Scribe: A Government Project Gone Wrong

The failures that led to the collapse of the Scribe project were so serious that sanctions were warranted.

The chamber of the dispute – the part of the Court of Auditors that rules on grave breaches of financial public order – decided as much. The penalties imposed are largely symbolic (ranging from €1 to €6,000). The identities of the individuals involved carry more significance. Among them are two former directors-general of the National Police.

An “emblematic example of poor project management”

Scribe began in December 2015. It was meant to yield a joint software solution for the police and the gendarmerie to draft procedures. Frozen in March 2021, it was officially abandoned the following year.

In mid-2022, the Court of Auditors revisited the factors behind this failure, described as an “emblematic example of poor project management.” Its analysis highlighted, in particular, a lack of framing, an unstable expression of needs, and an opaque, fragmented governance that diluted responsibilities.

Read also: At Social Security, aging and strained information systems

The chamber of the dispute – which judges grave breaches of financial public order – restated these main points. Its ruling distinguishes failures of the project owner (MOA) and the project manager (MOE).

Failings of the project owner (MOA)

The MOA failings cover three axes:

  • Structural failure in governance
  • Exercise of functional authority over the MOE
  • Ignorance of prerogatives at ministerial and interministerial levels

A problem of hierarchical placement

On the first axis, the chamber lists about ten charges. It begins with the fact that at the project’s launch, the National Police placed the MOA directorate within the cabinet of its director-general (DGPN) rather than in the first operational directorate concerned, namely the Central Directorate of Judicial Police (DCPJ). This arrangement persisted until early 2021.

A insufficient staffing and geographic dispersion

From March 2016 to November 2017, the MOA team consisted of only one person. It reached its peak at the end of 2019, with 12 people. The support provided by the DCPJ (delegated MOA) proved insufficient and only marginally functional, owing to the service’s location in the Rhône region.
Other DGPN directorates contributed to the project. The geographically dispersed configuration led the external contractor (Capgemini) to assume an MOA assistance role not foreseen in the contract. It thus found itself in an ambiguous “judge and party” role for defining functional specifications and organizing the steering governance.

Confusion of responsibilities with the MOE

During the period when the project was joint between the police and the gendarmerie, no document formalized the relationship between the two MOAs. Nor did it clarify their respective responsibilities in their dealings with the ST(SI) (Service for Information Technologies and Information Systems of Internal Security). This common service, housed at the DGGN (General Directorate of the National Gendarmerie), was responsible for the MOE.

In the absence of a formal definition and design of the project, the DGGN took it upon itself to disengage without notice or official information from the DGPN. Subsequently, each entity – the DGPN and the ST(SI) – defined its role autonomously, resulting in a confusion of responsibilities. The ST(SI) could indeed intervene in the overall organization of the project and in defining its functional needs, even though that responsibility belonged to the MOA.

A partially integrated team and no program director

For nearly two years, the project lead commuted two to three days a week to the ST(SI), before eventually settling there permanently. Meanwhile, MOA staff remained in the Rhône.

Read also: AI Strategy: France at the forefront of Europe… but still behind

The Court of Auditors also regrets the absence of a named program director. After the DGGN withdrew, the police’s MOA project lead was not officially vested with the duties she effectively had to perform. The same was true for her successor.

Incomplete initial framing

The execution also faltered due to incomplete initial framing. Not formalized, among other things, were:

  • Evaluation and assessment of the existing police software
  • Expression of the need
  • Study of the functional specifications that Scribe had to meet
  • Feasibility study

No clearly identified budget owner

Successive project leaders received no signing delegation on this matter, and no budget owner was designated within the MOA team until December 2021.

The mission letter issued in March 2016 to the project lead required her to monitor Scribe’s budget trajectory in connection with the ST(SI). She did not take up this aspect, considering that the ST(SI) handled this monitoring by delegation from the DGPN – which was not the case.
The second project lead did not receive a mission letter either, and did not exercise budget oversight, being informed only by the ST(SI) of the amount of credits committed by that service. The third project lead claimed budget responsibility but did not have the necessary signing delegation to exercise it.

