I found the FCC Consumer Complaints dataset through the allowlisted Data.gov catalog and opened the official FCC source and schema. Data.gov identifies the public dataset as U.S. government work. The source describes individual informal consumer complaints beginning October 31, 2014 and states that the information was selected by consumers and that the FCC does not verify the facts alleged in the complaints.
I queried complete calendar years 2015 through 2025 on October 31, 2026. The source metadata said the dataset had been updated at 05:22:41 UTC that day. The fetcher requests only four aggregates: year by topic, month by topic, year by topic and issue, and year by recorded state and topic. It never requests ticket identifiers, phone numbers, ZIP codes, locations, free text, or individual rows. Together with the official dataset and column metadata, the six pinned responses contain 5,701 aggregate rows in 484,597 bytes. Their SHA-256 values and byte counts are fixed in the reducer.
The reducer validates the source name, dataset identifier, update time, required columns, topic set, eleven complete years, 132 complete months, 56 recorded place codes, and 3,344,699 total reports. It checks that monthly and issue totals reproduce each annual topic count. Repeated runs produced the same 191,578-byte browser snapshot with SHA-256 `8868835eb1dd80d1751b7bfac209489d49e73bccf7df08cbbb7d02785f67a241`. The browser makes no request to Data.gov or the FCC.
The source schema says that, beginning October 23, 2016, the separate Robocalls and Telemarketing categories were collected as Unwanted Calls. I combine those three labels into one documented unwanted-calls series. I leave other issue categories as published, apart from small plain-language expansions. A missing issue row can mean no reports or that the label was not offered. Topic and issue definitions, form design, reporting behavior, media attention, service use, and data processing can change over time.
The place view includes reports with one of the 50 state codes, the District of Columbia, or five territory codes. It divides a selected topic count by all recorded reports from that place in the selected year. This reduces the visual dominance of large states but does not adjust for population, number of customers, service availability, internet access, or likelihood of reporting. The data cannot establish the number of affected people, frequency or severity of a problem, whether an allegation is true, company performance, FCC response quality, or why any count changed.