The case law on internet publications is chiefly concerned with dating the information found and the applicable standard of proof. The law on this seems now to have been established by the decisions in T 286/10 and T 2227/11, which were endorsed more recently in T 1711/11, T 353/14, T 545/08, T 1589/13 and T 1066/13. Those decisions also cited the EPO practice set out in its notice concerning internet citations (OJ 2009, 456 to 462) and the Guidelines for Examination at the EPO (G‑IV, 7.5 – November 2018 version). The appropriate standard of proof for internet citations is the "balance of probabilities" and not "beyond reasonable doubt".
The conclusion reached in the earlier decision T 1134/06 (followed by T 19/05 and T 1875/06) that the stricter standard of proof "beyond reasonable doubt" had to be applied to internet disclosures has been refuted. Nevertheless, examining a source's reliability as described in that decision has not become entirely obsolete; see, for instance, T 545/08 (points 15 and 18 of the Reasons), T 353/14 (point 2.2.7 of the Reasons) and T 286/10, in which the board found that, the Times Union newspaper's www.jacksonville.com was, prima facie, a known and reliable source of information. While the board in T 2227/11 agreed with the detailed reasoning in T 1134/06 that internet citations of prior art entail a number of difficulties in assessing the authenticity of notably the publication date and the content, there was no reason to impose a stricter standard of proof. These difficulties may require some far-reaching investigations into the matter and the provision of supporting evidence.
Observing that documents published on the internet presented a special problem not encountered with conventional publications, in that changes might be made to them over time that were not readily traceable, the board in T 286/10 endorsed the approach taken in T 2339/09 and T 990/09 and applied the principle of free evaluation of evidence as explained in T 750/94 (OJ 1998, 32).
According to T 545/08, it was correctly stated in the Guidelines (G‑IV, 7.5.2, unchanged in the November 2018 version) with respect to internet disclosures: "The standard for assessing these circumstances is the balance of probabilities. According to this standard, it is not sufficient that the alleged fact (e.g. the publication date) is merely probable; the examining division must be convinced that it is correct."
The burden of proof generally lies with who asserts. In the specific case of internet citations of prior art cited by the EPO, the burden of proof thus lies with the EPO. If the EPO, however, is satisfied that, on the balance of probabilities, an internet citation constitutes prior art, it is then up to the party to prove otherwise (T 2227/11, T 1589/13, see also on burden of proof T 545/08, points 12 and 13 of the Reasons; T 1066/13, "directory listing").
(ii) Illustrative case law
The facts on which any finding of public availability is based must be established with a sufficient degree of certainty in order to convince the competent organ of the EPO in view of all the relevant evidence that they have indeed occurred. This holds true even if the determination is made on the basis of probabilities and not on the basis of absolute certainty ("beyond any reasonable doubt") (T 545/08, point 11 of the Reasons, applied in T 1236/13 in the context of the date of finding a webpage, as displayed in the URL as a date stamp).
It is well-recognised policy of the Internet Archive's Wayback Machine to crawl the internet and archive crawled web pages using the date and time when the web page was crawled as part of the URL. The board did not see how the URL used by the Wayback Machine for archiving purposes can be considered to be dynamic, as argued by the appellant (applicant), nor did the appellant substantiate this argument any further. The board in T 523/13 was therefore of the view that the date stamp was established with a sufficient degree of certainty (cf. T 545/08, point 11 of the Reasons), unless proven otherwise (ibid., points 12 and 13 of the Reasons).
In T 286/10 (www.jacksonville.com / www.archive.org) the board stated that internet publications presented a special problem compared to conventional publications, given the potential for making changes that were not easily traceable. Internet publications did not in principle call for a different standard of proof; any uncertainty linked to such disclosures had to be overcome in a way that ensured a sufficient degree of probability and established a presumption of availability that would convince the judge.
In T 2339/09, according to the search report, D4 was an internet article dated 22 May 2006 and relating to a product catalogue of HBE GmbH, which had been found in an internet archive at www.archive.org on 21 March. The date of the catalogue's online publication was therefore 22 May 2006, i.e. before 17 November 2006, which was the filing date of the application in question. Moreover, the catalogue bore an imprint with the date "11.10.04", which suggested that it had been published "offline" even earlier. In any event, the relevant dates were prior to the filing date and it therefore had to be found that D4 formed part of the state of the art. The board held that the applicant (appellant) bore the burden of presenting the case for and proving the contrary, i.e. that D4 had not been published before the filing date.
In ex parte case T 1961/13 the board noted that it should not have been necessary for the appellants to investigate the relevance of Google's date indications. It was the task of the examiner to make an objective assessment of what a particular date indication was intended to represent and how reliable it was, and to make further investigations if necessary. If it was not understood how a particular date reported by a search engine was generated, it could not be used as evidence of a publication date. A date reported by Google was inherently unsuitable to serve as evidence of the publication date of a document.
The reliability of the information contained in the extracts of Wikipedia cited by the opponent could not be assessed and/or there was no evidence that the content of those documents was made available to the public before the effective date of filing of the patent. Accordingly, and independently from the question whether or not there was any justification to submit those documents on appeal, those documents could not be used to prove common general knowledge available at the effective date of the patent (T 378/15).
In T 1469/10 the board pointed out that the ETSI 3GPP organisation, a reputable standardisation body, had clear and reliable rules for publishing any meeting contributions, in particular as to the documentation of the uploading to the public file server. Therefore, a publication date indicated on the 3GPP document lists ("timestamp") was of high probative value and might serve as prima facie evidence as to the date on which the document was available to the public. Accordingly, the board pointed out that the dates ("timestamps") indicated on the 3GPP document lists corresponded to the dates on which the respective documents were uploaded to the 3GPP file server.