//
© 2026 Zoe. All rights reserved.
Zoe is a legal information platform. Always consult the Federal Register of Legislation for authoritative text.
[2018] HCA 25
appellant. Appeal allowed. Google LLC substituted as respondent. Orders 2 to 6 of the Court of Appeal of the Supreme Court of Victoria made on 20 December 2016 set aside and, in their place, order that the...
Michael Trkulja successfully appealed to the High Court after lower courts tried to throw out his defamation case against Google. When people searched for his name, Google's image results mixed his photo with mugshots of Melbourne gangsters like Carl Williams and Tony Mokbel under headings about the 'criminal underworld'. Autocomplete also suggested searches such as 'michael trkulja criminal'. Trkulja argued this implied he was a serious criminal too. The High Court ruled that an ordinary person could reasonably understand the results that way, so the case was not hopeless and deserved a full trial. The Court criticised the Victorian Court of Appeal for over-complicating the test, assuming users knew too much about how Google works, and deciding factual issues without evidence or a trial. The claim can now proceed.
AI-generated legal information, not legal advice. Zoe can make mistakes. Check the cited source, and for advice about your situation consult a qualified Australian lawyer.
7 cited instruments linked from this judgment.
3,083 words · generated 24/04/2026
Milorad (Michael) Trkulja commenced defamation proceedings in the Supreme Court of Victoria against Google Inc (now Google LLC). Between 1 December 2012 and 3 March 2014, persons in Victoria who entered Mr Trkulja's name or alias into Google's search engine or image search were presented with results that mixed photographs of Mr Trkulja with photographs of notorious Melbourne criminals including Carl Williams, Andrew "Benji" Veniamin, Tony Mokbel and Mario Rocco Condello. The results appeared under headings such as "melbourne criminals", "melbourne criminal underworld figure", "melbourne underworld crime photos" and "melbourne underworld killings". Autocomplete predictions generated by Google included phrases such as "michael trkulja criminal", "michael trkulja melbourne underworld crime" and "milorad trkulja shooting". One web result contained the direct statement "I hear Milorad 'Michael' Trkulja is a former hitman who shot a music promoter in the balaclava" followed by the word "Streisand'd".
On 3 December 2012 Mr Trkulja sent letters of demand by registered post to Google and Google Australia Pty Ltd. The letters identified the defamatory matter, demanded removal of images and links, requested source details and sought permanent blocking of searches for his names. Google Australia responded that the search products belonged to Google. Google itself provided a detailed response on 16 January 2013 stating that it had removed certain websites from web search results, blocked some autocomplete predictions on google.com.au, but declined to remove images of Mr Trkulja appearing in other image searches.
Mr Trkulja's holograph amended statement of claim pleaded that the "Google Images matter" (20 pages) and "Google Web matter" (seven pages) were defamatory in their natural and ordinary meaning and carried six specific imputations: that he was a hardened and serious criminal in Melbourne; in the same league as named underworld murderers; an associate of Veniamin, Williams and Mokbel; and such a significant figure in the Melbourne criminal underworld that events involving him were recorded on a crime-chronicle website. In the alternative, true innuendo and general association with organised crime were relied upon. Damages, aggravated and punitive damages (founded on Google's knowledge of falsity from December 2012), and a permanent injunction to block all future searches for his names were claimed.
Google applied to set aside service out of the jurisdiction and to summarily dismiss the proceeding under s 63 of the Civil Procedure Act 2010 (Vic) on three bases: no publication, no capacity to defame, and common-law immunity. McDonald J refused the application, holding that publication was strongly arguable, that it was certainly arguable a reasonable user would infer Mr Trkulja was a convicted criminal from the image compilations, and that the Defamation Act 2005 (Vic) defences made a new immunity unnecessary. The Court of Appeal (Ashley, Ferguson and McLeish JJA) allowed Google's appeal, concluding the matters were incapable of conveying any pleaded imputation. The High Court granted leave, heard the appeal and, in a joint judgment of Kiefel CJ, Bell, Keane, Nettle and Gordon JJ delivered on 13 June 2018, allowed the appeal, reinstated the proceeding and ordered Google to pay costs.
The High Court decided the matter on a narrow but decisive point: the Court of Appeal had erred in law in its approach to the capacity of the publications to convey the pleaded imputations and had therefore wrongly concluded the proceeding had no real prospect of success. The joint judgment emphasised that capacity to defame is a question of law but one on which reasonable minds may differ; accordingly, a defamation pleading should be struck out or a proceeding summarily dismissed on capacity grounds only with great caution. On a summary application the plaintiff's case must be taken at its highest.
The Court restated the classic test: whether the ordinary reasonable person would understand the matter as conveying the imputations, bearing in mind that such persons possess a range of temperaments, education and life experience. Some are suspicious, some naïve; the exercise is to envisage the mean and to ask what is the most damaging meaning that person could put on the material. The assessment is one of generosity not parsimony. The ordinary reasonable person reads between the lines, draws derogatory implications more freely than a lawyer, and is influenced by conspicuous headings and first impressions.
Critically, the Court held that the Court of Appeal had mischaracterised Mr Trkulja's pleading. Although the amended statement of claim was inelegant, it treated each distinct search and its responsive results as separate publications, consistent with Callinan J's observation in Dow Jones & Co Inc v Gutnick that each hit on a website is a separate publication. Even if the results in Annexure A could be aggregated among themselves (and those in Annexure B among themselves), the Court of Appeal went further by treating all of Annexures A and B as one composite and attributing to the hypothetical user knowledge of every page. That approach was not supported by the pleading and inflated the technical knowledge to be imputed to the ordinary reasonable user.
The Court of Appeal had also relied heavily on untested Google affidavit evidence about how the worldwide web, search algorithms and autocomplete function operate. The High Court ruled that, at the summary stage and without cross-examination or opposing evidence, it was inappropriate to make definitive findings of mixed fact and law drawn from that material. In particular, the Court of Appeal's conclusion that any user would "inevitably" perceive a "disconnect" between search terms and results, or would know results could not possibly be produced manually, assumed a uniformly sophisticated audience that the ordinary reasonable person test does not require. The High Court noted that the level of user comprehension of search-engine mechanics is itself a matter that may require evidence at trial and on which the law has not yet spoken authoritatively.
The Court distinguished Google Inc v Australian Competition and Consumer Commission. In that case the sponsored links were clearly labelled as advertisements from third-party advertisers; ordinary users understood Google was merely passing them on. No such clear demarcation existed here. A user searching for "melbourne underworld criminals" who receives Mr Trkulja's photograph alongside those of notorious killers under that heading could rationally infer that Mr Trkulja shares the same opprobrious characteristic. The presence of unrelated images (a tram, Marlon Brando as the Godfather, lawyers) does not destroy capacity; a jury could still conclude that unknown persons pictured with criminals are being presented as connected to the underworld. That a large and diverse class might be affected goes to the availability of the innocent dissemination defence, not to capacity.
