Agility & the Analytic Sandbox
View Comments: Newest First | Oldest First | Threaded View
Re: Sandbox or Petri dish or flight simulator?
  • 4/24/2014 11:04:50 AM

Lyndon,  I have not heard of the more pejurative interpretation of "sandbox" that you mention, but it appears that I might need to do a little research and be on the lookout for it.  At SAS, and the few times I've run across it in the literature, we use it to mean what you speak of at the end - simulation, or a test kitchen or the like.  A place for serious play (like you only find in four-year olds and serious adults) where you're doing data discovery, building models / scenarios, tweaking them, discarding them - often answering the question: "Tell me something I don't already know".  It would be disheartening to find out at this stage that the term as two, rather opposite meanings.

As I was thinking about this reply on the drive in, it made me remember my efforts at modeling our investment decisions while I was at Fujitsu.  It was a complex model, overly complex as I now realize, right down to modeling most of the income statement, when what I should have done was something less complex that let us "fail faster".  Because of the detail required, it might take a month to get through the six or more iterations required to take an initial proposal that came in with a ZERO IRR and play with it to find out what it might take to make it viable.  That's what I think of when I think of "sandbox" - a modeling or scenario tool/approach that would have let us work through various iterations and options all in one sitting, or at least within a week using email and webex to connect and get feedback.  I was trying to do the modeling AND the detailed planning at the same time - that's the finance guy in me, I suppose - which was largely unnecessary and unproductive at the early stages when we didn't even know if the proposal could be made viable.

Sandbox or Petri dish or flight simulator?
  • 4/23/2014 8:54:37 AM


Seems like an interesting and useful schema for guiding an organization's deployment of analytics into productive outcomes. But I'm a bit confused by the "sandbox" metaphor.

I'm used to "sandbox" meaning a closed environment where the participants with some kind of phantom "authority" discuss issues and make recommendations, but actually have no real power and are effectively ignored by the real decisionmakers.  (The metaphor evokes kids pretending to build things in a playground sandbox.) Like a "planning committee" or "community oversight committee" that's set up by the decisionmakers to keep critics occupied and gives the illusion of participation but the recommendations of which are generally disregarded or sent to another committee for "study".

If I understand Leo's concept, his "sandbox" is more like a Petri dish or flight simulator for generating and testing new approaches.