Should we consider that foundational knowledge to include broader tech knowledge like principles of software engineering, computer networking, etc.? Where do we draw the line?
Great question, I think it comes down to what it is being professionalised for. The need that I came from (I've since spoken to others who argue for other needs) is for data be professionalised from a business/management perspective - a bit like CFA and CIMA.
From that perspective you need a foundational knowledge or awareness of any topic is needed if you're trying to plan or manage how you get direct value from data i.e. doing business analytics or creating data products that deliver repeat value.
On the examples you mention - software engineering would be in as knowledge of coding and code management are important to managing successful data products but computer networking would be out as you don't need to know how that works to successfully plan or manage a data team or product.
Thanks for the response Benny. I agree with you for the most part, especially for data products that are analytics, reporting, dashboards, etc.
When we approach it from the engineering side, we may find it difficult to draw a line around what should be included for foundational knowledge.
Only reason why I'm playing a bit of devil's advocate is that I'm currently in the process of deploying ML models to production, and there are a wide variety of skills and knowledge needed to make this work that seem basic to the average developer. But maybe this wouldn't be considered foundational data knowledge.
Fascinating idea. I’m another one who’s ended up in data by interest and accident, and we have data science people in our org whose training is very mixed - analysis, signal processing, stats, operations, physics, computer science etc. Data management has some background knowledge through DAMA and the DMBOK but that’s more an accretion of practice than foundational principles. What’s your view on that?
While I fully agree with your argument here, Benny, especially the points you make about regulation and data ethics, I believe it's possible to push the professionalization pendulum too far in the other direction. Professional licensing, at least in the U.S., is all over the board, and as a 2015 discussion paper published by the Brookings Institution noted, "by making it more difficult to enter an occupation, licensing can affect employment in licensed occupations, wages of licensed workers, the prices for their services, and worker economic opportunity more broadly. Indeed, economic studies have demonstrated far more cases where occupational licensing has reduced employment and increased prices and wages of licensed workers than where it has improved the quality and safety of services (Kleiner 2013). These studies have shown, for example, that more-difficult requirements to earn a dental license (in the form of the pass rate of the required exam) do not lead to improved dental outcomes of patients but do result in higher prices of basic dental services, likely because the requirements result in fewer dentists (Kleiner and Kudrle 2000). Similarly, more-stringent licensing of mortgage brokers has no influence on the number of foreclosures, but does lead to higher prices of mortgages, again likely due to fewer providers of the service (Kleiner and Todd 2009)."
You've got me thinking Benny! I'll drop you an email with some detailed musings. I would argue data is a profession - it's just not regulated. But I agree it would benefit from foundational standards, which could be a tough nut to crack effectively!
Great article and great podcast with Joe Reis!
Should we consider that foundational knowledge to include broader tech knowledge like principles of software engineering, computer networking, etc.? Where do we draw the line?
Great question, I think it comes down to what it is being professionalised for. The need that I came from (I've since spoken to others who argue for other needs) is for data be professionalised from a business/management perspective - a bit like CFA and CIMA.
From that perspective you need a foundational knowledge or awareness of any topic is needed if you're trying to plan or manage how you get direct value from data i.e. doing business analytics or creating data products that deliver repeat value.
On the examples you mention - software engineering would be in as knowledge of coding and code management are important to managing successful data products but computer networking would be out as you don't need to know how that works to successfully plan or manage a data team or product.
Thoughts?
Thanks for the response Benny. I agree with you for the most part, especially for data products that are analytics, reporting, dashboards, etc.
When we approach it from the engineering side, we may find it difficult to draw a line around what should be included for foundational knowledge.
Only reason why I'm playing a bit of devil's advocate is that I'm currently in the process of deploying ML models to production, and there are a wide variety of skills and knowledge needed to make this work that seem basic to the average developer. But maybe this wouldn't be considered foundational data knowledge.
Fascinating idea. I’m another one who’s ended up in data by interest and accident, and we have data science people in our org whose training is very mixed - analysis, signal processing, stats, operations, physics, computer science etc. Data management has some background knowledge through DAMA and the DMBOK but that’s more an accretion of practice than foundational principles. What’s your view on that?
While I fully agree with your argument here, Benny, especially the points you make about regulation and data ethics, I believe it's possible to push the professionalization pendulum too far in the other direction. Professional licensing, at least in the U.S., is all over the board, and as a 2015 discussion paper published by the Brookings Institution noted, "by making it more difficult to enter an occupation, licensing can affect employment in licensed occupations, wages of licensed workers, the prices for their services, and worker economic opportunity more broadly. Indeed, economic studies have demonstrated far more cases where occupational licensing has reduced employment and increased prices and wages of licensed workers than where it has improved the quality and safety of services (Kleiner 2013). These studies have shown, for example, that more-difficult requirements to earn a dental license (in the form of the pass rate of the required exam) do not lead to improved dental outcomes of patients but do result in higher prices of basic dental services, likely because the requirements result in fewer dentists (Kleiner and Kudrle 2000). Similarly, more-stringent licensing of mortgage brokers has no influence on the number of foreclosures, but does lead to higher prices of mortgages, again likely due to fewer providers of the service (Kleiner and Todd 2009)."
You've got me thinking Benny! I'll drop you an email with some detailed musings. I would argue data is a profession - it's just not regulated. But I agree it would benefit from foundational standards, which could be a tough nut to crack effectively!