On autocomplete and the "Streisand'd" reference, the Court held that the most damaging meaning must be considered. A direct statement that Mr Trkulja was a "former hitman" is capable of conveying criminality even if accompanied by an allusion to his prior defamation victory. Snippets referring to earlier proceedings awarding him damages likewise do not necessarily neutralise the sting, especially where the snippets themselves repeat the underworld association.
Because the publications were at least capable of conveying one or more of the six imputations, the proceeding had a real prospect of success. The Court found it unnecessary to decide publication definitively but observed that McDonald J had been correct to hold that intentional participation in communicating the results made Google arguably a publisher. Questions of the precise degree of participation and any s 32 innocent dissemination defence could not be resolved summarily before discovery and a pleaded defence. The Court of Appeal's lengthy excursus into the juridical nature of search-engine liability and policy considerations was therefore premature.
Prior to Trkulja v Google LLC the Australian law of defamation had begun to grapple with internet intermediaries but remained unsettled on search engines. Beach J in Trkulja v Google (No 5) had held that Google could be a publisher of search results and autocomplete predictions once it had notice, and that certain image and web results were capable of conveying imputations of criminality. Blue J in Duffy v Google Inc reached similar conclusions after a full trial. Appellate authority was sparse. The Victorian Court of Appeal's decision under appeal had effectively treated those first-instance decisions as plainly wrong and had embarked on a prospective reshaping of defamation doctrine to accommodate the social utility of search engines, suggesting an almost automatic innocent dissemination defence before notification.
The High Court restored orthodoxy. It affirmed that the existing tests for publication and capacity to defame apply to search engines without need for immediate common-law immunity. Publication turns on intentional participation in communication; all degrees of participation constitute publication and the defendant must plead and prove any subordinate-distributor or innocent-dissemination defence. Capacity remains a jury question (or judge-alone equivalent) assessed generously by reference to the ordinary reasonable person. The "no real prospect of success" test under the Civil Procedure Act 2010 (Vic) is not to be equated with the stricter Dey and General Steel standard but still requires a high degree of certainty before a plaintiff is denied a trial.
After the decision, search-engine proprietors can no longer expect early summary dismissal on capacity grounds where results objectively link a plaintiff to criminality. The case confirms that autocomplete predictions and image compilations are not immune from scrutiny. It also signals that factual questions about how search results are generated, the state of user knowledge, and the practical operation of any innocent dissemination defence will ordinarily require discovery, evidence and trial. The decision has reinforced the caution with which intermediate appellate courts must approach summary judgment in novel but fact-sensitive areas of tort law. It leaves open, however, the precise content of the ordinary reasonable search-engine user's characteristics—an issue that later courts have been required to develop on fuller evidentiary records.
The joint judgment contains several passages that encapsulate the Court's reasoning. One central statement appears in the discussion of capacity:
"The question of whether words or images complained of are capable of conveying a pleaded defamatory imputation is a question of law which permits of only one correct answer. It is, however, a question about which reasonable minds may sometimes differ, and, consequently, it is only ever with great caution that a defamation pleading should be disallowed as incapable of bearing a defamatory imputation."
In plain English: deciding whether something can be defamatory is technically a legal question with one right answer, but because judges can reasonably disagree, courts should be very reluctant to kill a defamation case at the outset on the basis that the material is incapable of hurting the plaintiff's reputation.
On the ordinary reasonable person the Court said:
"The test for whether a published matter is capable of being defamatory is what ordinary reasonable people would understand by the matter complained of. In making that assessment, it is necessary to bear in mind that ordinary men and women have different temperaments and outlooks, degrees of education and life experience. … The exercise is, therefore, one of attempting to envisage a mean or midpoint of temperaments and abilities and on that basis to decide the most damaging meaning that ordinary reasonable people at the midpoint could put on the impugned words or images considering the publication as a whole."
Plain-English translation: we judge defamation by asking what an average person on the street would think. Because people vary—some are cynical, some gullible—we imagine a middle-of-the-road person and ask what is the worst (but still reasonable) meaning that person could take from the material when looking at it as a whole.
Rejecting the Court of Appeal's composite-publication approach and its reliance on untested evidence the Court observed:
"there can be no certainty as to the nature and extent of Google's involvement in the compilation and publication of its search engine results until after discovery. There are only the untested assertions of Google deponents."
In plain English: we cannot know exactly how much Google controls or contributes to its search results until the plaintiff has had a chance to see internal documents and cross-examine witnesses; it is therefore wrong to decide the case on Google's say-so alone at an early stage.
On the difference from the Google v ACCC case the Court stated:
"By contrast, this case is not concerned with sponsored links or misleading and deceptive conduct … There is no evidence here … that it would have been apparent to an ordinary reasonable person using the Google search engine that Google made no contribution to the elements or combination of elements of those of the search results that convey a connection between Mr Trkulja and criminality."
Plain-English translation: in the earlier ACCC case everyone knew the ads came from advertisers, not Google. Here the search results look like Google's own response to a query about criminals, so an ordinary user could think Google is implying Mr Trkulja is one of them.
Finally, on the "Streisand'd" reference and prior proceedings the Court held:
"even if the use of 'Streisand'd' could be regarded as suggesting some sort of connection between Mr Trkulja and a defamation proceeding … that would not bar the capacity of the words 'Milorad "Michael" Trkulja is a former hitman who shot a music promoter in the balaclava' to defame Mr Trkulja."
Plain English: just because the page mentions that Trkulja previously sued Google successfully does not cancel out the direct accusation that he is a hitman; the jury could still find the page damaging.
Trkulja v Google LLC is triggered whenever a plaintiff complains that an internet search engine has published results—whether organic web links, image thumbnails or autocomplete predictions—that a reasonable person could read as conveying defamatory imputations, and the defendant seeks to have the claim summarily dismissed on the basis that the results are incapable of being defamatory. The precedent applies with particular force where:
search terms entered by users include the plaintiff's name combined with words suggestive of crime or disrepute ("criminal", "underworld", "hitman");
the responsive results display the plaintiff's photograph or name alongside images or names of persons notorious for criminality under headings that echo the search terms;
autocomplete suggestions complete the plaintiff's name with criminal descriptors;
the plaintiff has given notice and the proprietor has not removed all offending material;
the defendant relies on the technical complexity of search algorithms or user sophistication to argue that no ordinary person could take the results at face value.
The case is not limited to Google; any search-engine operator or autocomplete provider operating in Australia falls within its reasoning. It does not, however, create strict liability. The proprietor may still succeed on innocent dissemination after notification, on qualified privilege, honest opinion or other statutory defences once the facts are fully ventilated at trial. The precedent is confined to the summary-judgment context; a different outcome may eventuate after evidence is heard. Fact patterns in which the results are clearly labelled as third-party advertisements or where the plaintiff is a public figure whose prior criminal convictions are notorious may be distinguished on their facts.
Subsequent decisions have treated Trkulja v Google LLC as authoritative on both the caution required in summary dismissal of defamation claims involving search engines and the application of the ordinary reasonable person test to digital publications. In Google LLC v Defteros the High Court again considered search-engine publication but focused on the innocent dissemination defence after notification; it cited Trkulja with approval for the proposition that publication is at least arguable. Victorian and New South Wales courts have applied the generous capacity test to social-media thumbnails, suggested searches and algorithmic compilations, refusing summary judgment where a jury could infer an adverse meaning.
