the scope of the arbitration agreement, or are needed to tie the claimants’ hands to make sure that they do not change their case once the jurisdictional challenge is brought. ⁸⁸ Terraform gave the example of its request for particulars as to whether the claimants had deposited UST on Anchor (at paras 6(g) and 8(h) of Annex A) – it argued that this request was “directly relevant to its jurisdictional challenge, even if this may not have been immediately obvious at the outset”. ⁸⁹ In this regard, Terraform in the Request Application stated that “the responses … by the Claimants to [Terraform’s] requests for F&BP are relevant to the grounds on which [Terraform] challenge[s] jurisdiction … and [Terraform] consider[s] it to be more efficient and consistent with the Ideals for the application for F&BP to be determined at the same time” [emphasis added]. ⁹⁰

89 However, save for making broad statements, Terraform did not explain how the requests for particulars at [86] above were relevant to its jurisdictional challenge. In so far as Terraform was attempting to argue that all of the Annex A requests were relevant to the jurisdictional challenge, it had earlier taken a contrary position: in the Request Application, under the header “Essence of intended application”, Terraform declared that “[the defendants] regard [the F&BP requests in Annex A] to be necessary in order that [the defendants] be able to understand and be able to prepare their defence to the Claimants’ claims” [emphasis added]. ⁹¹ This was similar to the situation in Dufferin (cited in Carona Holdings at [86]–[87]), where the defendant in its demand for particulars stated that it required them for the purpose of “drawing its statement of defence”. The court in Dufferin therefore concluded (at 25) that the defendant required the particulars not for the purpose of deciding whether to apply for a stay, but to defend the court proceedings.

90 Terraform also argued that it had made the Annex A requests on behalf of Kwon and Luna who were fully entitled to address the claimants’ claims in court (since they did not claim to be party to an arbitration agreement with the claimants). ⁹² This argument was not made in any affidavit, but only in written submissions before the AR and at the oral hearing before me. It was a non-starter. The requests were made also on behalf of Terraform and there was no reason for it to have joined in those requests if they were not intended for it. More importantly, it was clear that Terraform would directly benefit from the responses to the requests.

91 For the reasons noted above at [86]–[87], some of the Annex A particulars were clearly sought for the purpose of enabling Terraform to respond to the merits of the claimants’ pleaded case. The Annex A requests clearly went beyond the scope of what was necessary for the purpose of the jurisdictional challenge, and constituted a “step in the proceedings”.

92 In relation to Annex B, many of the F&BP requests centred on ascertaining the amounts which the claimants paid for their UST, the number of UST they purchased and sold, as well as the profit they made from their UST deposits on the Anchor Protocol. The aim of such requests appeared to be to ascertain the quantum for which Terraform may be liable to the claimants. For example: ⁹³

(a) para 1(a) requested for the claimants to “state in precise numerical terms” the amount paid for their UST, while para 1(d) requested for “the prevailing exchange rate” applied for each purchase of UST;

(b) para 1(b) requested for the claimants to state “the precise number of UST purchased in each tranche”, while para 5(a) requested for “the precise number of UST sold in each tranche”; and

(c) para 5(a) also requested for the precise consideration received for each UST sold and “the prevailing exchange rate” applied for each sale of UST, while para 6, similar to paras 6(g) and 8(h) of Annex A, requested for “a full account of the yield, earnings and/or profit received from such deposits on Anchor”.

93 Similarly, para 23 of Annex A requested for the claimants to state “precisely what ‘fiat currency or equivalent’ the [claimants] would have kept the precise sum in”. ⁹⁴ This request was in relation to the claimants’ claim in the SOC that if they had not purchased UST, they would likely have kept the purchase sum in “fiat currency or equivalent”. ⁹⁵ It appeared only relevant to the quantum of damages Terraform might be liable to the claimants.

94 Terraform’s act of requesting for particulars relating to the quantum of damages was inconsistent with its jurisdictional challenge. There was no reason why such information was needed unless Terraform intended to engage the court’s jurisdiction to assess the quantum of damages. Thus, such act constituted a “step in the proceedings”.

95 An application to strike out proceedings on the basis that it is unmeritorious is an act that signifies a submission to the court’s jurisdiction to resolve the dispute on the merits: L Capital Jones at [78].

96 In the SAPT Summons, Terraform sought to strike out and/or stay the Suit on various bases: ⁹⁶

(a) that the claimants’ pleaded claims are lacking in material averments and/or particulars and/or disclose no reasonable cause of action;

(b) that there is a lack of common interest or sufficient common interest between the claimants such that the action does not generate judicial economy, save costs, and/or it is otherwise inconsistent with the Ideals under O 3 r 1 of the ROC 2021 for the action to continue as representative proceedings, and the claims of all claimants other than the Representative Claimants be struck out or stayed;

(c) that the Representative Claimants are not in possession of and/or have not been authorised to bring these representative proceedings by written consents given by each of the other claimants;

(d) that some or all of the claimants are party to an arbitration agreement, and the claims of all such claimants be stayed pursuant to s 6 of the IAA; and

(e) that it is appropriate to strike out or stay all or part of the claimants’ claims as a matter of case management and/or in order to achieve the Ideals of the ROC 2021.

I deal with each of these bases in turn.

97 Basis (a) comprises two limbs – that the pleaded claims (i) are lacking in material averments and/or particulars; and (ii) disclose no reasonable cause of action. The first limb was clearly inconsistent with Terraform’s jurisdictional challenge – it was an invitation for the court to exercise its jurisdiction over the action by assessing whether the pleaded claims contained sufficient particularity.

98 In so far as an argument could be made, in relation to the first limb of basis (a), that a lack of particulars in the pleaded claims affected the propriety of the Suit as a representative action, and hence this basis for striking out went to the court’s jurisdiction over the Suit (and therefore was not inconsistent with Terraform’s jurisdictional challenge) – that is incorrect. To be clear, Terraform did not make this argument, but nevertheless I address it as it has implications for bases (b) and (c) of the striking‑out application.

99 In the first place, many of Terraform’s requests for F&BPs were not sought for the purpose of ascertaining whether the Suit was rightly brought as a representative action; for example, Terraform sought particulars relating to quantum of damages as well as the elements of deceit and conspiracy (see [87] and [94] above). In any case, and on a more fundamental level, a striking‑out application on the basis that the Suit was not properly brought as a representative action is inconsistent with a jurisdictional challenge. It is not a preliminary or threshold issue like the “fundamental procedural defect” in Maniach (at [84]) – even if the Suit was not properly brought as a representative action, the Representative Claimants are still rightfully before the court as far as their own claims are concerned. They will simply not be able to represent the other claimants. More importantly, if Terraform is contesting jurisdiction on the basis of the Arbitration Clauses, there is no reason for it to ask the court to determine the propriety of the Suit as a representative action. On the contrary, that expressly invites the court to exercise jurisdiction over the Suit.

