ABCDEFGHIJKLMNOPQRSTUVWXYZAAAB
1
Thank you to all who contributed to this consultation - your time and comments are very much appreciated. For any further queries please contact openenergy@icebreakerone.org
2
3
Section of documentCommentCommenter CategoryIB1 response
4
OE-SCS / OE-SBOverall I think the classes are well described and make sense. However I do wonder if these two classes below should be amalgamated into one class ‘requiring bilateral contract negotiation’. The considerations and range of outcomes for any form of bilateral data sharing negotiation are so broad that multiple sub-classes or categories could emerge to the extent that at the top level I think it would make sense to just have one class.SSEClassification changesIncorporated
5
OE-SCS / OE-SBI think bilateral contract negotiation data could be one class.Sara VaughanClassification changesIncorporated
6
OE-SCSI agree: For me, one of the main reasons for OE to exist is to share OE-SCS and OE-SP data, so I'm curious why OE-SCS and OE-SP data are not currently suitable for OE? :)Jack Kelly, OCFClassification changesOE-SCS is resolved as data classes are being collated, therefore this will be included in Phase 3. However, Open Energy will not enable the sharing of personal data in Phase 3. This is because the legal frameworks governing personal data in the UK are different to those governing non-personal data. These differences require careful consideration to ensure data-sharing is safe and compliant with the law. At present, we are prioritising development of non-personal data sharing in order to satisfy the data requirements of our use-case. Extension of the ecosystem to include personal data will be considered after Phase 3, subject to consultation.
7
OE-SAIt's not clear to me: Would access to OE-SA and/or OE-SB data be restricted via the OE Governance Platform? If so, then, in general, I must admit this table makes me a little nervous that we might accidentally reduce the quantity of open data in the wild by being over-eager to restrict access to data that really should be open. And, for data published under CC-BY-NC or ND licences, I'm not sure I see a role for OE to play for these datasets (other than to catalogue these datasets in the OE search)? These datasets already have a 'pre-emptive license'.Jack Kelly, OCFClassification changes1. Access to OE-SB data is currently moderated via bilateral contract negotation (i.e. not preemptive licensing). Open Energy governance aims to reduce time, financial and resourcing costs associated with this by opening up this category of data to preemptive licensing. 2. OE-SA data access will be moderated via the governance platform, subject to the lowest levels of access conditions. We acknowledge that some parties may find this data in the search then choose to access it independently of our ecosystem (e.g. where data requests are small or one-off). However, including this data within the Open Energy governance service offers several advantages: 1. Open Energy can carry out basic quality checks on the data (e.g. verifying provenence and API uptime) supporting better and quicker service. 2. OE-SA data providers will also be encouraged to use more standardised licensing arrangements where possible. 3. The data sensitivity class policy will be accompanied by guidance encouraging Data Providers to reduce access barriers where possible while maintaining safety. We hope that this will add weight to other existing guidance encouraging sharing of Open Data where possible.
8
Question 1b.There is a type of data (or use of data, perhaps it sits in another type of classification) which perhaps isn't noted, where it is not necessarily personal data by itself, but combined with other datasets could become personal/PII or other high risk data. It is possible that the mining of non personal data can be used to build identifiable personal profiles. Maybe this could be considered within the data classification. There may be some postcodes with a very small number of residents for example, so combining this information with aggregated postcode-level smart meter data could lead to building up a personal profile... might be a terrible example. This might be applicable for 3a) as well and may already be captured via the Personal Sensitivity columnCharles Roadnight, Zühlke Group AnonymisationIndividual re-identification is being actively considered by the Open Energy team. We have updated our guidance for Phase 3 to only permit aggregated data to be shared under OE-SB. Any aggregated data shared via the Open Energy ecosystem must have documented the aggregation method and show that this aligns with best practices for aggregation with a large enough 'n', such as those outlined by the ICO or ONS. Currently, personal data that has been anonymised using other methods will not be permitted to share under the category OE-SB. This is due to significant re-identification risks, combined with variable skill and knowledge levels across the sector. Sharing of data under different anonymisation mechanisms is not permanently ruled out, but will be subject to consultation beyond Phase 3. We appreciate that this may restrict immediate sharing of certain datasets, however we consider this approach vital to building a trusted ecosystem based on thorough consultation.
