It's a longstanding measurement, yeah. And it's not actually just excluding agriculture, the definition removes a whole cluster of jobs that are especially hard

It's a longstanding measurement, yeah. And it's not actually just excluding agriculture, the definition removes a whole cluster of jobs that are especially hard to track accurately, including all of nonprofit employment and unincorporated self-employment (i.e. most of construction, handymanning, etc.) Of course, farm labor and unincorporated self-employment are heavily male, and nonprofit work is heavily female. So it's a useful statistic for accurate change-over-time data, but stories like this treating 50% as an inherently meaningful tipping point are wildly misleading.