100 Hence, bases (b) and (c), which essentially challenged the propriety of the Suit as a representative action, were inconsistent with the jurisdictional challenge and therefore constituted a “step in the proceedings”. This was particularly evident in the case of basis (b), where Terraform was asking the court to strike out the action on the grounds of a lack of common interest between the claimants. This entails asking the court to (a) examine the claims of each claimant (or class of claimants); (b) compare the significance of the common issues between the claimants with the significance of the issues which differed as between them; and (c) consider whether it was expedient for the court to determine all the claims in the Suit: see Koh Chong Chiah and others v Treasure Resort Pte Ltd [2013] 4 SLR 1204 (“Treasure Resort”) at [29], [60] and [86]. Thus, asking the court to find a lack of common (or sufficiently common) interest, such that the representative action should be discontinued, was asking the court to assume jurisdiction over the Suit and was incompatible with a desire to arbitrate.

101 In relation to the second limb of basis (a), ie, that no reasonable cause of action was disclosed, this basis for striking out relates directly to the merits or elements of the relevant cause(s) of action – it requires the court to examine the allegations in the pleadings and determine if there is a cause of action which has some chance of success: see Gabriel Peter & Partners (suing as a firm) v Wee Chong Jin [1997] 3 SLR(R) 649 at [21]. Terraform did not argue that no reasonable cause of action was disclosed because there was a fundamental procedural defect with the bringing of the Suit (as opposed to a deficiency on the merits). In any event, there was no such defect here – as explained above at [99], the issue of the propriety of the Suit as a representative action does not constitute such a defect. In the circumstances, the application to strike out the Suit on the basis that there was no reasonable cause of action could only have related to the merits of the claim. It therefore constituted a “step in the proceedings”.

102 I observe that in any event, Terraform did not explicitly argue that the striking-out application in the SAPT Summons related only to its jurisdictional challenge. Rather, Terraform’s position on the whole was that the applications in the SAPT Summons were “fallback” applications which its intended jurisdictional challenge took precedence over. ⁹⁷ I address this argument below at [108]–[115].

103 For completeness, bases (d) and (e) were not “steps in the proceedings”: (d) was the basis of Terraform’s jurisdictional challenge and (e) appeared related to (d) in that it was presumably asking for the Suit to be struck out or stayed on case management grounds because all or some of the claimants may be subject to an arbitration agreement.

104 In the SAPT Summons, Terraform sought the production of all of the claimants’ RAAs (Representative Action Agreements). ⁹⁸ This application was related to basis (c) of the striking‑out application – the Request Application stated that the RAAs were pertinent to the question of whether the Representative Claimants had complied with the requirement under O 4 r 6 of the ROC 2021 to have obtained written consents from all the represented parties to represent them in the Suit. ⁹⁹ Hence, Terraform sought production of the RAAs in order to “ascertain what is the true nature/scope of the represented parties’ consent (if any) to this action”. ¹⁰⁰ Terraform argued that this issue was particularly important given that it appeared from the F&BPs provided by the claimants on 12 January 2023 that the factual circumstances of many of the represented parties were inconsistent with the claimants’ pleaded claims, and also disclosed considerable disparity between the cases of these parties. ¹⁰¹

105 In its further arguments, Terraform raised the additional point that the requirement for written consents “is a purely procedural requirement under O 4 r 6(3) of the ROC 2021”. ¹⁰² Terraform argued that since the issue of compliance with court rules cannot be referred to arbitration, seeking relief from the court in respect of this requirement was not inconsistent with choosing to arbitrate, and could not be considered a “step in the proceedings”. ¹⁰³

106 I consider that the application for production of the RAAs was a “step in the proceedings”. As noted above at [99], the issue of the propriety of the Suit as a representative action was not necessary to advance the application to stay the Suit in favour of arbitration or preserve the right to do so. Indeed, whether the Representative Claimants had obtained proper written consents to bring the Suit on behalf of the claimants was irrelevant to whether the claim should be arbitrated. There was no reason for Terraform to seek the RAAs if it was bringing a jurisdictional challenge. The application was therefore inconsistent with Terraform’s jurisdictional challenge and constituted a “step in the proceedings”.

107 In any case, Terraform’s argument that the requirement for written consents dealt with a purely procedural issue rather than one relating to merits was only raised in further arguments. Prior to that, Terraform only relied on the general argument that the applications in the SAPT Summons were “fallback” applications in case the jurisdictional challenge failed. ¹⁰⁴

108 I did not accept Terraform’s argument that the reliefs sought in the Request Application and SAPT Summons were merely “fall-back” applications to its jurisdictional challenge, and therefore did not amount to a “step in the proceedings”. In this regard, Terraform relied on Zoom Communications at [45], where the court held that the “making of a stay application on improper forum grounds as a fall--back would not prejudice the primary application challenging the existence of the Singapore courts’ jurisdiction”, where the applications were intended to “fall to be considered in a cascading sequence, and not together”.

109 This argument failed on the plain wording of the Request Application and the SAPT Summons.

110  First, the SAPT Summons prayed for F&BP orders “pending determination of the 1st, 2nd and 4th Defendants’ application to strike out or stay proceedings” [emphasis added] and prayed for orders for the RAAs to be produced “pending determination” of the same. ¹⁰⁵ The clear intention was for these applications to be heard and disposed of first, not as an alternative to the application for a stay in favour of arbitration.

111 Second, as noted above at [96], the SAPT Summons prayed for a stay or striking out of the Suit on several grounds. ¹⁰⁶ These included, inter alia, the “stay in favour of arbitration”, alongside other substantive grounds on the merits (eg, that no reasonable cause of action had been disclosed). These varied grounds were being invoked either conjunctively or disjunctively, as evidenced by the use of the words “and/or” in the SAPT Summons. This was irreconcilable with the interpretation that the “stay in favour of arbitration” was the primary ground for a stay, with the other grounds being prayed for only in the alternative.

112 Third, in its Request Application, under the header “Date of filing of intended application if permission is granted”, ¹⁰⁷ Terraform stated that it filed the SAPT Summons together with its “application to challenge jurisdiction”, and explained that “[t]his was done in the interests of achieving the Ideals as there is substantial overlap in the factual and legal grounds of all the applications and [Terraform, Kwon and Luna] believed that it would be more time and cost efficient for the court to consider all matters on a comprehensive/holistic basis [emphasis added]”. ¹⁰⁸ Clearly, the aim was for all the applications in the SAPT Summons to be considered together, and not in a “cascading sequence”.

