A View of Rebel Statistics, But What Were we Using?

Vincent Granville in Data Science Central:  Data Science Has Been Using Rebel Statistics for a Long Time.  Makes some fascinating comparisons between data science and classical statistics, while fashioning a view of 'rebel statistics'.  I learned my statistics to make the industrial engineering work.  I have never been a statistics insider, academically or in business, but have always felt there was a shakiness to the whole business that depended on a few too many assumptions.

  In statistics there were always too few observations, or a special distribution was assumed to make the math come out right,  or the complex software never warned you about all those assumptions or a forecast never included probable error or correlation was quickly assumed to be causation.

Still it was the bedrock of so much of what we did, and a Phd statistician was always on call, who could always cleverly  justify the method used.  Criticizing it sounds like blasphemy.  But now has 'Data Science' changed all this?  Or do we just have a new set of assumptions and 'bigger' data?  Good read that makes you think.