Open energy data roundup

21 views
Skip to first unread message

Robbie Morrison

unread,
Apr 26, 2022, 5:23:59 AMApr 26
to openmod list

Open energy data roundup

Robbie Morrison
robbie....@posteo.de
26 April 2022
Release 01


Hello all

Here is another roundup of issues related to open energy data, this time concentrating on recent developments in Europe. Apologies in advance for the dense material.

Context

Regarding Europe, access to clearly reusable data of public interest is becoming increasingly problematic. That includes the information required to analyze development pathways that honor rapid decarbonization. The Open Data Directive 2019/1024 was a clear step backward in this regard by neutering the legal notion of re‑use (see §2.11) despite high‑flying ideals (see recital 16) and a misleading title.

The focus of this data roundup is a novel property right for machine-generated data currently being considered by the European Parliament as part of the proposed Data Act. And as I and others argue, its adoption would be a disaster for transparent and reproducible public interest analysis.

These developments contrast with incremental progress toward open energy data in the United Kingdom. And a long‑established tradition of public domain data in the United States, together with an intellectual property regime that effectively excludes routine data from protection because of insufficient human creativity.

Proposed novel property right for machine-generated data

The proposed Data Act, currently before the European Parliament, contains provisions to create a new European property right for raw machine‑generated data. This new right is similar in intent to 96/9/EC database protection but without requiring that substantial investment, substantial extraction, and human agency be present. Notwithstanding, the legislative wording remains sketchy at this juncture.

Advocacy group Open Future provides an excellent critique of this proposal and documents the underlying motives as well:

Although the publication of Tarkowski and Vogelezang (2021) predates the latest official release of the proposed Data Act on 23 February 2022, their arguments stand nonetheless (I presume the authors relied on well‑publicized leaked drafts). The proposed Data Act, in its current form, can be found as follows (and search on "machine-generated" for the points under discussion):

Felix Reda (previously Julia), a former Member of the European Parliament (MEP), warned five years ago that this novel instrument was being justified on much the same faulty logic that underpinned the 96/9/EC database rights that entered into European Union law in 1996

Tarkowski and Vogelezang (2021:7–8) specifically highlight the dangers posed to science:

[T]he new right would strongly affect freedom of expression and information as well as freedom of scientific research and services, given that it would greatly reduce overall information availability. In this light, the European legislator would have to prove that a new property right would be socially and economically justifiable for information access by citizens and researchers.

And that generating data is increasingly a joint undertaking rather than an individual undertaking (p8):

[D]ata is increasingly seen as relational and co-generated. Salomé Viljoen (2021) proposes that the relational character of data means that for any exchange of data there are collective–even population-level–interests that cannot be reduced to individual interests. A related argument is made by the GPAI Data Governance Working Group, which recognizes that data is increasingly generated collectively by several different entities. These characteristics of data imply competing interests among various actors in the data economy. Exclusive property rights can therefore easily be questioned by other parties, asking for recognition of their rights in data.

The scope of this proposed instrument would surely include the outputs from energy system models. The good news is, I guess, that modelers can waive such rights by attaching Creative Commons CC‑BY‑4.0 licenses (or a CC0‑1.0 waivers or something else inbound compatible to CC‑BY‑4.0).

But this novel instrument would also give those supplying information under statutory reporting new and solid intellectual property rights. And that would include the transmission system operators who contribute grid status information to the ENTSO-E Transparency Platform and the EEX electricity market operator who makes public market clearance information. At present, there is no real basis for IP to attach to this kind of data. And indeed some in this community treat information sourced from the ENTSO‑E Transparency Platform as legally unencumbered. But that practice could not continue with this new provision without the prospect of actionable IP infringement.

Looking at the data processing pipeline, there is no agreement on how and under what circumstances rights attached to input data can transmit to output data. The legal question is whether the resulting output constitutes a new work or a derivative work. Most of the admittedly limited analysis on this question focuses on machine learning and big data, but more conventional numerics should not be much different in this regard in legal terms. Even if a new work was indeed created, researchers cannot republish their input data to facilitate open science and analytical transparency.

In summary, this proposed novel property right for machine-generated data provides a real threat to public interest analysis because such analysis will invariably draw on encumbered or potentially encumbered information. And that would, under current circumstances, include the statutory reporting of information on grid status data and market clearance.