113 This conclusion was further fortified by Terraform’s reasons in the Request Application for why the SAPT Summons was necessary at that stage of proceedings: ¹⁰⁹

114 It was therefore clear that the application for a stay in favour of arbitration was not intended to be the primary application, with the other grounds a “fall-back” or alternative remedy. Moreover, Terraform wanted the matter of the production of RAAs to be “determined as a preliminary issue before the matter is allowed to proceed any further”, ¹¹⁰ making it clear that it was not intended to be a “fall-back” application.

115 Finally, in response to my query as to why Terraform did not simply file a jurisdictional challenge on 6 February 2023 as directed by the SAR, counsel for Terraform explained that it was a “strategic decision”, because they “thought it might be efficient for everything to go before the Court and have it dealt with by way of an omnibus application, hearing all the applications in turn”. ¹¹¹ The first part of this reply was revealing – it betrayed the fact that the prayers sought were meant to go before the court together for the sake of (alleged) efficiency, as opposed to being “fallback” applications. As to the second part of the reply – that the applications were meant to be heard “in turn” – that did not make clear the order in which the applications were meant to be heard, and importantly, as just noted, the Request Application and the SAPT Summons made clear that the applications were meant to be heard together, not “in turn” or in a “cascading” manner.

116 The question remains why Terraform did not simply file its jurisdictional challenge by 30 January 2023 as directed by the SAR, like Platias did (see [21] above). ¹¹² In further arguments, Terraform argued that “[o]nce [it] received the [F&BPs] and became aware of its right to arbitrate, it duly applied for a mandatory arbitration stay. It should not matter that it did so under an omnibus application/SAPT instead of a standalone Section 6 (1) IAA application”. ¹¹³ This is misconceived for several reasons. First, as explained above at [91] and [94], Terraform’s application for F&BPs went beyond what was relevant for the jurisdictional challenge. Second, on Terraform’s own argument, it could have filed a standalone stay application, as it did on 6 March 2023, without asking for the other prayers. This completely undermines its position. Third, based on the structure of the ROC 2021, a Single Application Pending Trial (“SAPT”) is only relevant and invoked after the jurisdictional challenge is heard and dismissed: see O 9 r 8 read with O 9 r 9 of the ROC 2021, which makes clear that a SAPT may only be made after any jurisdictional challenge has been dealt with. Terraform was effectively re-writing the ROC 2021 to suit its purposes. Fourthly, although Terraform urges that its “omnibus” approach was intended to save time and costs and deal with the matter in a manner consistent with the Ideals, ¹¹⁴ it is evident that it was engaged in the complete opposite. It could, and should, have filed the standalone stay application as soon as possible, and not until its “strategic decision” had been rebuffed.

117 I was of the view that Terraform had taken a “step in the proceedings”. In this regard, I note that the court in assessing whether an act constitutes a “step in the proceedings” should consider the entirety of the circumstances surrounding the defendant’s acts in a practical and commonsensical way: Carona Holdings at [52]. As indicated above at [71] and [75], I considered Terraform’s filing of the defence on the merits and the counterclaim to be a “step in the proceedings”. This was confirmed or reinforced by its subsequent conduct in filing the Request Applications and the SAPT Summons, and the various reliefs sought therein. These plainly demonstrate that Terraform had employed court procedures to enable it to defeat or defend the Suit on the merits (L Capital Jones Ltd at [77]); affirmed the correctness of the court proceedings and its willingness to go along with the court’s determination (Australian Timbers at [19]); and waived its right to object to the court’s jurisdiction (Zoom Communications at [43]).

118 Terraform’s multiple reservations did not alter this conclusion – they were simply at odds with the way the applications sought to advance the court proceedings and demonstrated approbation and reprobation on the part of the defendant which should not be countenanced: Shanghai Turbo at [36], citing Carona Holdings at [101].

119 My findings dispose of Terraform’s appeal, and consequently, the other defendants’ applications for case management stays. I shall nonetheless deal with the second issue as it was the primary basis on which the AR decided the applications.

120 It is settled law that a court hearing an application for a stay of proceedings in favour of arbitration should grant such stay if the applicant is able to establish a prima facie case that:

(Tomolugen Holdings Ltd and another v Silica Investors Ltd and other appeals [2016] 1 SLR 373 (“Tomolugen”) at [63])

121 The court will only refuse to grant a stay where it is clear on the evidence that one or more of the requirements have not been satisfied: Tomolugen at [64]. The goal of the prima facie standard of review is to provide confidence that “the court will stay the proceedings in favour of arbitration except in cases where the arbitration clause is clearly invalid or inapplicable” [emphasis in original]: Tomolugen at [68]. Other cases have similarly emphasised that it is only in the clearest of cases that the court ought to find an arbitration agreement invalid or inapplicable: Tjong Very Sumito and others v Antig Investments Pte Ltd [2009] 4 SLR(R) 732 (“Tjong Very Sumito”) at [24]; Malini Ventura v Knight Capital Pte Ltd and others [2015] 5 SLR 707 (“Malini Ventura”) at [36]. Thus, the prima facie standard in Tomolugen is a low and unexacting threshold.

122 Pertinently, the same prima facie test is to be applied for all jurisdictional issues, be it the existence or application of an agreement to arbitrate the dispute that is before the court (see Tomolugen at [63] and Art 16(1) of the UNCITRAL Model Law on International Commercial Arbitration 1985 (“the Model Law”).

123 The principle underlying the prima facie approach is that of kompetenz‑kompetenz – ie, the jurisdiction of an arbitral tribunal is a matter reserved for the determination of the tribunal itself: see Tomolugen at [65]–[68]. This is a cardinal principle in the international arbitration regime (see Art 16(1) of the Model Law). As explained in the Explanatory Note on the UNCITRAL Model Law on International Commercial Arbitration 1985 with amendments as adopted in 2006, UN Sales No E.08.V.4 (2008) (“the Explanatory Note”) at para 25, “‘Kompetenz–Kompetenz’ means that the arbitral tribunal may independently rule on the question of whether it has jurisdiction, including any objections with respect to the existence or validity of the arbitration agreement, without having to resort to a court” [emphasis added]. Although the 2006 amendments to the Model Law have not been adopted in Singapore, Art 16(1) was left untouched by those amendments, and thus the comments in the Explanatory Note remain relevant in our context.

124 The second and third requirements in Tomolugen were not in issue – the claimants accepted that they were satisfied. The only point of dispute was the first requirement – ie, whether there was a valid arbitration agreement between the claimants and Terraform. ¹¹⁵

125 The claimants gave two reasons why a prima facie case of a valid arbitration agreement had not been shown by Terraform:

(a) first, there was no undisputed evidence of an agreement to arbitrate; ¹¹⁶ and

(b) second, the Terraform Terms of Use and the Anchor Terms of Service were not incorporated into the contracts between the claimants and Terraform, as users of the Terra and Anchor Websites would not have had notice of the Terms of Use/Terms of Service which contained the Arbitration Clauses. ¹¹⁷

126 As noted above at [30]–[32], the AR found that Terraform failed to show on a prima facie basis that there was a valid arbitration agreement between it and the claimants. She reasoned that the Arbitration Clauses had not been validly incorporated into the alleged contracts between Terraform and the claimants as a reasonable user of the websites would not have been put on notice of either the Terra Terms of Use or the Anchor Terms of Service, which contained the Arbitration Clauses.

