The ‘Coolglass Effect’: Some Implications for Public Sector Data Strategy 

“Data is the new oil”, “organisations need to be data-driven”, “Data is an asset”. These catchphrase calls to action should be familiar to anyone who has wandered past a conference or webinar about data in the last few year. The role of data and how it can deliver value has come under the microscope again with the hype surrounding Generative AI and emerging uses of both novel and long available machine learning and AI technologies.

But, as I have written before here and in other places, the catch phrases and buzzwords are meaningless and the hype machine an empty box making lots of noise if we do not step back from the technology and examine the strategic value of data in the round as an enabler for other things. However, as we step back we also need to consider how we avoid our use of data enabling things we don’t want to have happen and how we can better use data to minimise or mitigate the risks of things happening which we might not want. 

‘Value’ as a concept is a double edged sword. Delivery of great value on one hand can come with a cost on the other – the explosion in the use of fossil fuels and  chemical pesticides for example transformed industry and agricultural production (value) but at significant negative impacts to the environment. When developing organisation strategies or data strategies we need to consider those potential harms. Indeed, the GDPR explicitly requires consideration of the impact on the fundamental rights and freedoms of individuals when organisations are processing data about them, and the EU AI Act requires impact assessments for certain types of AI systems in certain sectors and scenarios. 

An additional perspective 

Public Bodies in Ireland have another strand of data-related regulation that might not be immediately obvious as having a data-related impact but which has been given a renewed focus arising from a recent decision of the High Court in Coolglass Windfarm Ltd v An Bord Pleanála (Coolglass Windfarm Limited v An Bord Pleanála [2025] IEHC 1).  

The legislation? The Climate Action and Low Carbon Development Act 2015 Act 2015 as amended. 

The decision? A blunt reminder from the High Court, in the form of Humphreys J, that relevant bodies under the Act have a statutory obligation “insofar as is practicable” to perform their functions in a manner that is consistent with, amongst other things, “the objective of mitigating greenhouse gas emissions and adapting to the effects of climate change in the State” (see paragraphs 68 to 76 of the judgement for some insight into the decision). 

The Impact? To assess the impact we need to first consider the scope of ‘relevant bodies’ under the 2015 Act, as amended. In summary, if you are subject to FOI you are a relevant body for the purposes of the Climate Action Act. With that out of the way we need to consider what the finding of the Court means. 

The Focus of the Decision – “should do” vs “must do”

The Court points out that the amendments to Section 15(1) of the 2015 Act introduced in the 2021 Act are “clearly a step-change from a mere have-regard-to obligation to a comply-with obligation” (emphasis added). In other words, it is a mandatory requirement for a relevant body, in the conduct of its functions, to consider the impact of those functions on the objectives set out in Section 15(1) of the Act. As Humphreys J. puts it in paragraph 72 of the judgement “what we find is that it imposes sweeping obligations across the public sector from the Government down.”  As a matter of statutory interpretation, the Court held that “the actual statutory context of s. 15(1) viewed in the light of the Act as a whole supports an interpretation that the subsection means what it says.” 

So, any ‘relevant body’ must, as a mandatory duty, consider the impact of their functions on the national climate objective and the other related objectives set out in Section 15(1), including Section 15(1)(e)’s objective of “mitigating greenhouse gas emissions” and then must insofar as is practicable, perform its functions in a manner consistent with those objectives. 

The ‘Coolglass Effect’ and Information Management 

The carbon footprint of Information management is increasingly recognised as being disproportionately affected by the common methods and practices for managing, or rather not managing, data and content in electronic and hard-copy forms. Processing of data and information was estimated to have generated 4% of global greenhouse gas emissions in 2020. This is about the same as the combined emissions of countries like Canada and Germany. Data centres have a combined impact (2.5% of human induced carbon) greater than the aviation industry (2.1% of human induced carbon). And it is growing, and growing faster since the adoption of Generative AI tools such as Microsoft CoPilot or ChatGPT. 

However, some 65% of data and content held by organisations is what researchers in Loughborough University term “dark data”. This is data which is sitting unused and obsolete in file shares, email inboxes, content libraries, databases, and hard drives. This is data which is still being indexed, searched, and stored. And each time an operation is performed on that data, whether by a human or an automated process, energy is consumed which contributes to the inexorable increase in carbon impact and water consumption. 

(Which, a cynic might say, makes a mockery of email footers telling people to think twice about printing the email out so as to save the environment.) 

The problem of ROT

Of course, this issue of Redundant, Obsolete, and Trivial data is an issue that is well known in organisations of all sizes, and the general actions that are needed to address it are equally well known.

Things like:

• a focus on Information Governance, with clear definition of and application of data and content retention schedules,
• clarity on processes and process definition, reduction in duplication of data capture, changes to work practices to reduce the creation of duplicate data,
• improvement of data quality to reduce information scrap and rework and reduce the need to check, correct, and redo data-related tasks,
• improvement in organisation data and digital literacy through proper training of staff on data and data-related processes,
• and a shift from treating data-related projects as technology implementations but instead as business function changes (enabled by technology)… these are some of the things that an organisation might consider. But all too often they get overlooked or pushed aside as there is a new or shinier priority initiative. 