For completeness, there were two submissions on the proposed Data Act from this community — although the issue of machine‑generated data was not traversed in either:

Stepping back, my general reaction on these current legislative reforms that fall under the rubric of the EU Digital Single Market are as follows. With this first point being the most important:

  • there needs to be a much sharper distinction between published information of public interest and other classes of information — moreover, public interest information must be explicitly and genuinely usable and reusable

And continuing:

  • there needs to be a greater recognition that data sharing and open data are markedly different concepts with very different underlying rationales, objectives, and use‑cases
  • the proposed new intellectual property right for machine-generated data (as discussed here) should be shelved
  • the existing 96/9/EC database right should be abandoned completely … or failing that, be subject to examination, registration, and annual fees (like trademarks and patents) and also expressly excluded from all material under statutory reporting
  • definition §2.11 for "re‑use" in the misleadingly‑named Open Data Directive 2019/1024 should be reworked to align with majority perceptions on that concept
  • proactive legislative support for an information commons should be considered — indeed there are no provisions in law in Europe or elsewhere (that I am aware of) that fill the legislative void that open content, data, and software licenses have been designed to fill

I wholeheartedly commend Tarkowski and Vogelezang (2021) to anybody wishing to understand the drivers that lie behind the current suite of data reforms by the European Commission. Those drivers clearly prioritize data commodification and the development of a European‑wide data market over an information commons and public interest usage and re‑usage of important information. Moreover I predict that the current policy trajectory will provide no end of headaches for open energy system modelers.

On the upside, France is advocating for a digital commons as part of its EU presidency. This is clearly counter to the direction the European Commission is currently pursuing. More here:

The core question is how far the French government wants to push the concept? Or will it remain just window dressing?

On Wednesday 18 May 2022, I will present these general ideas to the FSFE LLW 2022 Legal and Licensing Workshop under the title: "Open data for public policy: legal and political challenges". And will see if I can circulate a recording of that presentation for wider exposure.

European Commission language

I recently watched a webinar with a representative of the European Commission Copyright Office on the panel discussion. And that person never used the phrase "data owner" and instead always used the phrase "data holder. I think that is a much more helpful terminology and one that energy system modelers should adopt:

  • data owner → data holder

Policy databases

Researchers at the Wuppertal Institute and elsewhere recently released a paper describing a database of public policy measures that can support energy sufficiency. In relation to energy system analysis, such measures can be embedded in individual scenarios in direct or implicit form — or be used to help develop policy responses based on insights that derive from model outputs. That paper:

Public engagement, energy system analysis, policy formation, and the public sphere

This work‑in‑progress blog of mine, commissioned by the OSeMOSYS team as part of the U4RIA initiative, spans public engagement and energy system modeling:

Far too much numerical energy policy analysis today fails to even touch the public sphere. Indeed, many projects simply interact with the established NGOs and believe this suffices as public engagement. I would suggest otherwise.

Open data from a United States perspective

A recent blog by open source software lawyer Heather Meeker is worth reading. Lawyers tend to favor idealized arguments and Heather's advocacy of a stronger public domain that mirrors United States legal traditions follows this pattern (although I myself do not advocate abandoning CC‑BY‑4.0 licensing):

McHardy GPL software license settlement

Not related to data, but the Netfilter v McHardy GPL software license trolling case was fully settled on 27 December 2021. McHardy, a software developer had made minor contributions to the Netfilter codebase. That nonetheless gave him legal standing and he had netted circa one million euro by issuing cease and desist notices under German copyright law to companies that had technically and trivially infringed the conditions of the associated GNU GPL license. One lawyer involved (speaking under CHR so I cannot say who) opined that European courts were now more likely to admit the "soft law" associated with consensus policies on community enforcement, which treat civil litigation as the last resort of course. More on that settlement here:

McHardy's actions would have been more considerably difficult in the United States, if only because the transactions costs related to civil litigation are substantially higher than in Germany.

Closure

The Database Directive 96/9/EC, effective in various member states from 1997 onward, introduced the protection of non‑original human‑formed databases within the European Economic Area (EEA) and also the United Kingdom. The protection of machine-generated databases — should the proposal in the Data Act proceed — will again be limited to the EEA. There is no equivalent of either property right in the United States, for example. And the United Kingdom is currently promoting the principle of "presumed open" for energy sector information — which means, I imagine, that Creative Commons CC‑BY‑4.0 licensing or something inbound‑compatible should endure in the absence of personal and commercial privacy considerations. Europe therefore seems well out‑of‑step in this regard.

I believe the European Commission, Parliament, and Council should be far more cognizant of the importance of published information of public interest and explicitly provide for its free use and re‑use. And therefore abandon this novel proposal for machine‑generated data protection, at least in this important context.

Again, sorry for the complicated material.

References

Viljoen, Salomé (2021). "A relational theory of data governance". Yale Law Journal. 131: 573. ISSN 0044‑0094.


with best wishes, Robbie

▢

OPEN LICENSE

Copyright (c) 2022 Robbie Morrison robbie....@posteo.de

SPDX-License-Identifier: CC-BY-4.0
License-Text: https://spdx.org/licenses/CC-BY-4.0.html

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International (CC‑BY‑4.0) License.

-- 
Robbie Morrison
Address: Schillerstrasse 85, 10627 Berlin, Germany
Phone: +49.30.612-87617
Reply all
Reply to author
Forward
0 new messages