127 At the hearing before me, the claimants showed images of the state of the Terra Website as at December 2020 and December 2021.

128 With respect to the December 2021 image, the claimants showed that the hyperlink for the Terra Terms of Use was “buried” at the bottom of the webpage, and was relatively obscure and/or lacked prominence as compared to the other materials and hyperlinks (promotional material for the Terra Ecosystem) on that webpage. ¹¹⁸ The claimants submitted that the court had to approach the inquiry from the perspective of the “user experience”, and suggested that the other features on the webpage, including the link to the Terra White Paper at the top of the webpage, information regarding the Terra Ecosystem, and the other promotional banners and images thereon, were much more attention--grabbing than the hyperlink for the Terra Terms of Use. ¹¹⁹

129 As for the Anchor Terms of Service, Terraform submitted that some of the claimants, including Gan, concluded an arbitration agreement with Terraform by expressly or constructively accepting the Anchor Terms of Service when connecting their wallets to the Anchor Protocol. ¹²⁰ Terraform argued that these claimants admitted to depositing UST on the Anchor Protocol – in order to do so, they had to connect their cryptocurrency “wallets” to the Anchor Protocol, which required them to click on the “Connect Wallet” link on the Anchor Website. This would have triggered a pop-up containing a notice stating “[b]y connecting, I accept Anchor’s Terms of Service”. ¹²¹ Thus, by connecting their wallets to the Anchor Protocol, the relevant claimants would have accepted the Anchor Terms of Service.

130 The claimants pleaded, inter alia, that by purchasing UST, they had entered into an agreement with Terraform whereby Terraform guaranteed the terms set out in the Terra Representations (ie, relating to the stability of UST and Anchor), according to the applicability of each Representation to the individual claimants. ¹²² I note the possibility that the Arbitration Clauses applied as separate agreements between the claimants and Terraform, rather than being incorporated into the unilateral contracts allegedly formed upon the claimants’ purchase of UST. In any event, the law relating to the formation of contracts online – particularly the application of general terms of use – was pertinent.

131 In Dialogue Consulting Pty Ltd v Instagram Inc [2020] FCA 1846 (“Dialogue”), the court organised contracts formed online into three categories (at [241], [244] and [249]):

(a) “clickwrap” agreements require the user to scroll through the terms of use and affirmatively click a button or tick a box stating words to the effect of “I agree” before accessing the site;

(b) “sign-in wrap” agreements are where users are notified of terms that are available by way of a hyperlink, and are required to click a button, or sign in, in order to access the site; and

(c) “browse-wrap” agreements are where a website displays a notice or a banner notifying the user that they agree to the site’s terms of use by using the site. The user is not required to click any button, nor take any affirmative action to indicate their acceptance of the terms.

132 The defendants argued that in respect of “browse-wrap” principles, case law makes clear that the question of incorporation should be approached as a fact-sensitive inquiry: Lopez v Terras Kitchen LLC 331 F Supp 3d 1092 (SD Cal, 2018). ¹²³ In this regard, the claimants sought to rely on the holding in Dialogue (at [217]) that an arbitration agreement contained within the terms of use/service would bind a user of a website only if (a) the user of the website had a reasonable opportunity to consider the terms; and (b) by its conduct, the user indicated that it had accepted the terms. ¹²⁴

133 However, Dialogue should be applied with caution as the court explicitly stated (at [196]–[202]) that it was declining to apply the prima facie standard for assessing the existence of an arbitration agreement. Rather, the court applied the balance of probabilities standard: Dialogue at [202]. Hence, it would be incorrect to apply the threshold articulated in Dialogue for present purposes.

134 Several other cases were cited by the claimants to show that in the context of browse-wrap agreements, the courts would require substantial notice of the relevant terms of use to be given to the user of the website. These cases include Be In Inc v Google Inc 2013 WL 5568706 (ND Cal, 2013) (“Be In Inc”); Green v Petfre (Gibraltar) Ltd (trading as Betfred) [2021] EWHC 842 (QB) (“Green”); and Nguyen v Barnes & Noble Inc 763 F 3d 1171 (9th Cir, 2014) (“Nguyen”). ¹²⁵

135 I did not find these cases helpful or applicable. Be In Inc and Green did not involve arbitration clauses. In Be In Inc (at *30), the terms sought to be incorporated related to prohibitions on the use, copying, or distribution of the website’s content. In Green (at [166]), the relevant term was an exclusion of liability clause. The courts in these cases were addressing the issue of whether those terms had been incorporated into the contracts – this appeared to involve the usual civil standard of the balance of probabilities, as opposed to the prima facie standard which applies when assessing the existence of a valid arbitration agreement. The pronouncements in those cases relating to the incorporation of non-arbitration related clauses were therefore irrelevant to the present case.

136 Nguyen did involve an arbitration clause. However, as acknowledged by counsel for the claimants, the court in Nguyen did not mention the prima facie test. ¹²⁶ In fact, the reasoning in the decision made clear that the court did not apply a prima facie standard. The court delved into a detailed assessment of the proximity and conspicuousness of the relevant hyperlink and the case law on what constituted reasonable notice to a user of a website. The court concluded that “[i]n light of the lack of controlling authority on point, and in keeping with courts’ traditional reluctance to enforce browsewrap agreements against individual consumers, we therefore hold that where a website makes its terms of use available via a conspicuous hyperlink on every page of the website but otherwise provides no notice to users nor prompts them to take any affirmative action to demonstrate assent, even close proximity of the hyperlink to relevant buttons users must click on – without more – is insufficient to give rise to constructive notice”: Nguyen at **17. On that basis, the court held that the user of the website did not enter into an arbitration agreement with the operator of the website: Nguyen at **21. Clearly, the court’s conclusion in Nguyen was not based on a prima facie assessment of the issue.

137 I agree with the general approach adopted in the aforementioned cases of taking a fact-centric approach to the questions of whether a user had actual or constructive notice of the relevant terms as well as what constitutes constructive notice. Nonetheless, in the context of an application for a stay in favour of arbitration, the applicable test is whether the applicant can establish a prima facie case that the user did have notice of the arbitration clause. I found that on the facts of the present case, it cannot be said to be clear and obvious that no agreement to arbitrate was concluded between Terraform and the claimants.

