One weird thing in California

About 25 years ago, I reached over to a table crowded with earplugs. It was late, and I needed ear protection against California’s 101 freeway outside my window. One of the earplugs was looking right back at me. My young daughter had drawn a happy face on it. It was unexpected, just like a few things recently found in California’s District 28 voter rolls.

You can read my full paper on the God’s Five Stones website, or my Research Gate page.

The California report turned out better than expected, though it didn't look promising at first. The initial data covered just one district rather than a full county or state, and was missing crucial information - birthdates and registration dates. I figured I'd finish quickly but wouldn't have enough to show for it. When I reached out to the LAGOP representative who'd provided the data, I asked if they could get me those missing details.

While waiting, I decided to work with what I had. I loaded the data into my database, even though searching for duplicate voters by name and age felt crude compared to using birth dates like in New York, or contact details like Wisconsin's phone numbers and emails. But I was eager to see what patterns might emerge.

The initial search revealed nearly 5,000 matching records based on names and ages - meaning about 2,500 were duplicates. For a single district, this was significant. Extrapolating to the state level using New York's proportions, these 2,500 duplicates could suggest over a million statewide cases. But matching by age alone, rather than exact birth dates, risked false positives.

On November 15, 2024, I received updated data that included birth dates and registration dates. Running the search again with exact birth dates yielded just 10 matches across 5 names - suggesting my initial matches were mostly false positives. Except for one weird thing.

What I found defied probability: nearly 400 names appeared exactly twice in the database of 400,000 records, each pair sharing the same age but showing different birth dates. This pattern made my hair stand on end - statistically, it was virtually impossible to occur by chance.

To put this in perspective: Nassau County's data showed almost identical numbers of duplicate records whether matched by name and age (334) or by name and birth date (332). In fact, all but one of the name-age matches were confirmed by birth dates - a 99% confirmation rate. But in California's District 28, we saw the exact opposite pattern.

This pattern suggests an algorithm was deliberately randomizing birth dates to mask duplicate records, while carefully preserving ages. The complexity is revealing - maintaining matching ages between original and clone records requires sophisticated logic. For instance, if an original record shows a June 1967 birth date, the algorithm could assign the clone either a June 1966 or June 1968 birth date while preserving the same age in years. But this raises a crucial question: why go to such lengths?

Another anomaly emerged with registration dates - a pattern first spotted by Jeff O’Donnell in Arizona that I found mirrored in California. In just five days between the database snapshots (November 10-15, 2024), over 60,000 new registrations appeared. But their listed registration dates were distributed across more than 120 years of history, creating a seamless blend with the existing registration pattern.

For this to work, the age of the registrants had to be controlled in tandem with the registration date, so voters wouldn’t be too young to have plausibly or legally registered.

And of course, they had an ID number algorithm also. Take a look at the full report for the details. My opinion, this one is important because it strongly suggests three separate algorithms used in concert to achieve the same goal: obfuscation of attributes connected to suspicious records.