[lumost]

I watched the video, and I'll share some context.

I worked in the marketing and ad-tech industries for the better part of a decade. The data linking tools that Matt Braynard used to "enhance" the data available from states are intended for marketing purposes where even a 30% accuracy is sufficient for call, mail, and digital campaigns. They are not suitable for "auditing" an election, as you'll inevitably introduce bad data. Further, linking individuals from different databases is highly error prone - there are more than 500 thousand people in the US with my first and last name. Individuals may choose different writings of their name at different times such as including or excluding Sr./Jr. or including or excluding a middle initial.

The above limitations become apparent in the below hearing on Michael's evidence in the Georgia state legislature. Individuals were clearly misslinked by name, and inferred birthday - simple errors due to removing Sr/Jr. are also present.

https://youtu.be/1U8CuyfcnLs

These linking errors are inevitable in this analysis as the state does not publicly disclose sufficient information to uniquely identify individuals for privacy reasons.