138 The claimants argued that the Terra Terms of Use fell within the “browse-wrap” category, since users were not required to assent to the Terra Terms of Use before using the Terra Website. ¹²⁷ In contrast, Terraform argued that “browse-wrap” agreements usually concerned websites which offered items or software for purchase, ¹²⁸ and since the Terra Website did not offer UST for sale, there was no contract between the claimants and Terraform in the first place, to which the Terms of Use could be incorporated by “browse-wrap” or otherwise. ¹²⁹

139 First, I address the point of whether there was a contract between the claimants and Terraform in the first place. In this regard, the fact that the Terra Website did not offer UST for sale did not, by itself, mean that no contract could have been concluded between the claimants and Terraform. The pleaded contract was one under which Terraform guaranteed certain features of UST and the Terra Protocol, and the claimants purchased UST – it did not matter where the UST was purchased from. In the circumstances, it was “at least arguable” that contracts on the terms pleaded by the claimants had been formed between the claimants and Terraform, such that the prima facie threshold was satisfied in this regard (see Tjong Very Sumito at [24]).

140 Next, I address the question of whether a prima facie case could be made out that the Arbitration Clause found in the Terra Terms of Use was incorporated via the “browse-wrap” analysis. Given the way the claimants’ case was pleaded (see below), I did not consider it appropriate to approach the question of incorporation by attempting to fit the present factual matrix within one of the three categories set out in Dialogue. Instead, the real issue was whether a prima facie case could be established that the claimants had actual or constructive notice of the Arbitration Clauses.

141 It was the claimant’s pleaded case that all 377 of them had been induced to purchase UST in reliance on, inter alia, the Terra Representations. ¹³⁰ According to the claimants, the first and second Terra Representations were detailed in the Terra White Paper, which was accessed via a hyperlink on the homepage of the Terra Website. ¹³¹ The Terra Representations were also found on a webpage accessed by clicking on the Terra Website’s “Learn more” tab. ¹³²

142 Hence, the claimants’ pleaded case was premised on the claimants accessing the Terra White Paper via the hyperlink on the homepage of the Terra Website and going through other parts of the Terra Website. However, in the same breath, the claimants argued that the reasonable user would have stopped short of going to the end of the Terra Website or would not have noticed the hyperlink to the Terra Terms of Use. ¹³³ This appeared to be a highly selective proposition of what a visitor to the Terra Website would choose to read.

143 In response to this, counsel for the claimants argued that users of the Terra Website would have accessed it to find out more about UST. ¹³⁴ Hence, it was expected that they would seek to access the hyperlink to the Terra White Paper. ¹³⁵ Further, this hyperlink was located in a prominent position at the header of the Terra Website, whereas the hyperlink to the Terra Terms of Use was “buried” at the bottom of the page. ¹³⁶

144 This argument misses the point. This was not a case where the user was accessing the Terra Website to make a transaction and would therefore only have notice of those pages or portions relevant to that transaction. The claimants’ case was that they accessed the Terra Website to learn more about UST, and in the process, accessed and read the Terra White Paper and other portions on the Terra Website, which they say contained the Terra Representations. If that was the case, then it was not unreasonable that they would, in perusing the Terra Website, have also noticed the hyperlink to the Terra Terms of Use, which contained the Arbitration Clause.

145 Counsel for the claimants raised the case of Brett Long v Provide Commerce Inc 245 Cal App 4th 855 (2016) (“Brett Long”), where in the context of an e-commerce website selling flowers, the court held (at 866) that a user was not given reasonable notice of the terms of use hyperlinks in the flow of webpages in the checkout process. ¹³⁷ However, as noted above, the Websites in the present case were not e-commerce websites, where one enters into the relevant contract for sale of goods or services typically without accessing hyperlinks to the terms of use. In contrast, the Terra Website was an informational website which contained various documents accessible by hyperlink which users would access to gain information on the Terra Ecosystem and related products. There was no particular sequence of pages or links which the user would be directed or put through before exiting the webpage. Thus, the reasoning adopted in Brett Long was not applicable here.

146 The above discussion makes clear that a detailed assessment of the facts of this case, as well as the law governing the incorporation of arbitration clauses in online contracts, is necessary in order to reach a conclusion, on the balance of probabilities, as to whether the Terra Terms of Use were incorporated or applied in the (alleged) contract between the claimants and Terraform. The question of incorporation, which the claimants argued depended on whether the claimants had actual or constructive notice of the Terra Terms of Use, hinged on mixed issues of fact and law which are rightfully to be addressed by an arbitral tribunal:

(a) first, the issue of actual notice depends upon the weight and credibility accorded by the arbitral tribunal to the assertions of individual claimants as to whether they had seen the Terra Terms of Use when they visited the Terra Website to view the Terra Representations. Beltran and Gan both denied actual notice. ¹³⁸ No evidence was adduced as to the position of the other claimants. The tribunal would have to decide the weight to be accorded to such denials and the credibility of such claims when considered against the other objective evidence, including the relevant placing of the Terra Terms of Use hyperlink on the Terra Website, and whether it is believable that the claimants had actual notice of other hyperlinks on the Terra Website – such as that for the Terra White Paper – while having no actual notice of the hyperlink for the Terra Terms of Use. The answer to that question cannot be said to be clear and obvious, particularly in the context of how the claimants have pleaded their cause(s) of action and claim to have come across the Terra Representations; and

(b) second, for the issue of constructive notice, the claimants argued that it must be determined whether the Terra Website provided reasonable notice of the Terra Terms of Use to the claimants. ¹³⁹ This is a fact-sensitive inquiry, requiring the tribunal to determine whether the Terra Terms of Use were reasonably prominent or conspicuous, considering the location and labelling of the hyperlink amidst the broader layout and design of the Terra Website. The answer to this inquiry likewise cannot be said to be clear or obvious in this case.

147 Overall, the determination of these issues would require “the court to descend into a protracted examination of the evidence to make a finding on the merits that an arbitration agreement exists [or does not exist] on a balance of probabilities at the stay stage”: The “Titan Unity” [2013] SGHCR 28 (“The “Titan Unity””) at [34], cited with approval in Tomolugen at [58]. This would be inconsistent with a prima facie ascertainment of the existence of an arbitration agreement: The “Titan Unity” at [34]. Adopting the language of the Court of Appeal in Tjong Very Sumito at [24], it was certainly “at least arguable” that the claims in the Suit were subject to the Terra Terms of Use, and this case did not satisfy the threshold of “the clearest of cases … [where] the court ought to make a ruling on the inapplicability of an arbitration agreement”.

148 Consequently, I found that Terraform had made out a prima facie case that an arbitration agreement existed between it and all the claimants on the basis of the Terra Terms of Use.