9
Question 1b.In reply to Charles: +1 on this. Usage restrictions could theoretically prevent mining ('thou shalt not attemp to identifty...') but who polices?AnonymousAnonymisationSee comment F6
10
Question 1b.In reply to Charles: +1 re potential disclosure even with aggregated data too (when cross-linked to other data on same area)AnonymousAnonymisationSee comment F6
11
Question 3. Perhaps worth noting that different 'views' of the same data might require different licenses. For example, historical data might be OE-O, but real-time data might be more commercially sensitive, and hence might require OE-S{AB}. Likewise, 'raw' data might be sensitive, but aggregated data might be OE-O. More discussion of this point here: https://docs.google.com/document/d/1ox-3a4L_cob9otaHPC7hxJqXjf3c_7UJ2BiB24WRneo/edit#heading=h.y5xw9mw5p8hyJack Kelly, OCFClassification changesThis has been noted and will be considered when developing guidance for Data Providers. In these cases, it may be that different granularities or temporalities of data are considered 'different datasets' for the purpose of classification.
12
Question 3b.As per Charles' comment would be worth looking into statistical disclosure control methods/models as aggregated data (esp with small n) can be 'unravelled' in various ways using data at the same level of gepgraphy. Suggest a conversation with ONS re Census etc data? Mandating (or providing) some sort of formal Disclosure Assessment would give data owners (& customers) trust... @dataknutBen Anderson, University of SouthamptonAnonymisationSee comment F6
13
Question 4b.Our experience (as a University) of trying to persuade data providers in this sector to release data suggests this is optimistic. They don't have the capacity nor skills to make (e.g.) disclosure assessments and are likely to default to the safe 'no share' option unless someone else provides a suable resource or they invest in own capacity. Data sharing is never a technical issue - it is a social & political issue (small p political). @dataknutBen Anderson, University of SouthamptonAnonymisationSee comment F6. Provision of resources accompanying safe sharing of anonymised data could be covered under future consultation - we would be grateful for further comment at a time when we have resource to address this.
14
Question 4c.also re OE-SB - it is rarely possible to completely anonymise data, beware over-promising - see discussion at https://www.ukdataservice.ac.uk/manage-data/legal-ethical/anonymisation.aspx and and also https://www.ukdataservice.ac.uk/manage-data/legal-ethical/consent-data-sharing.aspxAnonymousAnonymisationSee comment F6
15
Question 4c.when some form of anonynmisation is applied the data often loses its value. Can be better to retain detail & apply stronger access controls (i.e. keep data in a different class)AnonymousClassification changesIn Phase 3, we are not able to permit sharing of personal data (class OE-SP) via the Open Energy ecosystem. Extensibility to this area is subject to future consultation. We would be grateful for further comment at a time when we have resource to address this.
16
Question 4c.In response to anonymous above: Can we encourage both? i.e. provide two views of the same data: A 'raw' version provided under tight access controls; and an anonymised version, provided openly (or, at least, under less restrictive access control)?Jack Kelly, OCFClassification changesSee comment F14.
17
Example datasetsPresumably these reflect the current access restrictions on data, rather than potential or target future access restrictions?Jeni Tennison, ODIMiscYes - this column is designed to provide practical examples based on current scenarios that may help organisations understand the sensitivity classification system and judge whether our proposal is acceptable. It may need updating in future if circumstances change (e.g. new regulation to make certain datasets open).
18
OE-CAdd phrase to description: 'May be security-critical information relating to operational technology supporting critical national infrastructure owned by an Operator of Essential Services.'AnonymousBasic editsIncorporated
19
OE-C / OE-SPOE-C and OE-SP - both mention "protected" or "special characteristics" in relation to personal data which seems confusing.AnonymousPersonal DataProtected characteristics removed from OE-C
20
OE-SPShared with whom? Data being shared back with those that provide it, or to third parties they permission (ie under the data portability right) has different restrictions around it than identifiable data about lots of people being shared between businesses or to academia, for example.Jeni Tennison, ODIMiscThis will require consideration for future extensibility, however sharing personal data is not in scope for Phase 3.
