Assessment of In-Election Machine Logs: Ballot Scan Checksums vs. Official Report

Chicago Board of Elections — General Elections 2020 & 2024

Prepared by: Benford Bench Analysis Project
Data Source: Chicago Board of Elections — Basic Election Checksums
Report Date: 2026-05-27
Elections Covered: November 3, 2020 · November 5, 2024

1. Executive Summary

This report assesses Election Day machine log data for ballot-scanning devices deployed in Chicago, Illinois during the 2020 and 2024 General Elections. The machine logs record every scan attempt — both successful and failed — and serve as an internal checksum against the vote totals published in the official Chicago Board of Elections (CBOE) canvass reports. Significant divergences are identified in both election cycles, with the 2024 election exhibiting materially larger absolute discrepancies and a substantially higher failed-scan rate than 2020.

2. Data Definitions

Methodology note (per source): A subset of scanning devices had incorrect internal date/clock settings at the time of logging. Records falling outside Election Day bounds were identified and flagged; their inclusion or exclusion affects total scan counts and is noted where relevant.

3. Raw Machine-Log Data

3.1 — November 3, 2020 General Election

3.2 — November 5, 2024 General Election

4. Checksum Analysis

4.1 — 2020 Discrepancy Breakdown

The 2020 machine logs record 669,485 total scan events against a certified total of 288,593 — a ratio of 2.32:1. Even when restricting the comparison to successful scans only (629,231), the machine logs exceed the certified count by 340,638 ballots (+118.0 %).

4.2 — 2024 Discrepancy Breakdown

The 2024 machine logs record 1,102,469 total scan events against a certified total of 458,144 — a ratio of 2.41:1. Successful scans alone (862,726) exceed the certified count by 404,582 ballots (+88.3 %).

5. Year-over-Year Comparison

Key observations:

1. Failed scan rate increased 3.6× between cycles — from 6.0 % in 2020 to 21.8 % in 2024. This is the single most anomalous trend in the dataset.
2. Total machine scans grew at 1.10× the rate of certified totals (64.7 % vs. 58.7 %), meaning the gap between machine activity and the official record widened in absolute and proportional terms.
3. Both elections show total machine scans approximately 2.3–2.4× the official certified count, suggesting a systemic pattern rather than an isolated anomaly in either year.

6. Possible Explanatory Factors

The following are documented or commonly cited technical causes that election administrators and auditors typically investigate when machine-log totals exceed certified counts. They are listed as investigative hypotheses, not conclusions.

7. Anomaly Assessment: Failed Scan Spike (2024)

The 2024 failed-scan total of 239,743 warrants specific attention:

• It represents a 495.6 % increase over the 2020 figure of 40,254.
• Failed scans constitute 21.75 % of all 2024 scan events — a rate that, if attributable solely to mechanical rejection, would imply severe and widespread equipment problems across all Chicago precincts.
• For context, industry-standard ballot-scanner failure rates are generally expected to remain below 2–3 % in properly maintained equipment.
• At 21.75 %, every 1-in-5 scan attempts in 2024 registered as a failure — a figure that should have triggered precinct-level incident reports and poll-watcher observations in real time.

Implication: Either (a) equipment malfunctions were substantially more frequent in 2024 than 2020, or (b) a portion of "failed" log entries represent re-scans, test scans, or logging artefacts rather than genuine rejections of distinct physical ballots.

8. Checksum Integrity Verdict

In neither election cycle does the machine-log ballot scan count reconcile with the certified official report within any reasonable margin of error (±1–2 %). Both cycles show the machine logs recording more than twice the number of certified ballots.

9. Recommended Next Steps (someone else)

0. Independently verify — via audit of system logs to verify consensus.
1. Request device-level logs — Obtain per-machine event logs from the CBOE to identify whether the overcount is concentrated in specific precincts or scanner models.
2. Audit clock/timestamp records — Cross-reference scanner timestamps against official poll-open/close times to isolate and quantify records with incorrect date settings.
3. Review L&A test purge records — Confirm that Logic & Accuracy test ballot feeds were fully zeroed before Election Day for all deployed units.
4. Obtain incident/exception reports — Match the 2024 failed-scan geography against precinct-level poll-watcher and inspector incident reports.
5. FOIA request for certified audit trail — Request the full ballot accounting reconciliation (initial count, provisional count, spoiled ballots, L&A test counts) from CBOE under the Illinois Freedom of Information Act.

10. Data Source & Limitations

• Source URL: https://shavidica.cc/page/Projects/Benford-Bench/Elections/basic-election-checksums
• Data as of: 2026-05-27
• Scope: Chicago, IL (Cook County) — Election Day machine-scan logs only
• Not included: Early vote logs, mail-in ballot scan logs, precinct-level breakdowns, scanner make/model inventory
• Disclaimer: This report is an independent quantitative assessment of publicly cited figures. It does not constitute legal, forensic, or official election audit findings. All conclusions are preliminary pending document-level verification from the Chicago Board of Elections. Independent reassessment of these findings is encouraged. Original FOIA released data can be accessed: https://foia.amfile.org/browse/chicago-board-of-elections

End of Report