Intermediate appellate courts have been careful not to attribute unrealistic technical knowledge to users without evidence. Trial judges have admitted expert evidence on search-engine operation and user behaviour, precisely the course the High Court contemplated. In cases involving autocomplete, courts have distinguished Duffy v Google Inc on the basis that Blue J's findings were made after a full trial, whereas Trkulja emphasises that capacity must be assessed at the pleading stage on the basis of the material itself. No court has yet overruled or relevantly distinguished the core propositions about generosity, the inappropriateness of composite-publication reasoning at the summary stage, or the need for discovery before determining the extent of a search engine's participation. Later authority has therefore reinforced rather than narrowed the decision, confirming that search-engine proprietors will ordinarily have to defend defamation claims on their merits once notice has been given and the results are objectively capable of conveying an adverse imputation.
Several important questions remain unresolved after Trkulja v Google LLC. First, what precise degree of knowledge about search-engine mechanics should be attributed to the ordinary reasonable person? The joint judgment left open whether there should be a single midpoint or whether different classes of user (novice versus habitual) require separate consideration, noting that evidence on this topic may be necessary. Second, the exact parameters of the innocent dissemination defence for search engines after notification remain to be fully worked through. While the Court indicated that the defence will often be available before notice, the position after a detailed letter of demand—particularly where automated systems continue to generate new results—requires further litigation.
Third, the interaction between capacity to defame and the statutory defences under the Defamation Act 2005 (Vic) (and its uniform counterparts) has not been exhaustively mapped. In particular, whether a search engine can rely on the s 32 subordinate distributor defence or the new public-interest defence introduced after 2021 in some jurisdictions is yet to be tested on a full factual record. Fourth, the appropriate form of injunctive relief against a global search provider—permanent blocking of all results for a name versus more targeted de-indexing—remains unsettled, especially given the risk of over-blocking legitimate content.
Finally, the decision leaves open how capacity should be assessed when results are generated by artificial-intelligence models rather than traditional keyword algorithms. As generative AI produces narrative summaries rather than mere links or thumbnails, the "most damaging meaning" a reasonable user might draw could become even more fact-sensitive. Until appellate courts have the benefit of evidence on these new technologies, Trkulja stands as a clear statement that plaintiffs must be given their day in court where the material is objectively capable of harming reputation, but many of the practical and doctrinal consequences for digital intermediaries are still to be worked out at trial level.
Trkulja v Google LLC [2018] HCA 25 13 June 2018 M88/2017
G O'L Reynolds SC with P A Heywood-Smith QC and D P Hume for the appellant (instructed by George Liberogiannis & Associates)
N J Young QC with L G De Ferrari SC for the respondent (instructed by Ashurst Australia)
Notice: This copy of the Court's Reasons for Judgment is subject to formal revision prior to publication in the Commonwealth Law Reports.
Defamation - Publication - Capacity to defame - Where application for summary dismissal of defamation proceeding - Where allegedly defamatory matter includes search results of internet search engine - Where allegedly defamatory matter includes autocomplete predictions of internet search engine - Whether respondent published allegedly defamatory matter - Whether matter capable of conveying allegedly defamatory imputations - Whether proceeding had no real prospect of success.
Words and phrases - "autocomplete", "capacity to defame", "composite publication", "defamation", "defamatory image", "no real prospect of success", "ordinary reasonable person", "publication", "search engine", "search results", "search terms", "summary dismissal", "summary judgment".
Civil Procedure Act 2010 (Vic), ss 62, 63. Supreme Court (General Civil Procedure) Rules 2005 (Vic), rr 7.01, 8.09.
KIEFEL CJ, BELL, KEANE, NETTLE AND GORDON JJ. This is an appeal from a judgment of the Court of Appeal of the Supreme Court of Victoria (Ashley, Ferguson and McLeish JJA), on appeal from an order of the primary judge (McDonald J). McDonald J ordered that an application by the respondent, Google Inc (now Google LLC ("Google")), to set aside a defamation proceeding brought by the appellant, Mr Trkulja, against Google, and its service out of the jurisdiction on Google, be dismissed. McDonald J rejected Google's contention that the proceeding has no real prospect of success. In allowing the appeal, the Court of Appeal held, to the contrary, that the proceeding has no real prospect of success. For the reasons which follow, McDonald J was correct to refuse to set aside the proceeding and, therefore, the appeal to this Court should be allowed. Mr Trkulja's claim Mr Trkulja's holograph amended statement of claim ("the Amended Statement of Claim") is not an elegant pleading. It is, however, sufficiently comprehensible to convey that Mr Trkulja alleges that Google defamed him by publishing images which convey imputations that he "is a hardened and serious criminal in Melbourne", in the same league as figures such as "convicted murderer" Carl Williams, "underworld killer" Andrew "Benji" Veniamin, "notorious murderer" Tony Mokbel and "Mafia Boss" Mario Rocco Condello; an associate of Veniamin, Williams and Mokbel; and "such a significant figure in the Melbourne criminal underworld that events involving him are recorded on a website that chronicles crime in [the] Melbourne criminal underworld". The pleading alleges that Google published the defamatory images between 1 December 2012 and 3 March 2014 to persons in Victoria, including several named persons, upon those persons accessing the Google website, searching for Mr Trkulja's name or alias (Michael Trkulja and Milorad Trkulja), and then viewing and perceiving the images presented on-screen in response to the search. The pleading particularises the allegedly defamatory matters as comprising two groups: "the Google Images matter" and "the Google Web matter" (reproductions of which are set out in the Amended Statement of Claim and in Annexures A and B, respectively, to the judgments of the primary judge and the Court of Appeal). The Google Images matter ("the images matter") consists of 20 pages which are individually described in the pleading. Pages one to 13 and 15 to 20 are described as Google images search results pages that display images of Mr Trkulja mixed with images of convicted Melbourne criminals. Those pages variously contain one of the following phrases: "melbourne criminals", "melbourne criminal underworld figure", "melbourne criminal underworld photos", "melbourne underworld crime", "melbourne underworld crime photos", "melbourne underworld criminals", "melbourne underworld killings" and "melbourne underworld photos". The pleading draws attention to a particular feature of the images matter, which is that some of the pages include an image that contains text stating, inter alia, "Google lawsuit in court", "COLOURFUL Melbourne identity Michael Trkulja" and "Mr Trkulja an associate of Mick Gatto". Page 14 of the images matter is described in the pleading as a Google "autocomplete" search results page. It shows a Google search for "michael trk" together with autocomplete predictions, namely, phrases including "michael trkulja", "michael trkulja criminal", "michael trkulja melbourne crime", "michael trkulja underworld" and "michael trkulja melbourne underworld crime". In addition, although it is not described as such in the pleading, the page contains an image referring to a "[w]ebsite for this image", stating that "[i]n a nutshell, Michael Trkulja's beef with both Yahoo and Google was that …" and other references to a defamation lawyer and an online solicitor. The Google Web matter ("the web matter") consists of seven individual pages. Page one is not described in the pleading but it shows what appears to be an online post by "Picklesworth" that says: "I hear Milorad 'Michael' Trkulja is a former hitman who shot a music promoter in the balaclava. 