Scope of the Coolglass Decision

Of course, from the Climate Action Act perspective, this only matters to a relevant body if the processing of data and content in electronic form is part of the functions of that body. So, if you are a leader or staff member in a ‘relevant body’ who does not capture or create data, doesn’t prepare or review documents, doesn’t create and manage spreadsheets, or otherwise access and use electronic data in the course of your functions, you can stop reading now. 

Still here?  

Practical Implication of Coolglass for data “stuff”

The practical implication of the Coolglass decision is that the unsexy tasks of managing data, defining metadata, applying retention schedules, and general ‘housekeeping’ are no longer optional ‘best efforts’ things, even if they were not already things that are required to be done under legislation such as GDPR or NIS2.  

The Climate Action Act makes it a mandatory requirement for relevant bodies to consider the carbon footprint of ROT when performing their functions, which means that the effect of Coolglas is to require proactive planning of data and information management activities in relevant bodies, to include consideration of the reduction and elimination of dark data through both the proactive management of data and records as assets of the organisation and the continuous improvement of data and content management processes to reduce and remove waste and duplication. 

In short: all those things that you should have been doing for GDPR and NIS2 now have a mandatory driver tied directly to stated Government policy on Climate Change. And at the National level in the Data Governance Roadmap for Ireland, we need to consider the Leadership, Policy, People, and Technology implications of the Climate Action plan in our design and planning for data stuff.

The Impact on AI and Generative AI 

Just as value is a double-edged sword, the impact of Section15(1) of the Climate Action Act and the Coolglass decision is also double-edged. Because artificial intelligence and related technologies can help improve efficiencies in data and information management, reducing waste, eliminating redundancies, and streamlining processes. But the same overarching family of technologies is also contributing to a massive increase in electricity consumption, with companies like Microsoft looking to build or acquire nuclear power stations to meet the demand.  

What Coolglass means for the adoption of AI and Generative AI tools in ‘relevant bodies’ is that there needs to be much greater scrutiny given to the use case and business need being met, the design of the process being implemented, and the consideration of the appropriateness of a technology to serve those needs. And this scrutiny needs to examine if the proposed solution will be more carbon intensive or less carbon intensive then an alternative, or then doing nothing and leaving a process unchanged so that the Climate Action impacts are understood and planned for.

Things to Consider

For example, the tagging and categorising of documents in content management systems like Microsoft SharePoint can be done manually (slow, time consuming, inaccurate) or it can be automated using machine learning and artificial intelligence, combined with careful rearchitecting and design of the underling information architecture model of your SharePoint sites. This can improve the speed and efficiency of search (carbon saving), reduce the risk of incorrect results being returned (carbon saving), simplify security and permissions management (improved compliance), with automated application of retention and review schedules (improved compliance, carbon saving). 

As a counterpoint, deploying Generative AI or Agentic AI tools in environments which are not appropriately architected, with relevant metadata, and reliable user access controls will result in slow search and extraction of data and information in response to prompts (carbon cost, efficiency cost), an increased likelihood of incorrect, misleading or hallucinated results leading to repeated prompt submissions (carbon cost, efficiency cost, decision making impact), and a risk of disclosing confidential or restricted information to unauthorised persons in response to a prompt (security risk, GDPR compliance risk). 

But this requires business leadership on data and a recognition that these are not technology projects but critical business asset management initiatives that can have a direct impact on Climate Action targets.

What Should (Public Sector) Data Leaders do? 

Imagine for a moment that you can invest in one thing that will contribute to your statutory obligation to act in a manner that is consistent with the Climate Action plan, improves the ability of your organisation to meet its obligations under GDPR and under NIS2, will improve timeliness and accuracy of data and reporting in your organisation, improves your data management maturity scores under the OGCIO’s recommended assessment framework, and helps mitigate the risk of expensive and high profile failures in the adoption of emerging technologies to improve delivery of services and internal efficiencies in your organisation? 

That thing is information asset management for data and content in your organisation. It’s not sexy or glamourous but it is the foundation on which everything else is built. It’s not a technology challenge. It is a whole-of-organisation opportunity. 

As W. Edwards Deming, the father of quality management, put it: 

“You don’t have to do this. Survival is not mandatory”. 

The High Court in Coolglass disagrees with Dr Deming. Public Sector Data Leaders do have to do this, and the rest of us should give it serious thought!

Note

A version of this Insight post appears in Public Affairs Ireland’s Review magazine

Join us at the 2025 Data Leaders’ Summit to hear an expert panel discuss the role of data management fundamentals in contributing to Net Zero goals. Details here: https://summit.castlebridge.info. Use discount code WEX15 to save 15%. Discounted accommodation available subject to availability.