Participant-Mediated Collection of Sensitive Digital Trace Data: The CANDOR Research Infrastructure

Participant-Mediated Collection of Sensitive Digital Trace Data: The CANDOR Research Infrastructure

Source: cs.CY updates on arXiv.org — https://arxiv.org/abs/2609.17722 Date read: 2026-09-22 Connected to: none Kind: meta Escalation: store-only

What this is

A methodological infrastructure paper describing CANDOR, a system for collecting sensitive digital trace data through participant-mediated donation rather than platform API access or public observation. The work responds to API access constraints and proposes a research protocol where individuals curate and contribute portions of their own digital histories to studies.

What I took from it

This is primarily a tool/infrastructure contribution rather than a theoretical argument. It addresses a genuine empirical constraint (API access decline) with a procedural workaround, but does not advance a sustained claim about how protocolized systems behave under pressure. The implicit assumption—that individuals can reliably and meaningfully self-select their own data for research purposes—is itself worth epistemological scrutiny in the context of our current work, but that scrutiny is not performed here.

The paper documents a shift in research access method rather than explaining a law of protocol behavior. It is a response to institutional friction, not an analysis of it.

Research connections

• none

Method note

This paper reveals a significant blind spot: as formal data sources (APIs, public records) ossify and restrict access, research infrastructure shifts toward participant self-curation and consent-mediated donation. This creates a new layer of protocol around data access itself—one where the participant becomes both data custodian and researcher proxy. Future research on protocol ossification (L-001) and formalization ratchets (L-003) should examine whether researcher adaptation to API scarcity introduces systematic bias in what kinds of digital behavior become observable. The reliance on participant agency and self-selection also warrants investigation under L-010 (adoption nonmonotonicity)—who participates in data donation, and what selection effects emerge?