'Streisand'd". Underneath that statement is an image of what appears to be predictions generated by Google's autocomplete functionality showing the phrases "michael trkulja", "michael trkulja criminal", "michael trkulja melbourne crime" and "michael trkulja underworld". Page two of the web matter is not precisely described in the pleading but appears to be a web search results page for the search words "melbourne-criminal-underworld-figure", and which displays both text results and image results. Pages three and four of the web matter are described in the pleading as web search results pages for the search words "melbourne criminal underworld photos" and "melbourne underworld criminals", and which display both text results and image results. The pleading draws attention to the fact that pages three and four display images of Mr Trkulja mixed with images of convicted Melbourne criminals. Pages five to seven of the web matter are described in the pleading as Google autocomplete search results pages. The substantive content of page five of the web matter resembles that of page 14 of the images matter, albeit page five does not have the additional images that are displayed in the latter. On page six, a Google search for the words "michael trkulj" is displayed together with autocomplete predictions, namely, the phrases "michael trkulja", "michael trkulja v google", "michael trkulja shot", "michael trkulja lawyer", "michael trkulja tony mokbel", "michael trkulja melbourne underworld crime" and "michael trkulja google". Similarly, on page seven, a Google search for the words "milorad trkulj" is displayed in combination with autocomplete predictions, namely, the phrases "milorad trkulja", "milorad trkulja criminal", "milorad trkulja shooting", "milorad trkulja google", "milorad trkulja lawyer", "milorad trkulja email", "milorad trkulja tony mokbel", "milorad trkulja wiki", "milorad trkulja yahoo" and "milorad trkulja melbourne". The pleading avers that the images matter and the web matter are defamatory of Mr Trkulja in their natural and ordinary meaning and, further, that they carry the following defamatory imputations: "(a) The plaintiff is a hardened and serious criminal in Melbourne[;] (b) The plaintiff is a hardened and serious criminal in Melbourne in the same league as convicted murderer Carl Williams, hardened notorious underworld killer Andrew 'Benji' Veniamin, hardened and serious and notorious murderer Tony Mokbel and the Mafia Boss Mario Rocco Condello[;] (c) The plaintiff is an associate of underworld killer Andrew 'Benji' Veniamin[;] [(d)] The plaintiff is an associate of Carl Williams Melbourne notorious convicted criminal murderer and drug trafficker; (e) The plaintiff is an associate of Tony Mokbel, the Australian notorious convicted murderer and drug supplier and trafficker; (f) The plaintiff is such a significant figure in the Melbourne criminal underworld that events involving him are recorded on a website that chronicles crime in [the] Melbourne criminal underworld[.]" In the alternative it is contended that the images matter is defamatory in its true innuendo for carrying substantially the same imputations, and also that the gist of the images matter and the web matter is to associate Mr Trkulja with organised criminal activity in Melbourne. The pleading then alleges that on or about 3 December 2012 Mr Trkulja sent a letter each to Google and Google Australia Pty Ltd ("Google Australia") (which at one time was the second defendant to the proceedings) by registered post drawing the allegedly defamatory matter to their attention, informing them of the nature of the defamatory matter, demanding that Google and Google Australia remove the images matter from their computers and servers, or to remove all links or direction from their computers and servers linking or directing internet users to the matter, requesting them to provide details including contact details of the source or sources of the matter, and demanding that they "block the name of Milorad Trkulja and Michael Trkulja from [their] computers and servers links or directing internet users to the name of 'Milorad Trkulja' and 'Michael Trkulja'". On 14 December 2012, Google Australia responded to the effect that the "search products" to which Mr Trkulja's "inquiry" related were owned by Google and that Google Australia was "unable to further assist" him with his inquiry. Then, on 18 December 2012, Google sent an email to Mr Trkulja to the effect that Google Australia had forwarded Mr Trkulja's letter to Google; that the "Google services" referred to in the letter were owned and operated by Google, to which all future correspondence relating thereto should be directed; and that Google was currently reviewing the complaint and would contact Mr Trkulja when it had completed its review. Mr Trkulja replied on the following day via email and on 20 December 2012 received a reply in the same terms as that sent by Google on 18 December 2012. Google provided a detailed response to Mr Trkulja on 16 January 2013. In substance, Google stated that it had removed certain websites from its web search results pages and, without admission, that it had blocked certain autocomplete predictions and search queries relating to Mr Trkulja from appearing as part of the autocomplete and search functions of "google.com.au". Google declined, however, to remove the images of Mr Trkulja which appeared in response to other image searches made using the Google search engine. The prayer for relief is for damages, including aggravated and punitive damages on the basis of Google's knowledge of the falsity of the imputations, at least from 3 December 2012, and its refusal to accept any responsibility for the allegedly defamatory publications, and also for an injunction against Google in the following terms: "that [Google] permanently block Google Images and web searchers [sic] of the Plaintiff's names 'Milorad Trkulja' and 'Michael Trkulja' from its computers and servers and remove all links from its computers and servers linking to the Google webs and images users from Australia." Relevant statutory provisions At relevant times and so far as is germane for present purposes, r 7.01 of the Supreme Court (General Civil Procedure) Rules 2005 (Vic) provided for service of originating process out of Australia, without order of the court, where the proceeding is founded on a tort committed within Victoria (r 7.01(1)(i)) or the proceeding is brought in respect of damage suffered wholly or partly in Victoria and caused by a tortious act or omission wherever occurring (r 7.01(1)(j)). The writ in this proceeding was served out of Australia on Google in the United States of America pursuant to r 7.01(1)(i) and (j). At relevant times, r 8.09 provided in substance that a defendant could apply before entering an appearance, whether conditional or unconditional, to set aside a writ or its service. Section 63 of the Civil Procedure Act 2010 (Vic) provides in substance that a court may give summary judgment in favour of a defendant on the defendant's application, if satisfied that the plaintiff's claim or part of that claim "has no real prospect of success". In Agar v Hyde, this Court essayed the test for determination of an application to set aside service of a proceeding out of Australia, pursuant to Pt 10 r 6A of the Supreme Court Rules 1970 (NSW), on the ground that the claims made in the proceeding had insufficient prospects of success to warrant putting an overseas defendant to the time, expense and trouble of defending them. The plurality concluded that the test should be the same as the test for summary judgment propounded in Dey v Victorian Railways Commissioners and General Steel Industries Inc v Commissioner for Railways (NSW): a party should not be denied the opportunity of placing his or her case before the court in the ordinary way, with the advantage of the usual interlocutory processes, unless there is a high degree of certainty about what would be the ultimate outcome of the proceeding if allowed to go to trial in the ordinary way. Subsequently, in Spencer v The Commonwealth, this Court considered whether the test for summary judgment prescribed by s 31A of the Federal Court of Australia Act 1976 (Cth), namely, that the court is satisfied that the other party has "no reasonable prospect of successfully prosecuting the proceeding or … part of [it]", differs from the test espoused in Dey and General Steel. All members of the Court except Heydon J emphasised that the power to dismiss an action summarily should not be exercised lightly but Hayne, Crennan, Kiefel and Bell JJ added that the evident legislative purpose revealed by the text of s 31A would be defeated if its application were read as confined to cases of a kind falling within the test in Dey and General Steel. In Victoria, the test for summary judgment is prescribed by s 62 of the Civil Procedure Act: whether the plaintiff's claim has "no real prospect of success". Consistently with Spencer, the view taken in Victoria is that the power to dismiss an action summarily is not lightly to be exercised but that, like the test applicable to s 31A of the Federal Court of Australia Act, the "no real prospect of success" test is to some degree more liberal than Dey and General Steel. It permits of the possibility of cases in which, although the plaintiff's case is not "hopeless" or "bound to fail", it does not have a real prospect of succeeding. The proceeding at