Hello, we have done a worldwide measurement to identify which ASes have propagated RPKI-invalid prefixes. We find an interesting phenomenon that some ASes may perform RPKI-invalid filtering only at partial interfaces (e.g., provider interfaces, customer interfaces, and peer interfaces). Although there are other studies [1, 2] noting the same phenomenon, little is known about why some network operators perform different filtering policies at different interfaces. We are trying to figure out the reasons and summarize the optimal deployment policy for RPKI-invalid filtering. To this end, we create an anonymous questionnaire and hope you can help fill out it. Questionnaire link: https://qfreeaccountssjc1.az1.qualtrics.com/jfe/preview/previewId/70bbaf01-b4a6-4f7d-aac4-33cff5b135a3/SV_dj62hVBXcEy5ryC?Q_CHL=preview&Q_SurveyVersionID=current Reference: [1] Help Validate ROV Adoption Measurements from RoVista. https://labs.ripe.net/author/tijay-chung/help-validate-rov-adoption-measurem... [2] 85% of MANRS Members Conformant to Actions 1 and 4. https://www.manrs.org/2023/01/85-of-manrs-members-conformant/ If you have any questions, concerns, issues, or comments, please send an email to qlc19@mails.tsinghua.edu.cn Best, Lancheng