21
OE-SPYes, it's worth underlining what Josh said: DCC are looking at improving access to system data, not the actual power demand data. The system data is just the 'header metadata', not the actual payload.Jack Kelly, OCFPersonal DataA significant amount of data that the DCC proposes to release is classed as personal data (thought not all). Personal data is out of scope for Phase 3 unless aggregated (see comment F6). However, please outline any relevant use-cases that could be developed beyond Phase 3 here: https://docs.google.com/spreadsheets/d/1qtXibya07HV0yC1pVEkjrQCeo1OukvzRZlOSw01wzBw/edit#gid=0
22
OE-SPFrom memory there has been work undertaken on sharing priority services register (PSR - e.g. customers with a health need for power) data between DNO's / GDN's / Water Companies so I'm not sure this should be excluded - I can take an action to check this and feedback.Rosie McGlynn, PassivPersonal DataProtected characteristics removed from OE-C.
23
OE-SPRosie is right, Ofwat and Ofgem launched a trial of this but it all seems to have gone quiet since 2018. I'm not sure whether that's because it worked or because it didn't! But it was shared in order to make it easier to serve customers in vulnerable situations, not for commercial purposes.Sara VaughanPersonal DataProtected characteristics removed from OE-C.
24
OE-SPData classifications seem to make sense – I agreed with the point that for OE-SP all currently legislated uses of energy data/personal data exemptions for public good etc should be included.AnonymousAffirms approachAmended
25
OE-SCS to me the root question is 'how many relationships need to be managed' and how hard/frictionless it is to gain access. e.g. if the market for a partial data set is 5 users, bilateral is very likely lower friction. if there is a potentially bigger market for data that is currently only shared via bilaterals, then Open Energy can reduce friction. We must assume that humans will route around any process to get things done, so if they are emailing 5 people a spreadsheet and no one 'cares' about the legality of it, they'll keep doing that. If they do care about the liability transfer and have a high-latency legal process, then OE helps.Gavin StarksMiscIB1 comment
26
OE-SARE CC BY-SA or GNU AGPLv3 - Note that these are both open licences so data published under them is open data. (Share-alike clauses are allowed under the open definition. Non-commercial licences such as CC-BY-NC or non-derivative licences are not open licences.)Jeni Tennison, ODIBasic editsTerminology updated.
27
OE-OAdd legal and regulatory data? E.g. licensing categories, institutions in charge, tax info, ...Jose CordovillaBasic editsIncorporated
28
Question 1a.Yes, I feel the categories are clear.Charles Roadnight, Zühlke Group Affirms approach
29
Question 1b.What about situations where the data provider expects payment for data, but where the data is not sensitive. For example, an energy forecasting company might want payment for their forecast. Where does that fit into the classes; and what's the mechanism for the provider to get paid? Maybe this doesn't exist inside the OE ecosystem, but instead the financial transaction is conducted just like any other bilateral financial transaction on the web (e.g. buying a book from Amazon)Jack Kelly, OCFMiscAddressed via the Advisory Group. Open Energy does not intend to be involved with how data is priced and (at least in Phase 3) does not intent to provide a payment service. Financial transactions will happen outside the Open Energy ecosystem, however the existence of a transaction or other payment agreement may form part of data access conditions considered by the Governance Platform.
30
Question 1b.I asked my question above on the OE Advisory Group meeting, and the reply was that OE don't plan to wade into setting prices for data, nor is OE a payment processing platform. So the payment will largely be done outside of OE. But OE will help by confirming the identity of the users.Jack Kelly, OCFMiscSee comment F27
31
Question 1c. have a look at https://www.ukdataservice.ac.uk/manage-data/legal-ethical/access-control ((although the ukds focus is on personal data...)AnonymousPersonal DataThis will require consideration for future extensibility beyond, however sharing personal data is not in scope for Phase 3.
32
Question 3. I'm not sure how relevant in this domain, but you might think about environmental sensitivity. There are sometimes datasets that hold data about eg sites of special scientific interest or the locations of rare / endangered animals that need to be protected.Jeni Tennison, ODIMiscThank you for the suggestion. At present we consider risks in this area as low for Open Energy's current domain. However, this may be something we consider in future if there is demand for data allied to energy to be shared. (e.g. water data may be higher risk).
