1 Introduction

Twenty years into its legal existence, it is still not simple to pinpoint what exactly is wrong with Regulation 1049/2001.Footnote 1 There is ample consensus that Regulation 1049/2001 has served but exhausted its purpose and urgently needs to be replaced.Footnote 2 More difficult to ascertain is what specific mechanisms in Regulation 1049/2001 need replacing and how to go about this in a way that is both appropriate and efficient.

There has been no further overarching legislative contribution to the European Union (EU) policy on public access to documents (PAD) since the adoption of Regulation 1049/2001. True, there are a couple of references in the Charter of Fundamental RightsFootnote 3 and there are dispersed self-adopted rules for specific institutions/agencies which have been adopted post-2001. In the meantime, with the legislative framework stalled, an incessant – and on occasion disruptive- amount of case-law continues to develop.

Case-law – both recent and less recent – has touched on the general interpretation of Regulation 1049/2001; on the nature of applicants; on the role of institutions and agencies; on the scope of exceptions and presumptions of harm;Footnote 4 on the consequences of legal silence; on the interests and legitimacy of interveners, and most crucially in 2022, case law has re-addressed the nature of documents. On that premise, it seems to me that the leading thought for 2022 within this policy is “pressing the right buttons”.

I will consider two distinct issues to anchor my contribution: first, the issue that case-law has partly altered the formal scope of the policy from a document-based approach to an information-based approach. Second, the issue that applicants and the academic community are currently disenchanted by the policy due to a generalised stasis affecting both the EU judiciary and the Ombudsman.

Tension exists between the legislative and judicial approaches concerning what tasks information officers are required to undertake in order to satisfy requests for information. When contained in digital databases, in 2017,Footnote 5 the information contained in a document began to be identified as a document unto itself, and thus both terms have since been used interchangeably, subject only to the caveat of extractability by way of pre-programmed search tools. Expressions such as “regular routine search”, “sufficient sophistication” of a database, “substantial investment” and “programmes easily available on the market” need urgent clarification by the Court in order for new boundaries to be drawn. The new framework however is not applicable to applications for information dispersed over several printed-on-paper documents.

Regarding the institutional stasis, on the one hand we highlight how the paucity of existing judicial remedies continues to inflict damage on applicants’ expectations and suggest that the inclusion of injunctive relief in the powers of the Court of Justice of the European Union should be given some thought. Lastly, albeit acknowledging that recommendations are insufficient to drive EU access policy, we conclude that the proposal that the Ombudsman be accorded binding powers carries an inherent serious risk of cognitive dissonance between the EU framework and national frameworks.

2 The formal scope of Regulation 1049/2001 concerns documents and not information

2.1 A Policy written for the “Material Paper World”Footnote 6

As was true of the 1993 Code of Conduct,Footnote 7 Regulation 1049/2001 concerns public access to documents. This is a formal boundary – born of legislative choice- that for many years excluded development into an EU policy of access to information.

In 2017, regarding this EU policy (albeit with the appropriate disclaimers) we qualified the quest for documents as generally interchangeable with the quest for information.Footnote 8 We did that -having the material paper world in mind- because, in that world, requests for documents are made simply because documents contain information. This acknowledged, within that world such a situation (the interchangeability between document and information) does not withstand all tests. When focusing more narrowly on the text of the 1993 Code of Conduct and on as much of that structure that carried over into Regulation 1049/2001, the document/information distinction is glaring.Footnote 9

From a comparative point of view, during the 2021 ERA seminar, regarding this distinction in the national laws of the EU, the difference between two contrasting approaches – the Scandinavian approach and the Anglo-Saxon approach – was highlighted.Footnote 10 The former approach was born of the Swedish Freedom of Press Act 1766 that focused on access to documents. Therein, PAD is, however, limited to definitive documents, thus excluding drafts. In contrast, the Anglo-Saxon approach is based on a Freedom of Information Act. Europe, thus, is acquainted with a panoply of options where some systems construe PAD policies around documents whilst others favour unclad information at the centre of transparency.Footnote 11

The EU policy’s identity as one drafted by the legislator for documents, is well-anchored in both EU law and case law. For the printed-on paperFootnote 12 version of documents that are produced or held by the EU institutions, it is established – and accepted – that applicants are not enabled to ask EU institutions solely for information. If information x is sought, applicants are required to identify the document(s) (containing information x). Success depends, first, on the off-chance that documents containing information x pre-exist the application.