149 For completeness, I address the claimants’ submission that the hyperlink to the Terra Terms of Use was absent on the Terra Website as of 26 December 2020. ¹⁴⁰ Counsel for the claimants was unable to confirm if and how many claimants purchased UST based on this version of the Terra Website. ¹⁴¹ Counsel for Terraform, Kwon and Luna informed me that none of the claimants purchased UST in or before December 2020. ¹⁴²

150 According to their affidavits, Beltran accessed the Terra Website “on or around 5 May 2022” to read the Terra White Paper, ¹⁴³ and Gan accessed the Terra Website, first, “in or around December 2020” to read the Terra White Paper, and “again in or around December 2021 and/or January 2022”. ¹⁴⁴ Hence, both Representative Claimants had accessed the Terra Website after December 2020. There was no evidence as to when the other claimants accessed the Terra Website.

151 Thus, for the purposes of the prima facie test, it was not clear and obvious that any or all of the claimants only visited the Terra Website before the hyperlink to the Terra Terms of Use was included.

152 I also found that Terraform had made out a prima facie case that arbitration agreements were formed between it and some of the claimants on the terms of the Anchor Terms of Service.

153 Pertinently, not all the claimants pleaded that they were induced to purchase UST in reliance on the Anchor Representations. The eight claimants under Schedule 1 of the SOC claimed to have been induced only by the Terra Representations. ¹⁴⁵ The remaining 369 claimants in Schedules 2 and 3 plead that they were induced by, inter alia, the Anchor Representations, ¹⁴⁶ and that they had entered into a contract with Terraform on the terms of those representations. ¹⁴⁷ Based on my finding that there was a prima facie case that the Terra Terms of Service applied to all the claimants, the issue of whether the Anchor Terms of Service applied to the relevant claimants was, strictly speaking, irrelevant. Nevertheless, for completeness, I address this issue.

154 The claimants argued that the Anchor Terms of Service were also incorporated via “browse-wrap”. ¹⁴⁸ However, the defendants argued that the Anchor Terms of Service were incorporated via “sign-in wrap”. ¹⁴⁹

155 I did not consider this distinction to be particularly important to the issue of whether there was a prima facie case that the Anchor Terms of Service had been incorporated. The key inquiry remained whether, on a prima facie basis, the claimants had actual or constructive notice of the Anchor Terms of Service.

156 It was the claimants’ pleaded case that the Anchor Representations were found either on the Anchor Website itself, or in the Anchor White Paper (which was accessible through a hyperlink on the homepage of the Anchor Website). ¹⁵⁰ The claimants argued that users of the Anchor Website were not required to assent to the Anchor Terms of Service when using the Website. ¹⁵¹ Similar to the argument in respect of the Terra Terms of Use, the claimants argued that the Anchor Terms of Service hyperlink was inconspicuous and the relevant claimants did not have reasonable notice of the Anchor Terms of Service prior to entering into a contract with Terraform. ¹⁵²

157 As described above at [129], Terraform pointed to the affidavit of Mr Amani, in which he gave evidence that the only way a user of the Anchor Protocol’s “wallet” service could “connect” his wallet would be to enter the Dashboard of the Anchor Website and click on the option to “connect” the wallet. This would trigger a “pop-up” notification that includes the acknowledgment “[b]y connecting, I accept Anchor’s Terms of Service”. ¹⁵³ The “pop–up” notice appears as follows (at the bottom-right corner of the image): ¹⁵⁴

158 Crucially, 351 claimants, including Gan, confirmed in response to F&BP requests that they had “staked” their UST on the Anchor Protocol. ¹⁵⁵

159 However, the claimants argued that the relevant claimants had purchased UST before connecting their cryptocurrency wallets to the Anchor Protocol. ¹⁵⁶ The claimants submitted that the subsequent connecting of these claimants’ wallets to the Anchor Protocol, after purchasing UST, did not incorporate the Anchor Terms of Service into the contracts which had been concluded at the point of purchase. ¹⁵⁷

160 However, I note that it was uncontested that at least some of the claimants “staked” their UST on the Anchor Protocol in “tranches”, and therefore purchased more UST after having first connected their “wallet” on the Anchor Website. ¹⁵⁸ This meant that they would have seen the “pop-up” notice referencing the Anchor Terms of Service before they purchased more UST. Gan was one such claimant – he purchased more UST on 18 February 2022, after he had staked some UST on the Anchor Platform on 2 January 2022. ¹⁵⁹ However, Gan claimed that he “did not notice the [Anchor] Terms of Service and so did not access it”, and that although he “did access the ‘Dashboard’ page of the Anchor website … [he did] not recall seeing any link to the Terms of Service on the ‘Dashboard’ page”. ¹⁶⁰ He stated further that he did not agree with Terraform’s assertion that he would have accepted the Anchor Terms of Service by clicking on the “Connect Wallet” link when connecting his wallet to the Anchor Protocol, since he “did not click the link in the [pop-up] notice that would bring [him] to Anchor’s Terms of Service”. ¹⁶¹

161 I found Gan’s statements to be vague. He essentially denied noticing or clicking the link to the Anchor Terms of Service when connecting his wallet – but this did not affect the issue of whether there was constructive notice of the Anchor Terms of Service. In my view, the “pop-up” notice on the Anchor Website constituted strong evidence that the claimants who staked their UST on the Anchor Protocol, and then purchased more UST, had either actual or constructive notice of the Anchor Terms of Service when they purchased their second and subsequent tranches of UST. In the circumstances, it suffices to say that it was “at least arguable” that on the basis of the Anchor Terms of Service, there was a prima facie agreement to arbitrate between some of the claimants and Terraform.

162 An interesting issue that arose was how the court should address a situation where a prima facie case of an arbitration agreement is made out only in respect of some of the claimants in a representative action. This issue was ultimately not engaged on the facts, since Terraform had taken a “step in the proceedings”, and in any event, all the claimants were subject to a prima facie arbitration agreement, at least on the basis of the Terra Terms of Use. Nevertheless, I make a few observations.

163 The claimants submitted that in the normal trajectory of a representative action, the court would determine the outcome at trial based solely on the claims of the representative claimants. ¹⁶² They argued that the court may take into account differences in the claimants’ respective positions at the assessment of damages stage, where the assessor “will examine each and every represented person’s claim”, and may vary the amount of damages an individual claimant is entitled to (or find that an individual claimant is only entitled to nominal damages) on the basis of those differences: referring to the approach in Treasure Resort at [77] and [110]. ¹⁶³ The claimants submitted that the same approach should apply when assessing the existence of an arbitration agreement vis-à-vis the claimants in a representative action – ie, the court should consider whether there is a binding arbitration agreement between the defendant and the representative claimants only, ¹⁶⁴ with any differences in the positions of the claimants (in respect of whether they are subject to an arbitration agreement) to be considered at the assessment of damages stage.

