Those who have attended one of my workshops on various aspects of reliability engineering and management, or have seen me present at conferences, may have heard me say, “Data is the difference between deciding and guessing.” It's a slogan I’ve used for at least 20 years. While I still believe it today, I’m modifying this slogan to replace the word “data” with the word “information.” That's because, over the past 20
years, we’ve transitioned from a condition of being “data starved” to one of being “data rich, but information-starved.”
Remember, data in and of itself has no value. Its value is only realized when it enables informed decisions, which is why we must now turn our attention to converting our data into meaningful information. But to get to information, we still require high-quality data.
Reliably operating a plant is a human activity that comes down to a series of decisions. While there's an element of luck involved, for the most part, good decisions produce favorable outcomes, and bad decisions tend to produce unfavorable outcomes with respect to productivity, safety, and environmental performance.
Extracting luck from the equation, the quality of a decision boils down to the nascent capabilities of the decision-maker and how informed he or she is. Given a choice, I’d rather live with the decisions made by a person of average capability who is well-informed than the decision of an uninformed or ill-informed genius; data and information separate deciding from guessing.
Data, the plural form of datum, is a collection of observations. These observations may be quantitative; quantitative, but relative; pseudo-quantitative (e.g., opinion-survey scales); or qualitative. They all have a place in decision-making. However, to be useful, the data must be incorporated into various descriptive or prescriptive models that can explain what has happened, what is happening, and/or predict what will happen. Predictive models are often the
most complex, but also tend to be very valuable.
Personally, I completed about 30 graduate hours in statistics and the quantitative sciences during the late 1980s and early '90s. Back then, we had the theories and the quantitative models, but simply didn’t have the computing power of today. The computer age has ushered in quite a transformation (particularly over the last 10 years.) This trend, which is sometimes referred to as the “Big Data” era, refers to the unleashing of a massive amount of processing power to solve a
broad range of very complicated problems. For example, the fast-tracked search for a COVID-19 vaccine has benefited significantly from sophisticated analytics and associated computer modeling.
The promise that Big Data holds for increasing the safety, productivity, and environmental performance of manufacturing and process plants is staggering. Still, to make informed decisions, we must have good data. Regrettably, most organizations simply lack the discipline to acquire quality data that will support better informed asset-management decisions.
The potential is there, but the structure is lacking. For example, standard taxonomies for operator notifications about a problem they’ve observed on an asset or a process don’t exist in most organizations. Many operations lack even a basic comprehensive definition of name-plate information or bills of materials. This type of sloppy data management keeps us in the dark when making decisions and keeps us
guessing.
Yes, information is the true difference between deciding and guessing, but without quality data, managing the assets in our plants remains guesswork and prevents us from fully embracing the age of “Big Data.”
dtroyer@theramreview.com