Regarding information contained only on printed-on-paper versions of documents, institutions are not required to draw up new documents by combining information sourced from disperse files to present the applicant with rearranged content. In this context, when an original document containing information x does exist, and cumulatively the applicant has identified the document, and still no exception is applicable, the applicant should either be sent the document or be invited to consult it in loco.Footnote 13

2.2 Part of the EU Public Access Policy “goes digital”

The ability to rely on the classical document/information distinction was rattled when, in 2009, a first case concerning access to a (digital) database arose. To structure seminal case law on this point, the existing case law acquis constructed for printed-on-paper documents was adapted by the General Court (GC) and the Court of Justice (the Court) to fit requests for information without a fixed material basis. This process shook the EU access tree and, over the next twelve years, six rulings fell out (Table 1: Sequence of case law regarding digital Databases): Dufour (2011), Typke (2015, Typke GC) (2016, Typke Op) (2017, Typke CJEU), Frontex (2019) and Kedrion (2022).Footnote 14

Table 1 Sequence of case law regarding digital Databases
Full size table

Each of these rulings became a building block in the disruptive path that the EU judiciary chose to take.

The rulings involved applicants litigating against the European Central Bank (ECB), the European Commission (the Commission), the European Union Border and Coast Guard Agency (Frontex) and the European Medicines Agency (EMA). The ripples of the case law that emerged in consequence spread wide and superimposed upon one another since there were multiple points of origin. In addition, since the case law progressed first through two of the most important institutions (the ECB and the Commission), it was thus more easily replicable at the level of the agencies.

The above-mentioned rulings focused on information included in databases. Dufour, running between 2009 and 2011, was certainly a case concerning a digital database. The same is true of the later Typke proceedings that were conducted between 2013 and 2017. In the Frontex case that ran between 2018 and 2019, it was the institution that paved the way for the court to treat requests for information contained in databases in a way that contrasts with the treatment of information kept outside a database. The institution moved first, admitting of its own volition, not to have worked from an original document. The information had accordingly been extracted from documents in a database and rearranged in another format/layout to comply with measures of inquiry ordered by the court. This allowed the EU judiciary (both the General Court and Court) to structure the seminal case law for databases for a decade: 2009-2019. In addition, the courts acknowledged and benchmarked differences in institutional easiness of manipulation between information dispersed over several printed-on-paper files and information held in digital form. The latter (say the courts) is easily manageable via search tools that are not available for the material paper world.

The easiness acquis is the most plausible explanation for the outcome of the Kedrion proceedings that ran from 2020 to 2022. It was held irrelevant, in this case-law, that the contested documents were not part of a digital database. If was sufficient, at that point, for the General Court, that the contested documents (nether contained in a database nor a database themselves) were digital: it was a document held as a PDF. The next conclusion drawn was that, since digital, it fitted the judicial definition of documents the information in which was easy to manipulate if the proper tools were available. Lastly in the cascade of argument, if the appropriate tools were not available, they were (said the court) easy to procure. Consequently, as had previously been held true for databases, manipulation of digital documents not included in databases does not amount to “excessive workload”.

The expressions “tools that make documents easy to manipulate”, “programmes easily available on the market” “programmes easy to procure”, translate the concepts that bridge the digital and the physical worlds and that impinge dramatically on the lengths that information officers must go to in order to properly discharge the burden of responding to applicants. Below is a more detailed explanation of the sequence of case law that disrupted the pre-1999 acquis.

2.2.1 Dufour 2009-2011

The Dufour proceedings began when a doctoral student of sociology lodged an action for annulment against a decision of the Executive Board of the ECB, in which the latter refused to grant the applicant access to the databases used as a basis for preparing ECB reports on staff recruitment and mobility. The ECB responded by claiming that the databases to which the applicant sought access were not documents within the meaning of Decision 2004/258.Footnote 15 The institution also argued that the copy-paste (electronic) operations deemed necessary to satisfy the applicants’ request amounted to creating a new document.