164 This approach is problematic and is not consistent with established principles.

165 First, in the context of an arbitration agreement, the court is concerned with a jurisdictional issue, which the court in Treasure Resort was not. This is a threshold issue, and not one which can be dealt with after the conclusion of trial. It therefore makes little sense, and is contrary to principle, to determine the existence of an arbitration agreement only in respect of the representative claimants, to proceed on that basis for all claimants, and then to make the necessary adjustments/corrections after the conclusion of trial (as the claimants suggested), when the issue of damages is considered. The principle of party autonomy demands that binding arbitration agreements be respected (see Tjong Very Sumito at [28]), and under the principle of kompetenz-kompetenz, the arbitral tribunal is the body charged with determining whether the relevant parties and the dispute before it are subject to an arbitration agreement (Tomolugen at [65]–[68]). The court would be acting in violation of both principles if it decides a dispute in respect of claimants who are prima facie bound by arbitration agreements. Further, in this situation, the defendant may be forced to take a step in the proceedings to defend itself in court, thereby prejudicing its right to arbitrate in respect of the represented claimants who are parties to an arbitration agreement. Claimants would also be encouraged to avoid arbitration agreements and game the system by putting forward representatives who are not clearly bound by an arbitration agreement.

166 Second, the claimants’ approach is inconsistent with how a representative action works. Under O 4 r 6 of the ROC 2021, a representative action may proceed where the claimants “have a common interest” in the proceedings. In conducting this assessment, the court must examine the claims and compare the significance of the common issues between the claimants with the significance of the issues which differed as between them: Treasure Resort at [29] and [60]. Further, the court must assess whether there are any other circumstances which justify discontinuing the representative action: Treasure Resort at [29] and [86]. This assessment is a preliminary one in the life cycle of a representative action and cannot be delayed until the assessment of damages.

167 In AnAn Group (Singapore) Pte Ltd v VTB Bank (Public Joint Stock Co) [2020] 1 SLR 1158 (“AnAn”), the Court of Appeal held, in the context of winding-up proceedings, that winding-up applications cannot be used as a method of circumventing parties’ agreement to submit their disputes to arbitration: AnAn at [75]–[80]. Thus, the applicable threshold for allowing a stay of a winding-up application, where the debt concerned was subject to an arbitration agreement, could not be the usual “triable issue” standard applied to claims which are not subject to arbitration. Rather, the “prima facie standard” applied, such that the winding-up proceedings would be stayed or dismissed if the debtor could show, on a prima facie basis, that the disputed debt fell within the scope of a valid arbitration agreement between the parties: AnAn at [56].

168 Similarly, representative actions cannot be used as a method of circumventing the principle of party autonomy. The court cannot, on the basis that the representative claimants are not subject to an arbitration agreement, proceed to determine the claims of other claimants who may be bound by arbitration agreements. The principle of party autonomy demands that the court assess whether a prima facie applicable arbitration agreement exists between each individual claimant and the defendant, and where it exists, such claims must be referred to the tribunal.

169 Fundamentally, the court’s assessment of the propriety of a representative action is at odds with the limited scope of the court’s duty where there is potentially an arbitration agreement covering the dispute. In determining whether to allow a representative action to proceed, the court exercises jurisdiction over the different claimants by examining their claims and circumstances. The court should not be doing so if there is a prima facie arbitration agreement governing the dispute.

170 There are also practical issues which arise where arbitration agreements interact with a representative action. First, the assessment of whether the claimants are subject to an arbitration agreement may require the court to conduct an individualised assessment of each claimant’s claim. To require the court to conduct such an assessment would contradict the very purpose of representative actions, which is to provide for “a final determination of the rights of all the claimants vis-à-vis the defendant … based solely on the claims brought by the representative plaintiff and the defence raised by the defendant” [emphasis in original omitted; emphasis added]: Treasure Resort at [26]. I note that the individualised assessment of whether each claimant is subject to an arbitration agreement is a more invasive inquiry than the general assessment of whether common issues exists across all claimants; the former requires an in‑depth inquiry into each claimant’s circumstances, while the latter involves a broader assessment which does not concentrate on individual circumstances – see Treasure Resorts at [112], where the court declined to find one of the claims to be common to all claimants where it would have required an examination of each claimant’s circumstances.

171 Second, represented claimants are not obliged to give discovery or evidence relating to the claims in the representative action, and the defendant will not be entitled to cross-examine the represented persons: Treasure Resort at [36]. Thus, all the court will have before it is the evidence of the representative claimants, from which it will be unable to ascertain with precision whether a prima facie applicable arbitration agreement exists between the represented claimants and the defendant.

172 Third, the representative claimants may not have the requisite standing to argue whether the represented claimants are each subject to a prima facie applicable arbitration agreement – to satisfy O 4 r 6(2) of the ROC 2021, it is only necessary that they consent to the representative claimant bringing an action in court on their behalf.

173 What then should the court do where the defendant asserts that some of the claimants in a representative action are bound by an arbitration agreement? This is a tricky issue as arbitration agreements pose significant problems for representative actions.

174 The courts in the US have grappled with some of these issues in the context of class actions: see Emily Villano, “Arbitration Asymmetries in Class Actions” (2022) 131 Yale LJF 742. The prevailing approach appears to be to deny certification of a class action (a threshold assessment before a class action may proceed) where some members of the putative class may be subject to an arbitration agreement: see, for example, Tan v Grubhub, Inc 2016 WL 4721439 (ND Cal, 2016) (“Tan”); Parrish v Gordon Lane Healthcare, LLC 2023 WL 7107267 (CD Cal, 2023) (“Parrish”); Berman v Freedom Fin Network, LLC 400 F Supp 3d 964, 986 (ND Cal, 2019). The threshold for denying certification on this ground is a low one – courts have explained that the “mere potential that the relevant arbitration provision is valid is sufficient to preclude a named plaintiff who opted out of the provision from representing a class largely made up of individuals that may be subject to the agreement”: Jensen v Cablevisions Systems Corp 372 F Supp 3d 95 (ED NY, 2019) at 123. Thus, US courts have denied certification on the basis that the claims of the members of the putative class are “likely barred”, “potentially subject”, or “may be bound by” arbitration agreements: see Villano at 746, citing, respectively, Forby v One Techs 2020 WL 4201604 (ND Tex, 2020) at *9; Tan at *3; and Conde v Open Door Mktg 223 F Supp 3d 949 (ND Cal, 2017) (“Conde”) at 961.

175 The reasons given by US courts for denying certification on this ground span the various issues identified above at [165]–[172] in respect of the incompatibility of collective proceedings and arbitration agreements. Courts have cited a lack of commonality of interests between the claimants (because some are subject to arbitration agreements) (see Tan at *2); a lack of standing on the part of the class representative to address the issue of whether the putative class members are subject to applicable arbitration agreements (see Conde at 960); and the fact that an individualised inquiry into the putative class members’ circumstances would be necessary to determine the applicability of an arbitration agreement to them (see Ehret v Uber Techs, Inc 148 F Supp 3d 884, 902-03 (ND Cal, 2015) at [21]).

