I understand the argument that this article is trying to make. But I feel like ultimately it is a semantic argument around the use of the word science in a job title that has little to do with what people actually in those roles accomplish.
I think its important to first understand that the idea of the data scientist is somewhat reactionary. For a time in data analysis a popular title was "quantitative scientist" these people were more often mathematician or statistician focused domain experts who synthesized results out of data-sets. They did cool things with with data and this helped sell many of companies on the need to expose data for use internally.
But most "quantitative scientists" were not software minded, that is to say the upper end of the average space touched on the "citizen developer" skill level and no higher. That meant that as companies scaled and collected data the work of the "quantitative scientists" broke down and couldn't be implemented in production contexts.
What emerged over the last 5 or 6 years is a group of people who came from the other direction. They generally had a more intense focus on the manipulation of data, but they melded this with a non expert but effective use of quantitative science to get great results.
Would calling these roles "quantitative analysis data engineers" be more descriptive sure but somewhere along the line someone or ones wrote articles and coined a term that captured the imagination of the industry.
The Author attempts to put forth the idea that "Data Science" is inexorably linked with "Big Data" and that as that trend cools so will the field. Again I would think that this is a misunderstanding of the field. Most companies hiring for the field use Big Data as a code word for, we aren't sure what to do about our problem. So although many companies seem to have that as a must have when approaching a problem the truth is that it is often a recognition of lack of internal proficiency. Part of being an expert in the field is trying to help companies realize what type of data and analysis actually brings the company value.
>Science creates knowledge via controlled experiments, so a data query isn’t an experiment. An experiment suggests controlled conditions; data scientists stare at data that someone else collected, which includes any and all sample biases.
As he said himself in the next paragraph, this would also not include things like astronomy. You can't just redefine science to mean "just doing experiments." Interpreting the data is at least half the work of science. If you are a bayesian, then doing experiments is nothing more than a necessary evil of getting more data so you can do more actual science.
And since when don't data scientists do experiments? A lot of work goes into collecting data, and there are things like active learning.
The conclusion he reaches article doesn't seem well supported, and in fact I'm not even sure he knows what a Data Scientist is based on his description.
3 comments
[ 5.2 ms ] story [ 16.3 ms ] threadI think its important to first understand that the idea of the data scientist is somewhat reactionary. For a time in data analysis a popular title was "quantitative scientist" these people were more often mathematician or statistician focused domain experts who synthesized results out of data-sets. They did cool things with with data and this helped sell many of companies on the need to expose data for use internally.
But most "quantitative scientists" were not software minded, that is to say the upper end of the average space touched on the "citizen developer" skill level and no higher. That meant that as companies scaled and collected data the work of the "quantitative scientists" broke down and couldn't be implemented in production contexts.
What emerged over the last 5 or 6 years is a group of people who came from the other direction. They generally had a more intense focus on the manipulation of data, but they melded this with a non expert but effective use of quantitative science to get great results.
Would calling these roles "quantitative analysis data engineers" be more descriptive sure but somewhere along the line someone or ones wrote articles and coined a term that captured the imagination of the industry.
The Author attempts to put forth the idea that "Data Science" is inexorably linked with "Big Data" and that as that trend cools so will the field. Again I would think that this is a misunderstanding of the field. Most companies hiring for the field use Big Data as a code word for, we aren't sure what to do about our problem. So although many companies seem to have that as a must have when approaching a problem the truth is that it is often a recognition of lack of internal proficiency. Part of being an expert in the field is trying to help companies realize what type of data and analysis actually brings the company value.
As he said himself in the next paragraph, this would also not include things like astronomy. You can't just redefine science to mean "just doing experiments." Interpreting the data is at least half the work of science. If you are a bayesian, then doing experiments is nothing more than a necessary evil of getting more data so you can do more actual science.
And since when don't data scientists do experiments? A lot of work goes into collecting data, and there are things like active learning.
The conclusion he reaches article doesn't seem well supported, and in fact I'm not even sure he knows what a Data Scientist is based on his description.