first instance Before McDonald J, Google put its application for summary dismissal on three bases: (i) that it did not publish the images matter or the web matter; (ii) that the matters in issue were not defamatory of Mr Trkulja; and (iii) that Google was entitled to immunity from suit. Based on a careful consideration of the present state of authority, including the decisions of Beach J in Trkulja v Google (No 5) (against which there was no appeal), and of Blue J in Duffy v Google Inc, McDonald J concluded that it was strongly arguable that Google's intentional participation in the communication of the allegedly defamatory search results relating to Mr Trkulja to users of the Google search engine supported a finding that Google published the allegedly defamatory results. McDonald J also rejected Google's contention that a Google search engine user or a person looking over his or her shoulder would not think less of a person such as Mr Trkulja because his photograph is included in the search results or because his photograph or references to his name appear in "snippets" and hyperlinks returned by web searches and autocomplete predictions. His Honour illustrated the point by reference to a compilation of images of Mr Trkulja among images of convicted criminals Judith Moran, Matthew Johnson and Tony Mokbel, which appeared at page four of the web matter as reproduced in Annexure B, and concluded that it was certainly arguable that a reasonable search engine user would look at the compilation and assume that Mr Trkulja was a convicted criminal. McDonald J further rejected Google's contention that Google should be held immune from suit as a matter of public interest, observing, correctly, that the range and extent of the defences provided for in Div 2 of Pt 4 of the Defamation Act 2005 (Vic) militate heavily against the development of a common law search engine proprietor immunity. The proceeding before the Court of Appeal Before the Court of Appeal, Google advanced essentially the same three grounds. The Court of Appeal found it unnecessary to decide the first ground and rejected the third. But the Court of Appeal upheld the second ground, ruling in relation to the images matter that Mr Trkulja "would have no prospect at all of establishing that the images matter conveyed any of the defamatory imputations relied upon", and, in relation to the web matter, that Mr Trkulja "could not possibly succeed in showing that the web matter upon which he relies carried any of the pleaded defamatory imputations". For the reasons which follow, the Court of Appeal were wrong so to hold. Assessing capacity to defame The question of whether words or images complained of are capable of conveying a pleaded defamatory imputation is a question of law which permits of only one correct answer. It is, however, a question about which reasonable minds may sometimes differ, and, consequently, it is only ever with great caution that a defamation pleading should be disallowed as incapable of bearing a defamatory imputation. The potential for difference about the capacity of matters to convey different meanings is an equally strong reason for declining to set aside a proceeding on the basis that an impugned publication is incapable of bearing the defamatory imputation alleged. And it is to be remembered that on an application for summary dismissal such as this, the plaintiff's case as to the capacity of the publications to defame is to be taken at its highest. The test for whether a published matter is capable of being defamatory is what ordinary reasonable people would understand by the matter complained of. In making that assessment, it is necessary to bear in mind that ordinary men and women have different temperaments and outlooks, degrees of education and life experience. As Lord Reid observed in Lewis v Daily Telegraph Ltd, "[s]ome are unusually suspicious and some are unusually naive". So also are some unusually well educated and sophisticated while others are deprived of the benefits of those advantages. The exercise is, therefore, one of attempting to envisage a mean or midpoint of temperaments and abilities and on that basis to decide the most damaging meaning that ordinary reasonable people at the midpoint could put on the impugned words or images considering the publication as a whole. As the Court of Appeal of England and Wales observed in Berezovsky v Forbes Inc, that exercise is one in generosity not parsimony. The question is not what the allegedly defamatory words or images in fact say or depict but what a jury could reasonably think they convey to the ordinary reasonable person; and it is often a matter of first impression. The ordinary reasonable person is not a lawyer who examines the impugned publication over-zealously but someone who views the publication casually and is prone to a degree of loose thinking. He or she may be taken to "read between the lines in the light of his general knowledge and experience of worldly affairs", but such a person also draws implications much more freely than a lawyer, especially derogatory implications, and takes into account emphasis given by conspicuous headlines or captions. Hence, as Kirby J observed in Chakravarti v Advertiser Newspapers Ltd, "[w]here words have been used which are imprecise, ambiguous or loose, a very wide latitude will be ascribed to the ordinary person to draw imputations adverse to the subject". The Court of Appeal approached the matter on the basis that Mr Trkulja's claim is a composite claim wherein all of the search results comprised in the images matter (Annexure A) are to be looked at as one single composite publication, all of the search results comprised in the web matter (Annexure B) are to be looked at as another single composite publication, and, in determining whether any of the searches comprised in Annexure A is capable of conveying the allegedly defamatory imputations, the ordinary reasonable search engine user is to be attributed with knowledge of the contents of all of the searches comprising Annexure A and Annexure B, and vice versa. As appears from the Amended Statement of Claim, that is not the way in which the case is pleaded. The Amended Statement of Claim conveys that each search and the result which appeared in response to it are to be considered together but separately from each other separate search and response, for the reason that each search may have been conducted by a different person without engaging in any of the other searches. That accords with the view expressed by Callinan J in Dow Jones & Co Inc v Gutnick that each hit on a website is a separate publication. Before this Court, counsel for Mr Trkulja did not seek to make anything of the point. He appeared to accept that it was open to aggregate all of the search results in Annexure A and all of the search results in Annexure B, although not A and B, but, if the matter goes to trial, the difference could prove significant. Be that as it may, it is evident for the reasons given by McDonald J that at least some of the search results complained of had the capacity to convey to an ordinary reasonable person viewing the search results that Mr Trkulja was somehow opprobriously associated with the Melbourne criminal underworld, and, therefore, that the search results had the capacity to convey one or more of the defamatory imputations alleged. Whether or not the search results are viewed individually or as a composite does not affect that conclusion. As will be explained, the Court of Appeal's reasoning to the contrary must be rejected. The Court of Appeal's reasoning The Court of Appeal's judgment is of extraordinary length and complexity for the resolution of an appeal against dismissal of a summary disposition application in which the only real question was the capacity of the published matters to defame. It ranges across a broad tract of the law of defamation extending to a substantial, proleptic analysis of the juridical basis of primary and secondary publication in relation to computer search engine proprietors, of the application of innocent publication defences to computer search engine proprietors, and of how and why, in view of the social utility of computer search engines, the existing law of defamation might better be shaped to relate to search engine proprietors or relieve them from liability. Problematically, it also effectively treats the judgment of Beach J in Trkulja v Google (No 5) as if it were plainly wrong (despite the fact that Google did not appeal against that judgment and that it has been considered with implicit approval in another common law jurisdiction), and, in relation to the question of capacity of the autocomplete predictions to defame, treats the observations of Blue J in Duffy as if they went to capacity to defame, notwithstanding that Blue J was describing the process of reasoning by which his Honour, sitting as trial judge, reached findings of mixed fact and law in the trial of a defamation proceeding before judge alone. The Court of Appeal's judgment is also