33
Question 4b.Agree 100%Jack Kelly, OCF
34
Question 4b.This is exactly why 'pre-emptive' licensing is so critical — a collectively agreed framework moves us from 'no share' to 'share a bit' ... and yes, OE is not a tech project, but a culture-shift-supported-by legal, tech, etc. projectGavin StarksIB1 comment
35
Question 4c.don't underestimate the difficulty of obtaining informed consent especially if OE-SP turns out to be the most commercially useful/valuable class (any opinions on that?). Retrospective consent is usually not possible...AnonymousThis will require consideration for future extensibility beyond, however sharing personal data is not in scope for Phase 3.
36
Question 4c.How does (might) OE-SP relate to the DCC?AnonymousMiscSee comment F19
37
OE-SPobviously this is just an example but worth ensuring alignment with DCC on this as they are committed to improving access to their system data and this should be coordinated with Open Energy: https://www.smartdcc.co.uk/media/4699/21037-dcc-data-for-good-paper_v8-final.pdfJosh D'Addario, ODIMiscSee comment F19
38
OE-SB dataset examplesCouldn't a lot of this data be moved into the OE-O or OE-SA categories? 😀 Power output from large generators is already publicly available from BMRS, MicroGen DB, and PVOutput.org (and probably other places). UKPN have recently built a public dashboard which shows power output aggregated by GSP for each fuel type, and they are planning an API. (Although, AFAICT, none of these sites use standard open data licences like CC-BY-4.0 or CC0. For example, here's the BMRS licence: https://www.elexon.co.uk/operations-settlement/bsc-central-services/balancing-mechanism-reporting-agent/copyright-licence-bmrs-data/)Jack Kelly, OCFClassification changesBeyond Phase 3 Open Energy will produce guidance - complementing the MED Energy Data Best Practice Guidance once finalised after the recent consultation - regarding how Data Providers can classify their data. We hope that this encourages data sharing at the lowest levels of restriction, while also respecting the sensitivities listed. However, responsibility for data classification ultimately rests with the Data Provider therefore Open Energy is not able to mandate classification. This may change if regulatory obligations are updated in future.
39
Question 4c.How does this document handle situations in which data held by a Data Provider is not owned by them, but is important for understanding the networks? e.g. suppliers hold customer data such as smart metering but are not authorised to share this without consent. There may also be instances where consent (or other conditions) prior to sharing is required for non-personal data. Would this be the role of what MED data best practice guidance terms the 'Data Custodian' within the Data Provider's organisation? i.e. where would responsibility/liability for consent-seeking lie? Would this have to be covered internally prior to allocating a dataset to a particular data class?SSEMiscYes - it is the responsibility of the Data Provider to ensure that data is appropriate for sharing via the Open Energy ecosystem. Personal data consents are currently out of scope for Phase 3. Non-personal data consents may be determined through data licensing (e.g. conditions or resale, combination and onward sharing) and/or negotation with the data owner. However, all of this must be established prior to sharing data via the Open Energy ecosystem as liability for compliance lies with the Data Provider.
40
OE-SCSAdded this comment in the operational guidelines doc too, but I think it's worth thinking about how the governance framework can provide even just high-level support for datasets on the more sensitive side.Naaman Tammuz, BitfountMiscThese comments are about ensuring that the OE ecosystem enables various distributed computation methods in future. For consideration by the tech team. No changes implicated for data sensitivity classes.
41
OE-SCSsharing metadata about closed and less-shared data sources could be something too. which could then provide an opp for FL-style data sharing as you [Naaman] intimated.Josh D'Addario, ODIMiscThese comments are about ensuring that the OE ecosystem enables various distributed computation methods in future. For consideration by the tech team. No changes implicated for data sensitivity classes.
42
OE-SCSYes, this is exactly how our product works. I can imagine us providing an 'FL-ready' version of a dataset via OE, likewise discoverable via metadata, and accessed via our (Bitfount's) governance mechanism (which takes into account which algorithms can be run, etc) which would be made fully-compatible with OE and the license with the data owner.Naaman Tammuz, BitfountMiscThese comments are about ensuring that the OE ecosystem enables various distributed computation methods in future. For consideration by the tech team. No changes implicated for data sensitivity classes.
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100