176 There is no certification process under Singapore law for representative actions. Rather, the structure of the ROC 2021 is such that the propriety of a representative action appears only to be assessed when a defendant brings a challenge against it: see O 4 r 6. Hence, the issue of whether the court should allow a representative action (where some claimants may be subject to an arbitration agreement) to proceed necessarily comes into play in an application to stay the proceedings in favour of arbitration.

177 Thus, if the representative claimants themselves are subject to a prima facie arbitration agreement, that would be the end of the matter – they would not be capable of representing the other claimants in the action: see the Ontario Federal Court’s decision in Stephanie Difederico and Jameson Edmond Casey v Amazon.com (2022) FC 1256 at [2]. However, if the representative claimants are not subject to a prima facie arbitration agreement but some of the represented claimants are, the only courses of action appear to be:

178 I observe that policy concerns arise where collective proceedings are discontinued or stayed because some claimants may be subject to an arbitration agreement: see Villano at pp 751–755. In the US, arbitration agreements pose a formidable obstacle to the prosecution of class actions: see the decisions of the US Supreme Court in AT&T Mobility LLC v Concepcion 563 US 333 (2011) and American Express Co v Italian Colors Restaurant 570 US 228 (2013) (“American Express”). Class arbitration is not permitted unless the arbitration agreements clearly provide for it (Lamps Plus Inc v Varela 139 S Ct 1407 (2019)), and thus consumers who are precluded from bringing their claims to court are unlikely to ever bring arbitral proceedings against defendant companies, given the typical gulf in resources at the disposal of either side and the small value of individual claims. As observed by the court in American Express (citing Green Tree Financial Corp-Alabama v Randolph 531 US 79 (2000) at 90), “[i]t may well be that the existence of large arbitration costs could preclude a litigant … from effectively vindicating her federal statutory rights in the arbitral forum”. Thus, arbitration agreements potentially signal a death knell for representative actions, which exist precisely to bridge the gulf in financial resources and motivation between individual consumers and large companies.

179 The Canadian courts have attempted to address these policy concerns by striking down an arbitration agreement where it gives rise to an issue of accessibility – ie, where referring a dispute to arbitration would result in a real prospect of the challenge not being resolved: see Uber Technologies Inc v Heller (2020) SCC 16 (“Uber”) at [32]–[46]. In Uber, the fees payable by the representative claimant to initiate the arbitration were equivalent to his annual income and were effectively a barrier to bringing a claim. The court found that the arbitration agreement was unenforceable because of unconscionability as (a) there was clearly inequality of bargaining power between the defendant and the representative claimant in entering into the arbitration agreement; (b) the bargain concluded unduly advantaged the stronger party and unduly disadvantaged the vulnerable party; and (c) the arbitration agreement violated the claimant’s reasonable expectations by depriving him of remedies: Uber at [93]–[98].

180 Such an approach appears to significantly extend the doctrine of unconscionability (as noted in Brown JJ’s concurring opinion in Uber at [103]) and be out of step with the prevailing stance in Singapore to support and facilitate arbitration. Our courts have consistently emphasised the principle of party autonomy, which involves giving effect to parties’ agreed method of dispute resolution: see Tjong Very Sumito at [28] and AnAn at [75]–[80]. Subject to legislative intervention, it is questionable whether the court can or should strike down arbitration agreements on the grounds relied on in Uber. Given the increased reliance on arbitration clauses in e-commerce sites and consumer contracts, the question arises whether retail consumers will be left with no effective remedy in the event of a breach.

181 The issues above are worthy of future consideration in a case which squarely engages them, and the court has the benefit of detailed and directed arguments. They were not engaged in the present case since the Representative Claimants here were both subject to prima facie arbitration agreements, as were the rest of the claimants. In any event, Terraform had taken a “step in the proceedings”.

182 I therefore dismissed Terraform’s appeal in RA 185 against the order denying a stay in favour of arbitration (in SUM 1427) on the basis that it had taken steps in the proceedings.

183 Given that the “case management stays” requested by Kwon and Luna (in SUM 1427) and Platias (in SUM 235) were predicated on Terraform obtaining an arbitration stay, their appeals (in RAs 185 and 186 respectively) against the order denying their applications are dismissed. Accordingly, I dismissed the defendants’ appeals with costs.

184 There was one aspect of these proceedings which was troubling. Although the Suit was instituted on 7 September 2022, and the defendants served on 9 September 2022, ¹⁶⁵ Terraform’s jurisdictional challenge was only heard by the AR on 28 June 2023, more than nine months later. This delay is regrettable. As highlighted above at [51], the ROC 2021 is specifically designed to ensure that jurisdictional objections are filed and disposed of expeditiously and as early as possible, to ensure that time and costs are not wasted and that parties may then proceed to deal with the merits of the claim only if the challenge fails. That is also sensible from a case management standpoint, and is in the interests of all the parties. Indeed, the ROC 2021 was meant “to simplify and expedite applications and appeals on procedural matters so that disputes … do not become procedural skirmishes which waste time and costs and often do not bring the parties any closer to the main battlefield”: Civil Justice Commission Report at p 2. Unfortunately, that philosophy was not embraced in this case, and parties (particularly Terraform) engaged in the “old” ways of procedural and “strategic” manoeuvres unrelated and unnecessary to resolving the jurisdictional challenge.

185 All litigants should comply with the language and spirit of the ROC 2021, as embodied in the Ideals in O 3 r 1, and work with the courts to ensure a fair and expeditious resolution to their disputes (see O 3 rr 1(b) and 1(c)). While the basic structure of civil litigation was not altered by the ROC 2021, it is not old wine in new skins. It was intended for parties to approach the resolution of civil disputes with a new mindset – for cases to be managed closely and effectively so as to facilitate the crystallisation of key issues in dispute at an early stage, which will in turn enable cases to progress more efficiently: see Civil Justice Review Committee, Report of the Civil Justice Review Committee (2018) (Chairperson: Indranee Rajah SC) at pp 6–7. In this regard, I note Lord Bingham MR’s observations in Costellow v Somerset Country Council [1993] 1 All ER 959 at 959 (cited by the Court of Appeal with approval in Leong Mei Chuan v Chan Teck Hock David [2001] 1 SLR(R) 261 at [19] and The Tokai Maru [1998] 2 SLR(R) 646 (“The Tokai Maru”) at [21]), that “the rules of court and the associated rules of practice, devised in the public interest to promote the expeditious dispatch of litigation, must be observed … [they] are not targets to be aimed at or expressions of pious hope but requirements to be met” [emphasis added]. As far as a party’s actions in delaying the judicial process are not intentional or disingenuous, some lenience may be granted: see The Tokai Maru at [22]–[23]. However, where such conduct is deliberate or unreasonable, parties should expect to be met with firm sanctions from the court.