Decision engines refer to software that can be programmed to make decisions based on a number of indentifiable factors. They have been around since at least 1965 when the IBM 360 and other, more powerful computer systems made it possible and practical to develop large-scale Management Information Systems which processed data (usually from accounting and transaction systems) and provided managers with periodic, structured reports that were useful for planning and decision-making.

By the late 1960s, mathematical modeling of decision support systems (DSS) or management decision systems became practical. In 1971, academics had already published studies on using computers and analytical models in making decisions for production planning or to help investment managers in their daily task of administering a client's stock portfolio.

From the 1970s to the 1990s, DSS methods and capabilities continued to improve as new software and applications, combined with improvements in computer hardware, came about. One major development was the concept of data warehousing; enabling DSS developers to expand the list of variables included in their models.

In recent years, DSS systems have given rise to actual decision engines – software designed to make decisions for routine tasks which follow a fixed set of rules such as the granting of credit ratings, investment management, program trading in securities, and other, mostly finance-based systems.

Decision Support SystemsDecision Engines

Decision support systems have not been phased out. They are still around and are still undergoing continuing upgrades and improvements. Newer systems can provide users with assistance in decision-making using 'fuzzy' logic rather than data crunching fixed or measurable variables. The new systems can generate options or suggestions using incomplete data.

Current DSS systems allow uses to do several things unheard of only a decade ago:

  • allow users (whether individuals or groups) to 'doodle' or 'scratch pad' ideas in decision-spaces and allow the sufficiently 'fuzzy' boundaries to go off in unexpected directions
  • deal with ill-formed or incomplete questions, allowing users to look for patterns and help them arrive at the correct decision model
  • simulations can be generated using these DSS models, adding a time element to the process. This, in turn, allows users to ask 'what-if' questions with responses or developments that can be mapped out over time and again provide additional information for reaching a decision;
  • most DSS systems apply theory-of-constraints to the outcome of simulations, allowing users to judge the feasibility of planned actions; and
  • they can help generate new rules which can be fed into a decision engine. This can help users judge whether the activity or process can be migrated into the latter, rather than continue being an element in a human decision cycle.

Developers insist that DSS systems are not decision makers or decision engines, they are support systems that can help users analyze, review and organize data needed for decision-making. They are not supposed to be substitutes for key human attributes such as analysis, insights and, most of all, intuition.