I'm convinced that the best way to teach stats is to start people with data and graphing. The formal mathematics of statistics is certainly a fascinating discipline (my grandfather was a professor of statistics) but for the rest of us, the formulas are a work-around to the problem of dealing directly with the data and doing repetitive calculations by hand.
The formulas and proofs are certainly valuable, but could come later. And the graphs remain useful (at least for a hack like me) for confirming our understanding of the formulas.
Also, data and graphing would be a way to ease students into programming.
I agree, but have to caution that the abstract and dry stuff is still part of math. It was what turned me on to math. I'd favor a blending of math and science curricula, and use more data analysis when teaching science.
It's a dilemma because math is so broad, and has both pure and applied sides. And a challenge for changing curricula is that the public defines "math" as the precise sequence of topics that they studied in school (and hated).
The formulas and proofs are certainly valuable, but could come later. And the graphs remain useful (at least for a hack like me) for confirming our understanding of the formulas.
Also, data and graphing would be a way to ease students into programming.