Ad-blocking DNS cuts data flows, but consent opt-out can increase traffic in Android games
The claim
User-side countermeasures generally reduce data flows, yet withholding consent may cause more outgoing traffic
Published as: “Investigating Data Flows from Android Games”
So what
Users cannot rely on simply refusing consent to protect privacy; effective mitigation requires network-level blocking tools, though these may affect app functionality.
The five-minute version
Refusing consent in free-to-play Android puzzle games can increase outbound traffic rather than reduce it, while most user-side countermeasures like ad-blocking DNS services do cut data flows. The study measured 16 GB of network traffic from 50 games, finding each app contacts about 22 unique domains and sends device identifiers, timezone and battery info. Countermeasure effectiveness varies: DNS blocking lowers request counts but its impact on privacy-relevant payloads is inconsistent, and consent refusal sometimes adds traffic and new destinations.
What they found 2
- Ad-blocking DNS servers substantially reduce the number of advertising-related requests.
“Figure 1 shows that enabling the DNS server significantly decreases the average number of requests for all three applications compared to consent refusals and the base cases.”
- Each app contacts on average 22.2 unique domains per run, sending device identifiers and other personal data.
“On average, each app contacts 22.2 unique domains, 49.4 unique hosts, and 95.6 unique endpoints in a run.”
What undercuts it the paper’s own results
- All refusal scenarios increase traffic for Screw Sort Puzzle and Water SortPuz 3D.
- DNS blocking reduces overall request volume but the number of keyword matches per payload request drops for Screw Sort Puzzle, but increases for Sudoku and Water SortPuz 3D.
Stated limits 2
- Our findings should be interpreted within the limits of our methodology: the number and selection of apps target only one category to prioritize analytical depth over broad generalization.
- The string-matching-based payload search is also sensitive to and biased toward the selection of keywords and regular expressions.
What’s new here 3
- First large-scale network-level measurement of free-to-play Android puzzle games in 2026.
- Empirical evaluation of multiple user-side countermeasures (flight mode, DNS blocking, AdGuard, consent refusal) on real traffic.
- Release of a 16 GB open dataset of 500 network traces for further analysis.
Artifacts & methods 7
- mitmproxyprimitive
- LDPlayer emulatorsystem
- Android Studio virtual devicesystem
- ad-blocking DNS server (AdGuard DNS)system
- AdGuard appsystem
- keyword-based payload searchevaluation
- dataset of 16.14 GB network tracesdataset