replete with direct and indirect references to Google's affidavit evidence regarding the "world wide web", search engines, and the systems and processes by which Google claims that its computer search engine results are generated; and, despite the summary nature of the application and, therefore, the impracticability of affording Mr Trkulja access to an opportunity for meaningful cross-examination of Google deponents, ordinary interlocutory processes and tendering opposing evidence, the judgment includes a range of purportedly definitive findings of mixed fact and law drawn from Google's affidavit evidence adverse to Mr Trkulja. (i) Publication McDonald J was correct to hold that it is strongly arguable that Google's intentional participation in the communication of the allegedly defamatory results to Google search engine users supports a finding that Google published the allegedly defamatory results. Properly advised, that was all that the Court of Appeal needed to say on the subject. Instead, although the Court of Appeal did not decide the appeal on the question of publication, their Honours made a purportedly determinative finding of mixed fact and law that a search engine proprietor, like Google, is a publisher of search results, including of autocomplete predictions, but that an innocent dissemination defence will almost always, if not always, be maintainable in a period before notification of an alleged defamation. That was not an appropriate way to proceed. In point of principle, the law as to publication is tolerably clear. It is the application of it to the particular facts of the case which tends to be difficult, especially in the relatively novel context of internet search engine results. And contrary to the Court of Appeal's approach, there can be no certainty as to the nature and extent of Google's involvement in the compilation and publication of its search engine results until after discovery. There are only the untested assertions of Google deponents. Furthermore, until and unless Google files a defence it cannot be known what defences will be taken (whatever Google might now say is its intention regarding the defences on which it will rely). Nor does it profit to conjecture what defences might be taken and whether, if taken, they would be likely to succeed. For whatever defences are taken, they will involve questions of mixed fact and law and, to the extent that they involve questions of fact, they will be matters for the jury. Given the nature of this proceeding, there should have been no thought of summary determination of issues relating to publication or possible defences, at least until after discovery, and possibly at all. The Court of Appeal were also incorrect to say that it was incumbent on Mr Trkulja to plead that Google is a primary or secondary publisher of the allegedly defamatory matters. It is not the practice to plead the degree of participation in the publication of defamatory matters, for the reason that all degrees of participation in the publication are publication. As Isaacs J held in Webb v Bloch: "The term published is the proper and technical term to be used in the case of libel, without reference to the precise degree in which the defendant has been instrumental to such publication; since, if he has intentionally lent his assistance to its existence for the purpose of being published, his instrumentality is evidence to show a publication by him." If Google wishes to invoke the defence afforded to "subordinate distributors" by s 32 of the Defamation Act or otherwise contend that the degree of its participation in the publication of the impugned search results was such that it should not be held liable, it is for Google to plead and prove the relevant facts. (ii) Capacity to defame The Court of Appeal's process of reasoning as to the capacity of the impugned web searches to defame Mr Trkulja proceeded from dual premises that whether "any of the defamatory imputations which are pleaded [are] arguably conveyed" is to be determined by reference to the understanding of "an ordinary reasonable user of a search engine such as the Google search engine, without which the facility to navigate the trillions of pages on the world wide web would be gravely compromised", and having regard to the "entirety of the matter relied upon". Their Honours next identified the impugned matter as being, in the case of the images matter, "the composite of the search terms and the images compiled in response" and, in the case of the web matter, "the composite of the search terms and the results which the Google search engine produced", with the result that all of the search results comprised in the images matter (reproduced in Annexure A) were to be looked at as one single composite publication and all of the search results comprised in the web matter (reproduced in Annexure B) were to be looked at as another single publication, upon the basis that the ordinary reasonable search engine user was to be attributed with knowledge of the contents of all of the search results comprising Annexures A and B. The third step in the Court of Appeal's process of reasoning was that the allegedly defamatory material was to be viewed in a context which comprised the world wide web; the particular search engine website; the ability of any internet user to access that website using a web browser to input search terms; and the form of the search engine's response to the terms inputted by the user, because, according to the Court of Appeal, "there would scarcely be an internet user in Australia (or in the 189 countries where the Google search engine is used) who would not recognise that context". The fourth step was to reason that, because of the extreme speed with which search engine results are generated, and the number of search results produced, any user of a search engine would know of the enormous scale of the search which has been made and that it could not possibly be made manually. Thus, according to the Court of Appeal, any user who inputted the words "melbourne criminal underworld photos" and received in response a compilation of images such as the allegedly defamatory search results appearing on page four of Annexure A, which included some images of known criminals, some images of the late Marlon Brando (in his role as "the Godfather"), a tram, actors (presumably in a film about serious crime) and a solicitor, or, on another occasion, results including images such as those appearing on page one of Annexure A of a former Victorian Chief Commissioner of Police, a murder victim, a crime reporter and a Google logo, "would inevitably give thought to just what relationship there could possibly be between the words inputted and the compilation produced, and very probably perceive a disconnect between the images and the search terms" and would "recognise ... that the search results in their entirety did not reflect the meaning of the inputted words considered as a phrase". The Court of Appeal also relied, in part, on observations of this Court in Google Inc v Australian Competition and Consumer Commission ("Google v ACCC") as support for the Court of Appeal's findings as to the knowledge to be attributed to the ordinary reasonable user of a search engine. From that, it was said to follow that, considered by reference to the understanding of an ordinary reasonable user of a search engine, and, in particular, the Google search engine: "the plaintiff would have no prospect at all of establishing that the images matter conveyed any of the defamatory imputations relied upon. … It might be said, if a contrary conclusion was to be reached, that the list of persons potentially defamed would be both large and diverse. We do not accept that such a conclusion would be sound." Likewise, in relation to page 14 of Annexure A, which contained an image of autocomplete predictions, an image referring to a "[w]ebsite for this image" and stating that "[i]n a nutshell, Michael Trkulja's beef with both Yahoo and Google was that …", and advertisements for a defamation lawyer and an online solicitor, the Court of Appeal held that the content was incapable of being defamatory. Their Honours found on the basis of Google's affidavit evidence that autocomplete predictions which are returned in respect of particular search terms entered by search engine users are strongly influenced by the particular user's previous searches. In the Court of Appeal's view, it was also "crystal clear" that the image of the webpage related to Mr Trkulja's earlier successful defamation proceedings. Viewing page 14 in the context of the whole of Annexure A, it could not be considered capable of carrying any of the pleaded imputations. As to the web matter, the Court of Appeal referred first to pages five to seven of Annexure B, which also contained autocomplete predictions, and relied on its earlier conclusion that the autocomplete predictions were incapable of being defamatory. Their Honours concluded that page one of Annexure B was essentially of the same character as