1. Forum
  2. Usenet
  3. LINUX.DEBIAN.PROJECT

  • Re: Fresh thread - 2023 Debian Project survey: Sustainability

    From Joerg Jaspert@21:1/5 to Mathieu O'Neil on Thu Nov 16 23:00:01 2023
    On 17049 March 1977, Mathieu O'Neil wrote:

    • Survey responses are anonymous, IP and HTTP information are not
    logged, and all questions are optional. As it is still likely possible
    to determine who a respondent is based on their answers, results will
    only be distributed in aggregate form, in a way that does not allow de-anonymisation.

    That "likely" is better worded as "for a great number of people
    anonymity will not exist for anyone seeing the real answers".
    Also, that aggregate form would need to somehow mangle the following 3 questions up pretty good, I think.

    A03 formal status
    A04 which 3 teams most involved
    A05/A05a which other foss projects involved in

    Those are enough to ditch about any anonymity I think, combine it
    with B01c to make it easy enough for even an AI to find out who the
    submitter is. I might be an extreme example, but for me it would be
    enough to truthfully answer A04 (archive, new-maintainer, salsa) and
    confirm it with A05a putting OFTC...

    B01c only loses when someone works for a real big company.

    --
    bye, Joerg

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Joerg Jaspert@21:1/5 to Mathieu O'Neil on Fri Nov 17 15:30:01 2023
    On 17050 March 1977, Mathieu O'Neil wrote:

    That "likely" is better worded as "for a great number of people
    anonymity will not exist for anyone seeing the real answers".
    Also, that aggregate form would need to somehow mangle the following 3 >>questions up pretty good, I think.

    A03 formal status
    A04 which 3 teams most involved
    A05/A05a which other foss projects involved in

    Those are enough to ditch about any anonymity I think, combine it
    with B01c to make it easy enough for even an AI to find out who the >>submitter is. I might be an extreme example, but for me it would be
    enough to truthfully answer A04 (archive, new-maintainer, salsa) and >>confirm it with A05a putting OFTC..."

    All response collected with LimeSurvey will be downloaded to a password-protected device in the University of Canberra and the data
    on LimeSurvey will be deleted after downloading. All the data analysis
    will be done locally, no cloud computing will be applied to prevent AI
    from obtaining the data. To better prevent reidentification, the raw
    data will not be shared to the Debian developers in our research team.
    Only aggregated results will be shared with the whole team for interpretation.

    Are you an AI, or used one to write this? :)
    Me mentioning AI was just a sidekick on how bad AI is, but the general
    point is the *ease* of identifying people answering the question set and
    that pretending it will be anonymous is a nice wish only.

    B01c only loses when someone works for a real big company.
    The reason we kept this question from the 2016 Survey is to
    investigate how firms’ engagement in Debian has changed over time, but
    all the questions in the survey are voluntary, if you feel
    uncomfortable answering any question feel free to skip it.
    We could add a sentence in this question like “If you feel
    uncomfortable to answer it, feel free to skip it and move on to the
    next one”. Would that help?

    I think the whole questions I mentioned should have something above that
    the combination of the answers may make it really easy to identify the
    person answering.

    --
    bye, Joerg

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)