The applicant argued that the entire database was a document and furthermore that granting his application would put the institution to almost no trouble at all.Footnote 16 The General Court defined database,Footnote 17 agreeing with the applicant, and went on to state that if sufficiently sophisticated,Footnote 18 a database’s search tools could easily provide an applicant not only with the entire dataset, but possibly with the same dataset arranged in a variety of ways.

The moment §153 of the Dufour ruling was written and published, the EUFootnote 19 policy on public access to documents changed dramatically. It went digital. The General Court ruled that if data is stored electronically and the request for access is compatible with the database’s classification scheme and can be extracted by way of a “normal or routine search” it may be the subject of an application for access. Agents in institutions were left to press the right buttons.Footnote 20 Since then, agents not only have to test databases’ classification schemes, but, additionally, must establish what a “normal or routine (electronic) search” is.

2.2.2 Typke 2013-2017

Less than two years later, on 15 April 2013, very different proceedings began. They were initiated by a member of staff of the Commission, Rainer Typke, who had taken part in tests for admission to a different grade.

The applicant asked that the Commission (draw up and) grant him “access to a table containing a series of anonymised data on the tests in question, which had been taken by approximately 45 000 candidates”. The Typke General Court proceedings would signal – building on Dufour – that a great divide had been wedged into the EU policy on public access to documents.

On the one hand, the General Court would not stray from the principle that applications pertaining to the material paper world remain attached to the acquis that distinguishes documents and information.Footnote 21 Conversely, the General Court went on to make a bold, new statement: applications for data stored digitally -because of their immaterial basis- cause the EU policy on public access to documents to take on a new breadth.

The General Court’s interpretation was unexpected since the table Mr. Typke asked for was complex.Footnote 22 And it required anonymisation of each participant. Moreover, the applicant clearly irritated the Commission’s Secretary General who stated that Regulation No 1049/2001 was not intended to oblige the Commission to perform IT (information technology) operations in order to extract information stored in various databases.Footnote 23

The applicant-institution conversation quickly escalated from this into technical jargon including terms as “strict mark-up syntax; item banks, structured query language (SQL queries) and new identifiers”. New language to make progress on the only topic that matters today: what degree of sophistication do databases’ pre-programmed search tools allow? It is a fascinating new analysis and a far cry from the traditional disputes about how many pages long a document is, the time a document would take to photocopy, if the applicant is able to travel to the institution’s headquarters, if someone is available to supervise in loco consultation etc…

The General Court then left it to the Commission to decide whether a “routine search” was, or not, feasible in that specific case. However, a word of warning transpired since the General Court also added that had the applicant Typke requested access to the “entire” database, the outcome might have taken a different turn.

The Typke case went on to appeal, and very early on, Advocate General Bobek, in his Opinion, trumpeted again that the digital era required the enforcers of Regulation 1049/2001 to adapt. This was the fourth time the matter had been discussed by the EU courts. It had been discussed first in Dufour, secondly during the first instance phase of the Typke proceedings, thirdly in AG Bobek’s Opinion delivered within the Typke appeal proceedings, and finally it was to be addressed once again by a judgment that would bring the Typke appeal to a close. After three hints that an about-turn might take place, it was without surprise that the public read in the judgment that that the Court concurred with the Advocate General’s Opinion. The Court masterfully altered the scope of the EU policy on public access to documents by kneading together the words ‘information’ and ‘document’. Subject to the caveat of its extractability by way of pre-programmed search tools, in the digital era the words ‘information’ and ‘document’ may be used interchangeably.

The only nod to the prior acquis constructed for the material paper world, involved the concept of excessive workload.Footnote 24 For the digital world, this was reconceptualised as “substantial investment”,Footnote 25 and explained to mean “a required alteration either to the organisation of the database or to the search tools.” If either step is required to satisfy an applicants’ request, then the institution may legitimately refuse to undertake those measures.

