Friday, July 29, 2016

Data Science, Coding, the Automation Paradox, and the Silicon Valley State

My July post @All Analytics.

Indeed, if what Linus Torvald is admiringly quoted as saying is true -- "Don’t ever make the mistake [of thinking] that you can design something better than what you get from ruthless massively parallel trial-and-error with a feedback cycle. That’s giving your intelligence much too much credit." -- why bother with education, science, and theory at all? Just “plug your coding skills” by getting a Microsoft or IBM “Professional Degree”, or even take a free coding course by your favorite billionaire.

Read it all. (Please comment there, not here) 


  1. Can't seem to comment there unfortunately. I enjoy reading your "contrarian" viewpoints. Especially this article. I work at a large cloud provider with extensive ML/data science offerings. The common refrain is: "how do I break into this industry?" I often reply with very similar responses as your remarks. I see people run a few algos on tiny data sets that show beautiful ROC curves and they have no clue in the myriad of errors in their thinking. I, on the other hand, have enough self-actualization to realize that I can do the mechanics of an experiment but I would never trust it because I know I'm not fully-grounded in the underlying math theory.

  2. Can you describe the problem you have commenting? I will pass it to my editor.

    I will try to post your comment and respond to it there.


View My Stats