pages five to seven of Annexure B and, further, that the apparent reference to Mr Trkulja being "Streisand'd" indicated that: "The whole point of this page is that the plaintiff's successful defamation proceeding had produced the Streisand effect. Far from carrying any of the defamatory imputations pleaded by the plaintiff, the commentator was pointing out that the plaintiff's successful defamation proceedings - in which he had been awarded damages in respect of an imputation that he was somehow connected with the Melbourne underworld - had not brought matters to an end." The Court of Appeal thereafter referred to page two of Annexure B, in which, according to the Court of Appeal, the only reference to Mr Trkulja was in connection with his earlier successful defamation proceeding against Google, under the heading "Google defamation case and 'publishing' in the digital age - Crikey". That led their Honours to observe: "How that could possibly be said to be defamatory of the plaintiff we do not understand. The fact that the reference to the plaintiff's earlier successful defamation proceeding was on a results page which adverted to the television series, 'Underbelly', which contained thumbnails of persons associated with the Melbourne underworld (none of which were the plaintiff) and which referred to a reputed criminal named Arico, could not possibly deflect attention from the import of the only reference to the plaintiff." Finally, the Court of Appeal referred to page four of Annexure B, which, as mentioned, was the page in respect of which McDonald J drew the conclusion that it was certainly arguable that a reasonable search engine user would look at the compilation of images and assume that Mr Trkulja was a convicted criminal. The Court of Appeal took the entirely opposite view: "It may be regarded as the high water mark of the material relied upon by the plaintiff, because of the fact that the return of images included the plaintiff and three criminals. Pausing, and underlining the random nature of the images displayed, the four images in the particular sequence are the first four images from the left on the top line of page five of Annexure A … The heading under which the thumbnails on page four appear is 'Images for Melbourne underworld criminals - report images'. It is a similar heading to that which appears above the compilations of images on pages two and three of Annexure B. A reasonable user of the internet, aware of the unpredictable results which are generated by an image search - well exemplified by the 20 pages of Annexure A - would immediately apprehend, in our opinion, that the thumbnails on page four of Annexure B were of no different character. They could not convey the defamatory imputations pleaded by the plaintiff. But there is a further matter. … It is scarcely conceivable that assumed secondary publication prior to [notice being given to Google on or about 3 December 2012] would not attract a successful innocent dissemination defence. But according to the plaintiff's particularised case, it is impossible to say whether page four of Annexure B (and the same is the situation with page five of Annexure A) was published after the giving of notice." Those conclusions are unacceptable. As has been observed, the test of capacity of a published matter to defame is, in this case, whether any of the search results complained of are capable of conveying any of the defamatory imputations alleged. It is not, as the Court of Appeal stated, whether "any of the defamatory imputations which are pleaded [are] arguably conveyed". To express the test as the Court of Appeal did runs the risk (which appears to have eventuated) of judging the issue according to what the court may think the allegedly defamatory words or images say or depict rather than what a jury could reasonably think they convey. Further, although it might be correct to say that the capacity of the search results to convey the alleged defamatory imputations is to be judged by reference to the "ordinary reasonable user of a search engine such as the Google search engine", by analogy, say, to the way it is said that the capacity of a newspaper article to defame is to be judged by reference to the standards of an ordinary reasonable reader, to do so would be correct only so long as the expression were understood to mean an ordinary reasonable person who has made the Google search in issue. No doubt, as the Court of Appeal said, it can be assumed that the ordinary reasonable person who has used the Google search engine to make a search contemplates that the results of his or her search bear some connection to the search terms. But in the absence of tested, accepted evidence to the contrary, it must also be allowed that the ability to navigate the Google search engine, and the extent of comprehension of how and what it produces, whence it derives, and how and to what degree Google contributes to its content, may vary significantly among the range of persons taken to be representative of the hypothetical ordinary reasonable person. Additionally, the question of law of whether the standard of knowledge and comprehension of the processes involved should be taken as some hypothetical midpoint in the range of understanding is yet to be authoritatively determined. It may well be that the answer will turn on evidence as to the standards of knowledge and comprehension among users of the Google search engine (be they first-time or experienced participants, and recognising that the two classes may require separate consideration for the purposes of the law of defamation), and on inferences to be drawn from that kind of evidence as to the implications, particularly derogatory implications, that a user with that degree of knowledge and comprehension would likely attribute to the results of a Google search of the kinds in issue. As Kirby P (as his Honour then was) observed in another context, in Wickstead v Browne, appellate courts should be loath to consider the application of the law to evidence in novel contexts without the benefit of the evidence having been adduced and a trial concluded. Testimony "gives colour and content to the application and development of legal principle", and out of the detail of the evidence ultimately proved may arise an insight which aids understanding whether and how principle should be developed. The Court of Appeal were further in error in treating the decision of this Court in Google v ACCC as supportive of the conclusion that, although an image of Mr Trkulja may have appeared in responses to Google searches which included the words "criminal", "melbourne" and "underworld", that was simply because those terms appeared in a webpage which contained that image, and for that reason were not capable of conveying to the ordinary reasonable user of a search engine the imputation that Mr Trkulja was a criminal or part of the Melbourne criminal underworld. The question in Google v ACCC was whether Google had engaged in misleading and deceptive conduct contrary to s 52 of the Trade Practices Act 1974 (Cth) by displaying misleading and deceptive "sponsored links". At first instance, it was held that Google had not done so because it was simply a conduit which passed on the sponsored links without any adoption or approval of their contents. On appeal, the Full Court of the Federal Court of Australia held that Google had engaged in misleading and deceptive conduct by displaying sponsored links because the sponsored links were "Google's response to a user's insertion of a search term into Google's search engine", which meant that Google did not merely pass on the contents of the sponsored links without adoption or approval in the sense essayed in Yorke v Lucas. On appeal to this Court that holding was reversed. It was considered to be axiomatic that the Google search mechanism operates according to search terms chosen by the user for the purpose of generating "organic search results". According to the primary judge's findings of fact, which were not impugned, Google was not the maker, author, creator or originator of the information in any of the sponsored links. Given evidence adduced at first instance, it was held that the primary judge was right to find, as he had, that ordinary and reasonable users of the Google search engine would have understood that the sponsored links were advertisements which Google did not endorse but was merely passing on for what they were worth. The reason for that being so was that: "[o]n its face, each sponsored link indicates that its source is not Google, but an advertiser. The heading 'Sponsored Links' appears above both top left sponsored links and right side sponsored links, and the URL of the advertiser, appearing within each sponsored link, clearly indicates its source. Ordinary and reasonable users of the Google search engine would have understood that the sponsored links were created by advertisers. Such users would also have understood that representations made by the sponsored links were those of the advertisers, and