2.2.3 Frontex 2018-2019

The Frontex case builds on Typke and contributes to thickening the growing acquis on public access to information contained in digital documents. Interestingly, the applicants in this case had simply asked for “information (x) contained in documents” without identifying specific documents. The information requested was “the name, type and flag of all vessels deployed by Frontex between 1 June and 30 August 2017 in the central Mediterranean under Operation Triton 2017”. In a pre–1999 setting such a request might have been stonewalled by the combined interpretation of Regulation 1049/2001’s Articles 2(1) and 6(1) – written for the material paper world – that burden the applicant with a higher level of specificity.Footnote 26

It was the institution’s inertia regarding cooperative duties with the applicant combined with a self-incriminating institutional statement that kicked the Frontex case into the digital arena.Footnote 27 The institution declared to have recovered and rearranged the contested information from a digital database not for the purpose of aiding the applicant but, conversely, to satisfy a measure of inquiry ordered by the General Court.Footnote 28

It was but a short throw for the General Court to recover the Typke-CJEU para. 36 paradigm established by the Court: “all information that can be extracted (…) through pre-programmed search tools (…) must be regarded as an existing document.”Footnote 29

2.3 Easy does it! 2022 – More bites, less pieces:Footnote 30 informatics and public access to documents beyond databases

2.3.1 Kedrion 2020-2022

Finally, a third ruling Kedrion (2022) tore even harder at the document/information boundary.Footnote 31

The object of the Kedrion ruling was not a database. However, in this case, the new argument that had emerged specifically for databases – “the easiness of electronic manipulation” – was thrown over the entire world of electronic documents whether catalogued in a database or not.

Kedrion is a company that manages the collection and manipulation of blood plasma in five Italian regions that participate in a consortium called “Planet”. It asked the European Medicine Agency (EMA) for access to the list of centres for collection and manipulation of blood plasma contained in the plasma master file (PMF) of a pharmaceutical company, Takeda. Among other arguments, the agency held that even a mere list of names and addresses of Takeda’s centres would require manipulation of the PMF file by the agency since the order according to which centres appear in the original PMF (as main, secondary, or mobile centres) would reveal management strategy (which was commercially sensitive information).

Again, the question was: in such a context is the EMA compelled to create a new document?

The EMA objected that the contested list was only held in PDF format (thus the contested document was neither in a database nor a database unto itself) and that to re-transcribe it would have excessively burdened the institution.

The Kedrion ruling recalls that differently to the past, specifically since the TypkeFootnote 32 ruling, in general, institutions may be required to present information to the applicant in a rearranged form.Footnote 33 That reasoning is equally applied to documents not catalogued in databases since the General Court states that opportunities arising from informatics (might and do) reach beyond databases.Footnote 34

Still, the General Court pointed out that by resorting to “programmes easily available on the market “it is possible to convert “PDF image plus text”Footnote 35 into another format that allows information to be easily rearranged by way of standard procedures. Finally, after the appropriate program is procured, this double manipulation: conversion + rearrangement (said the General Court) could not be considered a substantial investment.

2.3.2 Conclusion: digital documents in the EU Access Policy: from PAD to PAI

Concluding on this point, first, regarding information in databases new buttons must be pressed to perform normal routine searches. Secondly, for documents held merely in electronic format albeit not within a database, programmes easily available on the market are to be bought and utilised by institutions. Correspondingly, in 2022, for digital documents, it is information, not a document that is placed at the centre of the public access debate.

After Kedrion, when information is available in a digital version, it is unlikely that the EU judiciary will ever revert to interpreting the formal scope of Regulation 1049/2001 as one still connected to (classical) documents. All digital information held by the EU institutions has irreversibly slid, at the hands of the judiciary, into a new framework of public-access-to-information: PAI. Therein, the language of access speaks in bites rather than pieces.

