IASC: Institute for Ascertaining Scientific Consensus
STATEMENT
Regarding Ethics and Anonymisation
Votes are anonymous. Votes are not connected to a voter, nor even to a unique voting token.
IASC ensures that each targeted scientist can only vote once, by linking each vote to a unique voting token. As each token is used (a targeted participant votes), the IASC server deletes that token from its storage. If anyone tries to vote a second time, using that same token, the absence of that token from the server is used to block the attempt.
Thus when a vote takes place, the vote itself is recorded in one place (a spreadsheet), whilst the fact that a given targeted scientist has voted is recorded in another place, in effect, by the fact that the unique voting token assigned to that scientist is erased from the IASC server. Thus, during the two weeks that the survey is open (but not after), it is possible in principle to establish whether or not a given targeted scientist has voted, by ascertaining whether or not their voting token has been erased from the server.
Two points are important here: (1) it still isn’t possible to ascertain how the scientist voted, except in edge-cases such as when only one targeted scientist participates, or when all scientists vote the same way; (2) after two weeks, when the survey closes, all voting tokens are erased from the IASC server, whether they have been used or not. Thus, at the moment the survey closes, after two weeks, it becomes impossible to ascertain even if a targeted scientist participated, never mind how they voted. Again, there are edge-cases, such as when the response rate is 100%, and all scientists vote the same way. If this were to happen, we would know how every scientist voted. Such cases are vanishingly unlikely; we expect a response rate between 40% and (optimistically) 70%. A 100% response rate is unthinkable, and we include it here only because it is logically possible.
Concerning the data collected, IASC records and retains some key pieces of demographic information for research purposes. Responses are tagged with type of scientist (broadly construed), and institution, for example that the vote-caster was a chemist residing at Durham University (UK). For the purposes of this system, scientists are put into five categories: Physics, Chemistry, Biology, Earth Sciences, and Health Sciences. We balance aggregation of data with the necessity to satisfy statistical queries about voter demographics.
FOI Request
A Freedom of Information Request (FOI) can be issued to a public institution, such as a university, by anyone at any time. In theory, during the two weeks that the survey is ‘live’, an FOI request could be submitted in an attempt to find out if a scientist did/did not participate. During these two weeks (but not after), such information is, in principle, available, since a unique voting link is assigned to each targeted scientist, and the server records which links have been used by deleting them as they are used.
(N.B. Even during these two weeks there is no way to link a participant with their vote, since this information isn’t recorded by IASC. A rare edge-case exception is when all participants respond in the same way. In that case, knowing if a scientist voted is sufficient to know how they voted.)
However, various exemptions are listed in the Freedom of Information Act: for example, a request can be refused if it would disclose personal information in contravention of the UK Data Protection Act 2018, or the General Data Protection Regulation (GDPR). This must be balanced against the Public Interest Test, or PIT: an exemption can only hold if it is not in the public interest to release that data.[1] GDPR would provide protection against this, as the responses of individuals could be argued to be their personal data, which as the data controller we would be unable to release. It would be difficult to argue that it is in the public interest to reveal information about how any given individual voted, and it would likely be allowable to release summary information (demographics, etc) at an aggregate level to satisfy a FOI request.
Subpoena and Witness Summons
A subpoena is a coercive summons issued by a court requiring an individual or organisation to produce evidence. Subpoena are used in the United States to compel testimony. The 1970 Hague Evidence Convention would be the fallback for a lawyer in the USA attempting to exfiltrate data from a British company[2], using a witness summons which performs a much narrower discovery function than the US subpoena.
A witness summons is the equivalent UK process for compelling an individual to produce evidence at trial: the word ‘subpoena’ is no longer used in UK law. A witness can refuse to comply with a witness summons on the grounds that they cannot produce the requested evidence, or that they have a duty of confidentiality to any person to whom the evidence relates which outweighs the reasons for the court summons, warrant, or order.[3]
In the case of an IASC survey, an attempted subpoena or witness summons would be highly unlikely to be an effective way to extract information on survey participants and votes cast. After two weeks, when the survey closes, nobody could possibly identify whether or not a targeted scientist voted, never mind how they voted. (Data is deleted professionally by a software engineer.) And before two weeks, when the survey is still ‘live’, IASC would have a duty of confidentiality to its participants which would outweigh any reason for the court summons, warrant, or order.
Ethics and Anonymisation – FAQs
Q1. Who can see the data?
- Although the survey invitation emails are sent from various different local institutions, the data itself goes directly to the project hub (Durham, UK). The spoke representatives in the IASC hub-and-spoke network, who send the emails, cannot access any data. This includes knowledge of who has/hasn’t chosen to participate at their own institution.
Q2. What encryption method is used?
- Responses are encrypted with an https:// secure connection.
Q3. Why aren’t participants provided with an informed consent form?
- In many projects, such as this one, the choice to participate counts as (implicit) consent for one’s vote to be included in the survey data. Targeted scientists are at liberty to ignore or delete the survey email(s).
Q4. Does this project have ethics approval?
- This project has full ethics approval from Durham University (UK), equivalent to IRB in the US.
Q5. What will be done with the collected data?
- Data is collected purely for the purposes of ascertaining strength of scientific community opinion, regarding a given statement of interest.
Q6. Why aren’t targeted scientists given the opportunity to “unsubscribe”?
- Targeted scientists aren’t subscribed to anything. They will be emailed only twice during the 2022-23 academic year, with the email coming from somebody internal to their own institution.
—
Any further questions regarding ethics and anonymisation can be directed to the project leader, Peter Vickers: peter.vickers@durham.ac.uk.
—
[1] https://ico.org.uk/for-organisations/guide-to-freedom-of-information/refusing-a-request/
petervickersphilosophy.com 