were not adopted or endorsed by Google." By contrast, this case is not concerned with sponsored links or misleading and deceptive conduct in relation to the content of sponsored links, but rather with the law of defamation in relation to responses to Google searches of another kind. There is no evidence here, on the basis of which it is possible to be persuaded to the level of satisfaction necessary for the summary disposition of the proceeding on the ground of lack of capacity to defame, that it would have been apparent to an ordinary reasonable person using the Google search engine that Google made no contribution to the elements or combination of elements of those of the search results that convey a connection between Mr Trkulja and criminality. And in contradistinction to the "sponsored links" in Google v ACCC, there is not here the indication axiomatically implicit in a third party advertisement that the author of the advertisement is the advertiser. Just as importantly, to say that a user of the Google search engine would "inevitably give thought to just what relationship there could possibly be between the words inputted and the compilation produced" or "very probably perceive a disconnect between the images and the search terms" (emphasis added) does not gainsay that it would be open to a jury to conclude on the balance of probabilities that an ordinary reasonable person using the Google search engine would perceive the compilation to convey one or more of the defamatory imputations alleged. To the contrary, it is to be assumed that such a person would contemplate that there is a connection between the terms of the search inputted into the search engine and the contents of the results displayed. Ex hypothesi, since he or she has conducted the search for criminals and members of the Melbourne criminal underworld, he or she would rationally suppose that there is something in the response which correlates to criminals and members of the Melbourne criminal underworld. And prima facie the most obvious, logical connection between the terms of the search and the response is that those persons whose images or names appear in the response, under headings such as "melbourne criminal underworld photos", "melbourne underworld crime" and "melbourne underworld killings", or at least some of them, are criminals or members of the Melbourne criminal underworld. It is true, as the Court of Appeal observed, that in some of the search results comprising Annexures A and B, some of the persons shown are plainly not criminals or members of the Melbourne criminal underworld. A former Victorian Chief Commissioner of Police, the late Marlon Brando in his role as "the Godfather", criminal law barristers and solicitors and other persons who are not criminals but whose professions have something to do with crime are obvious examples. But in each of the pages on which images of such persons appear, there are also images of persons who are notorious criminals or members of the Melbourne criminal underworld coupled with images of persons, such as Mr Trkulja, whose identity is relatively unknown. Depending upon the totality of the evidence adduced at trial, it would be open to a jury to conclude that an ordinary reasonable person using the Google search engine would infer that the persons pictured whose identities are unknown are persons, like the notorious criminals with whom they are pictured, in some fashion opprobriously connected with criminality and the Melbourne criminal underworld. So to conclude, as the Court of Appeal observed, might result in the list of persons potentially defamed being large and diverse. But contrary to the Court of Appeal's apparent reasoning, that does not mean that the conclusion is unsound. It means no more than that, in such cases, the liability of a search engine proprietor, like Google, may well turn more on whether the search engine proprietor is able to bring itself within the defence of innocent dissemination than on whether the content of what has been published has the capacity to defame. The Court of Appeal further erred, in relation to the autocomplete predictions which appear on page 14 of Annexure A and pages five to seven of Annexure B, in adopting the findings of mixed fact and law made by Blue J sitting at trial as judge alone in Duffy in relation to the autocomplete publications in that case as a basis for concluding that the autocomplete predictions in this case were incapable of conveying the imputations alleged. Contrary, too, to the Court of Appeal's reasoning, the apparent references in Annexure B to previous defamation proceedings involving Mr Trkulja do not significantly if at all detract from the conclusion that the impugned searches are capable of conveying the defamatory imputations alleged. On page one of Annexure B, there appear the words "Milorad 'Michael' Trkulja is a former hitman who shot a music promoter in the balaclava" followed by the word "Streisand'd". Ex facie, those words are capable of imputing some criminality on the part of Mr Trkulja. The Court of Appeal reasoned by reference to a meaning of "Streisand'd" which their Honours appear to have derived from Wikipedia that "Streisand'd" implied a reference to Mr Trkulja's earlier successful defamation proceeding against Google. But that is not an inference that was open to be drawn. It had not been suggested that the meaning of "Streisand'd" was notorious or would be known to an ordinary reasonable person viewing the search results; and the fact that the word may have appeared in Wikipedia is in itself irrelevant. The capacity of a published matter to defame must be assessed by reference to the most damaging meaning that could reasonably be put upon the words in question. Moreover, even if the use of "Streisand'd" could be regarded as suggesting some sort of connection between Mr Trkulja and a defamation proceeding (which, as the matter stands, is a dubious proposition), that would not bar the capacity of the words "Milorad 'Michael' Trkulja is a former hitman who shot a music promoter in the balaclava" to defame Mr Trkulja. Admittedly, there appears on page two of Annexure B a "snippet" of a webpage relating to a Google defamation case which mentions Mr Trkulja's name and a connection to the "criminal underworld", and, on page three, a "snippet" of a webpage which, under the heading "Trkulja v Yahoo! - Defamation Watch", refers to a "music promoter" winning "225000" followed by the words "To the right of the article was a large photo of Trkulja and then an article … the plaintiff is such a significant figure in the Melbourne criminal underworld that events …". Possibly, the latter suggests that Mr Trkulja succeeded in a defamation action against Yahoo and was awarded $225,000. But, even if that be so, it does not necessarily detract from the sting of the words that the plaintiff in that action, namely, Mr Trkulja, was a "significant figure in the Melbourne criminal underworld". It also remains that the search results reproduced at pages two and three were the result of searches for the words "melbourne criminal underworld figure" and "melbourne criminal underworld photos", which of itself would be capable of conveying to the ordinary reasonable person using the Google search engine that there is some opprobrious connection between those terms and Mr Trkulja. And lastly, given that the "snippets" provide little by way of detail as to the defamatory imputations which were the subject of the previous proceedings, or as to the falsity or otherwise of Mr Trkulja's alleged criminal connection, it might be thought that there is little about the "snippets" that a jury would necessarily regard as significantly ameliorative. This observation applies equally to the material appearing on page 14 of Annexure A which the Court of Appeal held was a reference to Mr Trkulja's earlier defamation proceedings that undermined the capacity of the publication to convey the pleaded imputations. Conclusion and orders The Court of Appeal erred in concluding that the matters upon which Mr Trkulja relied were incapable of conveying any of the defamatory imputations which were pleaded and therefore erred in concluding that Mr Trkulja's proceeding had no real prospect of success. It follows that the appeal should be allowed. Pursuant to r 42.07.1 of the High Court Rules 2004 (Cth) Google LLC is made the respondent to this appeal in substitution for Google Inc and the appeal is determined as so constituted. Orders two to six of the Court of Appeal should be set aside and in their place it should be ordered that the appeal to the Court of Appeal be dismissed with costs. Google should pay the costs of the appeal to this Court.
Trkulja
Google LLC
Appeal allowed. Google LLC substituted as respondent. Orders 2 to 6 of the Court of Appeal of the Supreme Court of Victoria made on 20 December 2016 set aside and, in their place, order that the appeal to that Court be dismissed with costs. Respondent to pay the appellant's costs of the appeal to this Court.