3 An institutional stasis affecting both the Judiciary and the Ombudsman

Turning now to the second issue of our contribution, 2022 is also the time in the history of public access to documents in the EU in which two institutions openly struggle with their formal bounds. The Courts of the EU, already unpopular because proceedings seem to take too long, are (still) unable to offer injunctions to applicants. At the same time, a speculative insinuation has been publicly voiced regarding the Ombudsman: that, within the field of public access to documents, this institution would do a better job if given binding powers.Footnote 36

First, and regarding the EU courts, PAD-related litigation in the EU supranational courts is governed strictly within the bounds of the action for annulment, i.e., Art. 263 of the Treaty on the Functioning of the EU (TFEU). The only other (litigious) alternative is to engage the Ombusdman. Bluntly put, the best outcome of an Art. 263 TFEU action is that after one or two years of litigation, an institution might be told by an EU court that it has erred in denying access to a document and must – as a consequence – take a new decision on the matter.

Insofar as concerns the Ombudsman, just as bluntly put, the best outcome is that after one year, a non-binding recommendation is issued stating that the institution erred in denying access to the document. These narrow bounds have led to an institutional stasis. Institutions have no legal path to overcome their inadequacy to assure what we may call a “EUropean”Footnote 37 that if (after a year or two) a first reason for denial is not validated (by the judiciary or by the Ombudsman) an obligation to hand-over the document will arise.

When determined to gain access to a document EUropeans will find this institutional panoply unattractive: slow, obstacle-ridden and of low remedial intensity: in sum, ill-equipped to help applicants exercise a right of access to documents at full potential.

3.1 The Judicial stasis

Concerning the Judiciary, there is a triple jeopardy: time, cost and remedial depth.

3.1.1 First, actions for annulment, from lodging to ruling take a long timeFootnote 38

Even a very superficial analysis of a small sample of cases regarding access to documents will result in the conclusion that litigating these cases in the EU courts is an (excessively) lengthy process. If we refer to Table 1, which describes six episodes, we observe the following: Dufour took 24 months, Typke (from first instance to a final ruling from the CJEU) took 44 months, Frontex took 22 months, and, finally, Kedrion took 16 months. The numbers raise a crucial question of incentives: can we really ask EUropeans to wait between two to three years for a validation or cassation of an institutional refusal to disclose a document?

3.1.2 Secondly, the access litigation imposes significant cost on applicantsFootnote 39

Funding may also be a significant deterrent for applicants to engage in litigation. A recent order on expensesFootnote 40has brought to light that a case may impose costs upwards of 10,000 euros on an applicant. Even though (some) expense is inherent in litigation, within a policy that seeks to foster institutional transparency vis-à-vis the public at large, this is an often-neglected topic worth looking into.

3.1.3 Thirdly, even twenty years into Regulation 1049/2001, the public has a right to have illegitimate refusals annulled: no less, and no more

Existing remedies are confined to discussing whether a refusal was legitimate or not, but fall short of admitting that applicants might be handed the documents directly by the EU courts.Footnote 41 In addition, the galloping growth of presumptions of harmFootnote 42 has flung blanket-bans far and wide over masses of information. As a consequence, when considering pursuing the judicial route, the results of a cost-benefit analysis are heavily swayed against the popularity of Art. 236 TFEU as an efficient tool to gain access to documents.

It is not by chance that litigation has mostly been preferred by those with plenty of time on their hands, or who have access to in-house counsel and/or to abundant funds.Footnote 43 These points taken together (time, cost and the absence of injunctive relief) have allowed another player’s role within the policy area, i.e., the Ombudsman, to compete for applicants’ attention.

3.2 Stasis also affects the Ombudsman

Concerning the Ombudsman, here too an institutional stasis arises. However, here it is much more from the lack of remedial intensity and much less due to lengthy procedures. Conversely, lack of expense acts as an incentive to pursue this route as a preferential one to the judicial alternative.

A classical analysis of Ombudsman’s powers dictates that -here too- we must be honest about the remedial paucity involved: Ombudsmen (and Ombudswomen), at the most, after stating that an institution is at fault, provide complainants with a non-binding recommendation that the institution involved amend its ways. After all, the judicial alternative (litigation via Art. 263 TFEU) also falls formally short of coercive powers over the institution, i.e., an order to make the contested documents available.

It is not surprising then that two proposals, both bold, were discussed -among other topic s- in 2021 at a debate organised by the Ombudsman on altering institutional design.Footnote 44 The first one was that the Ombudsman be granted binding authority in proceedings concerning access to documents, the second one is that an Access Commissioner be added to the EU’s institutional Design. Both proposals involve risk.

