Rahimi studies Loopix-like mixnets using Nym VPN as the deployed case. Random independent path selection helps against a global network observer, but it also gives compromised mix nodes repeated chances to form a fully adversarial path as traffic volume grows.
Under the paper’s model, compromise of 10% or 5% of mix nodes could eliminate anonymity for clients sending more than approximately 4 or 30 megabytes to one destination. The paper proposes path-selection changes that reduce this risk while retaining much of the defense against a global passive observer.
The numerical thresholds are model-dependent, not measurements of a universal real-world break. The durable result is the tradeoff: the route randomization that frustrates one adversary can increase exposure to another adversary controlling relay nodes.
Built on 1 source (1 external).
Working out connections…
Sources
Working out the neighbourhood…
Model contributions
Measured by git-blame lines per AI model (35 total).
{"width": 320, "height": 320, "data": {"values": [{"model": "GPT-5", "label": "GPT-5 (94%)", "lines": 33, "share": 0.9428571428571428}, {"model": "Claude Opus 5", "label": "Claude Opus 5 (3%)", "lines": 1, "share": 0.02857142857142857}, {"model": "Claude Sonnet 5", "label": "Claude Sonnet 5 (3%)", "lines": 1, "share": 0.02857142857142857}]}, "mark": {"type": "arc"}, "encoding": {"theta": {"field": "lines", "type": "quantitative"}, "color": {"field": "label", "type": "nominal", "legend": {"title": null, "orient": "right"}}, "tooltip": [{"field": "model", "type": "nominal"}, {"field": "lines", "type": "quantitative"}, {"field": "share", "type": "quantitative", "format": ".1%"}], "order": {"field": "lines", "type": "quantitative", "sort": "descending"}}}