Third-party infrastructure of the ten most visited Polish news websites: measurement data and analysis code

This deposit is anonymised for double-blind peer review. Author, affiliation and funding metadata will be added once the review process concludes. The DOI and the files will not change. Measurement data, code and derived results for a study of the third-party infrastructure of the ten most visited Polish news websites. Each site was crawled six times with an instrumented browser, capturing every network request at protocol level together with its initiator stack, and distinguishing three states of the same page: before the consent dialogue is answered, after it is accepted, and after simulated scrolling. Sixty crawls, 41,135 network requests. Contents. The full pipeline (crawler, enrichment, classification, analysis, content-based library scan), 60 raw captures plus 61 earlier captures taken without scrolling, enriched and classified request tables, all derived result tables, and the figures as published. A README, a data dictionary documenting every column and every caveat, and SHA-256 checksums for every file are included. Reproducibility. The crawling and enrichment steps reach the live web and will not reproduce the deposited captures, because the advertising ecosystem rotates suppliers between page loads, which is itself one of the findings. Every subsequent step is deterministic and reproduces all reported numbers and figures from the deposited raw captures. Personal data. None. The measurement observes publicly accessible websites from a laboratory browser and records the traffic that browser generates. No human participants, no accounts, no circumvention of access controls. Captured addresses belong to servers; response headers were checked and contain no client-side addresses. Licensing. Data and figures produced by this study are released under CC BY 4.0. The analysis code is released under the MIT licence (see LICENSE-CODE.txt inside the archive). External classification lists (Disconnect, Retire.js) retain their own licences, documented in NOTICE-third-party.md.