3.2.1 Binding powers for the Ombudsman? Risk: cognitive dissonance

The first risk – duly acknowledged by the current Ombudsman Emily O’Reilly – is that of too much cognitive dissonance between the EU system and national legal systems: an Ombudsman with binding authority is (for most EUropeans) no longer an Ombudsman.Footnote 45 Even a “new legal order”Footnote 46 such as that of the EU must mirror basic tenets of its component members’ legal systems to avoid the risk of (too much) cognitive dissonance between national law and the supranational EU framework. A new legal order “ma non troppo” is a philosophical limitation that the Union must acknowledge if it means to keep a cultural umbilical cord attached to its peoples.

Moreover, the EU courts issue binding decisions and even that is not enough to exclude the possibility that an institution is able to invoke one reason after another to ground non-disclosure. In truth, it’s not just the binding nature of authority that matters but the injunctive muscle that for now is altogether absent in the EU system.

3.2.2 An information commissioner? Risk: sludge

The discussion on changes to Regulation 1049/2001 also included the idea of the advent of an Information Commissioner.Footnote 47 If taken up, this would bring about a new institutional framework (after the requisite Treaty reform) featuring two courts, an Ombudsman, and an Access Commissioner. It is respectfully submitted that it would possibly add (another!) layer to what is already a “sludge-filled”Footnote 48 route to obtaining documents from the EU.

If an Information Commissioner were designed for the EU where would it fit? Would it be a stand-alone institution, and thus would we discard the Ombudsman’s past role altogether? Would it amount to a (new) structure within the Ombusdman’s structure? Would there be an appeal from Access Commissioner to Ombusdman (thus mimicking the judicial setting that comprises the General Court/Court with the associated risks of lengthy proceedings)?

During the debate, Ombudsman O’Reilly recounted (on the basis of a comparative law analysis) that where an Information Commissioner steps up, an institution that does not wish to follow a recommendation to disclose must then sue the Access Commissioner. Is this where we want to go? Into a place where institutions sue institutions while the applicant waits?

3.3 Disbelief in the judicial route?

Both limbs of the proposal to release the Ombudsman from the present institutional deadlock (i.e., those of binding authority and of an Information Commissioner) give up on the judicial route. Relegating (judicial) litigation to the blessed few with time and funds on their hands and/or to the simply brave. Such a message is twofold and delivers a dismal prognosis. On the one hand, it rests on the belief that the judicial “tempo” cannot be bettered. On the other hand, the message suggests that we should relinquish hope in a different future status for the EU judiciary.Footnote 49 Today, formally confined by Art. 263 TFEU, it might yet grow up into a “proper” administrative court with competence to enjoin institutions (albeit confined, in doing so, to the access policy) or ultimately even to substitute itself for the institutions.Footnote 50

In the fight against this institutional stasis, we concur that a balanced solution may well lie in improving existing routes – but not beyond recognition. Under the current remedial frame (limited to annulment by the judiciary, and dispensation of non-binding decisions by the Ombudsman) both the courts and the Ombudsman could still work on the “tempo” of the dispensation of justice. Possibly by way of setting up of, say, a special section/chamber in each, thus creating a public access procedural highway as opposed to the current status quo under which access to documents cases are discussed alongside a panoply of cases covering different topics and are not ear-marked to be readily dealt with and swiftly gotten out of the way.Footnote 51

Specifically, as regards the EU’s judicial branch, the time is also right to discuss and consider an about-turn in competence that – as regards the access policy- would transform the General Court and the Court into administrative courts able to dispense the more intense remedies known to national law – namely, injunctions.

4 Conclusion

In sum, several issues must be resolved in 2023. First, further clarification is needed from the courts on the meaning of the expressions “normal or routine searches” and “programmes easily available on the market”. Secondly, the width and depth of the EU courts’ competence should be revised. Thirdly, the extension or not of the Ombudsman’s role in access policy needs to be discussed. And, finally, the recasting of Regulation 1049/2001 can no longer be postponed.