Subscribe to DSC Newsletter

There are a number of movies that I consider much more suitable for television than theatre: Alien Versus Predator; Doom; Snakes on a Plane; The Cave; The Colony. The stories of these movies play out in environmentally-limited sometimes enclosed settings. In the theatre, I considered Alien Versus Predator the worst movie I ever had to sit through in my entire life, second perhaps only to the original Tron. But at home on a 36-inch display, it has become one of my all-time favourites. On the other hand, I think that a person is rather deprived watching Lord of the Rings or a Marvel movie such as the Avengers on a small screen. In these movies, the scenery or setting represents a significant aspect of the entertainment experience. Indeed, the setting is sometimes an important part of the story. There could be a long journey. Or perhaps there are huge crowds and all sorts of buildings. Superman wouldn't be able to fly without space to do so. Likewise in business, the setting is sometimes an important part of the story; it represents a challenge that has to be confronted or overcome. In some production environments, the narrative is mostly "inside": how to organize and motivate people to achieve specific outcomes. The idea of "thinking big" or "thinking outside the box" makes sense as far as discussion points go; but at times the problem is right inside the box. The question is really about the decision frame: the strategic use of data to capture important details while dismissing the less critical points - a priori, not knowing what will make or break the fortunes of the business.

How much of the narrative is inside or enclosed - how much of it outside or open? I think for business people, the idea of making use of big data has really been about having a panoramic or unfettered decision frame; there is some desire to catch every possible business concern whether small or big. Although I can't speak for the entire data science community, I believe that some data scientists remain focused on a small decision frame. As I have noted, a small frame might be precisely what is needed. Returning to my analogy, having a small television screen doesn't actually mean that the data is limited in quantity: in Doom, there are endless pathways - full of mutated people. But as one walks through this scenery, more pathways containing mutants open up. The pathways can go on forever although the elements of the story remain fairly static. A data scientist - or let's call this person a statistician in this case for the sake of clarity - might want more data; but it is more of the same data. I doubt that many businesses would be well-served by such a static decision frame unless of course operations never change.

While big data "might" eliminate the need for a decision frame, the ensuing decision must itself be framed. Even if an organization could routinely sift through millions of variables each day and research all of its data, coming to a decision requires a frame. Will the cost of postage stamps in the shipping department affect the decision to hire a new person for the accounting department? Well, if the frame is improperly constructed in relation to data gathering, there would be inadequate data to render a proper decision. This is not to say there would necessarily be insufficient data. There might be a lot of data, but little of it is applicable. On the other hand, there could be a great deal of applicable data, but the frame is not constructed to enable the rendering of desirable decisions. I find it humorous the concept of pumping a system full of data hoping for fabulous decisions - I guess because it reminds me of Reagan's supply-side economics. There is some hope that the best data should brings about the best outcomes. Is it really a done-deal: manna enters a system, and milk and honey logically flows out? I can imagine terrific data nonetheless resulting in miserable decisions. This is why I call this blog The Decision "Frame" - underlining the framing of decisions rather than the making of them.

An operation sometimes faces limited decision options. For example, the U.S. central bank typically has three options to consider: raising interest rates, lowering the rates, or keeping them the same. Many manufacturers face similar decision options on any product: making more, making less, or making about the same. Due to the simplicity of the decision, I feel that it can be tempting to limit the data aspects in the decision frame. It is possible to play a big movie on a small screen. However, the finer details would likely be missed. One takes only the main points from the storyline. Conversely, playing a movie depicting an enclosed environment on a big screen seems like it might be a boring movie. If the movie is about a guy buried alive in a coffin, that is pretty much what I expect to see on the big screen. Consequently, it seems desirable when there are limited outcomes to stick to a small decision frame. But then the outcomes are indeed limited. There isn't a single rule-of-thumb or particular line of reasoning that I can offer as guidance. There is a difference between making limited decisions and having. If I "have" limited decisions, then it is only a limited decision that can be made. The decision frame therefore enables or disables business strategies. Sometimes, having more options is well worth the trouble of maintaining a wider frame.

Because a coherent decision must be framed, the perception of clarity through abundance of data is not entirely valid. Quantity does not mitigate the need for strategy. Enormity of data might be useful. However, a decision can be rendered even on the total absence of data. The decision frame (on the input of data) allows for the framing of decisions (on the output); this sets the context in which decisions are made, but it does not set the decisions. I have therefore in this blog identified three components in decisions perhaps for a computer program intended to make decisions: 1) the framing for data input; 2) the use of data to frame the output; and 3) framing the decision from the output. Note my rather repetitive use of a particular term. It might seem academic to discuss the use of data in this a manner when even dice can be used to make decisions. For a large organization making coherent decisions over a period of time, perhaps over different teams of people, I believe the differences matter. What data should be collected (#1)? To what data should the company attach what amount of relevance (#2)? What aspects of the data should influence what aspects of which decisions (#3)? The data doesn't just appear out of nowhere already pointing to specific conclusions. Consequently, a great deal of thought tends to be necessary to reach decisions.

With a decision frame, it is possible to assess the possible value of a decision even before the decision is made. I would argue that the construction of decisions is often mostly about the framing. For some people that want to decorate their homes, they seek out a particular type of print or painting to fit in a specific frame to occupy a particular setting. There is a basic understanding of what will fit and why. This is a step forward from having no idea what fits and why. Having data, having data that fits, and making good decisions from that data are separate issues. Issues of framing are pervasive; but these can lead an organization towards its decisions. Sadly there is no magic leap to transport a person from empty void to convincing decisions. Even bad decisions leave its residual framing; and the supporting data hardly seems to matter after the fact. For this reason, I question the value of data collected by people that don't have the foggiest idea how to make use of it. The decision-making ability precedes the use of data to make decisions. I expect the question of how to make decisions using big data to be a major conversation going forward; and framing will likely be its cornerstone.

Views: 402


You need to be a member of Data Science Central to add comments!

Join Data Science Central


  • Add Videos
  • View All

© 2019   Data Science Central ®   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service