Is big data all it is cracked up to be? 9 reasons the answer may be: No

Submitted on: Thu, 04.10.2014 06:27pm - Annie Eissler |

We recently ran across an interesting New York Times op-ed piece on big data by two New York University professors, Gary Marcus and Ernest Davis. While the authors agree that big data can be a valuable tool, they point out 9 things big data can’t do or isn’t very good at, summarized here:

Big data identifies correlations but it doesn’t indicate which are meaningful.

  1. It is rare when big data can completely replace scientific inquiry.
  2. Tools based on big data can easily be “gamed.”
  3. When not gamed, big data analysis results are often less impactful than what they appear to be.
  4. When the source of big data analysis is big data itself (e.g., the web), errors can be reinforced.
  5. Statistically significant correlations can be found even when there is no real meaningful connection.
  6. Even imprecise questions can lead to very scientific sounding solutions.
  7. Big data is better at analyzing common things vs uncommon things.
  8. Innovations from big data (so far!) are nowhere near as significant as those of the 19th and 20th century (e.g., cars and planes).

The point of all this? As the authors conclude:
“Big data is here to stay, as it should be.  But let’s be realistic: It’s an important resource for anyone analyzing data, not a silver bullet.”

Read the op ed piece and let us know what you think.