The overall budget tracking in full cost suffered from the absence of a responsible individual. The ST(SI) itself had incomplete visibility. It did not know about orders the DGPN and the DNum (Interior Ministry’s Digital Directorate) were placing directly, nor the total cost of MOA staff.

Steering committee governance with episodic operation

At the project’s outset, there was no steering committee structure. The governance bodies were established “progressively and empirically,” in varying configurations. They lacked a methodical calendar and minutes of meetings and decisions.

Failings of the project execution (MOE)

On paper, the DGGN and DGPN had joint authority over the ST(SI), MOE of the project. In practice, its organizational root in the DGGN, leadership by a gendarmerie general, and a staff largely made up of gendarmes made things difficult for the police.

Read also: Digital sovereignty: a distant goal… according to the Court of Auditors

The late involvement of DINSIC

The ignorance of prerogatives at ministerial and interministerial levels showed itself mainly through the late referral to MGMSIC (Ministry Governance Mission for information and communication systems, responsible for approving IT project launches) and, as a result, to DINSIC (now DINUM). Even though the project immediately met the sensitivity criteria and potential implications for the Interior Ministry’s IT systems. And that crossing the €9 million threshold (beyond which DINSIC’s opinion of conformity was required) was foreseeable “before the end of 2018.” This late referral delayed the MOA and MOE from formulating complete project dossiers detailing the project’s prospects. It also deprived the DGPN of well-grounded opinions on possible options.

Internal management shortcomings…

The MOE also lacked adequate human resources. Four years elapsed between the project’s launch and the appointment of a full-time project manager. The staff rose from 2 FTE in 2016 to 4 FTE in 2020-2021. This was not sufficient, despite reinforcements sought from another consultancy besides Capgemini.

Another issue: insufficient mastery of the technical elements used. The ST(SI) was competent only in two of the chosen technologies. All others were identified as requiring external expertise.

… and the control of Capgemini

This latter point takes on even more significance in light of Capgemini’s lack of proper oversight. Two contracts were awarded successively under different terms. First, in December 2015, for four years, a TMMA (tiered multiservice maintenance) contract and ACS (assistance to service centers). It provided support resources to the MOE. There was no obligation of result for the ACS; with the TMMA providing services on demand. The second contract, issued in March 2020 for two years, covered the maintenance of five applications, including Scribe.

The first contract was awarded while the project remained joint between the DGPN and the DGGN. When the latter withdrew, the ST(SI) continued to rely on it even as the MOE’s nature changed, with the project limited to the DGPN and little engaged staff. Capgemini effectively became responsible for developing the application. It then worked alone, with no performance obligation for the ACS services billed on a time-and-materials basis. The TMMA services, meanwhile, could not be mobilized until Scribe reached the deployment phase. Despite this misalignment, the contract ran to its term.

The second contract required Capgemini to deliver a result for the services ordered by purchase order. But its scope appears premature given Scribe’s state of development, according to the Court. The team Capgemini mobilized proved, moreover, to be insufficiently skilled and organized.

A collective fault amounting to €30 million

The chamber of the dispute estimates the state’s direct loss at around €30 million. It says that it originates entirely from the management from 2016 to 2022. Given the stakes of Scribe, the whole constitutes, it adds, a grave… collective fault: it is impossible to attribute the failure to any single person.

The unilateral and sudden withdrawal of the National Gendarmerie served as mitigating circumstances, as did the governance shortcomings in the Interior Ministry’s secretariat general.

The judgment dates from June 30, 2026. It could be appealed to the Financial Court of Appeal within two months.

* The first was in office from June 2014 to August 2017. He received a €4,000 fine. The second succeeded him, serving until February 2020. He received a €2,000 fine. Their names do not appear in the Court of Auditors’ decision, which we respect here. The four other people concerned are:

– A former advisor to the DGPN for technologies and digital affairs (€6,000 fine)
– A former head of the ST(SI) (also €6,000)
– Two former project managers (€1 symbolic each)

Dawn Liphardt

Dawn Liphardt

I'm Dawn Liphardt, the founder and lead writer of this publication. With a background in philosophy and a deep interest in the social impact of technology, I started this platform to explore how innovation shapes — and sometimes disrupts — the world we live in. My work focuses on critical, human-centered storytelling at the frontier of artificial intelligence and emerging tech.