The Distinction Between Open Problems and Closed Problems
When attempting to categorise the problems encountered, it can be useful to distinguish between two types of problems:
- Open problems
- Closed problems
Definition Elements
Closed Problems
This distinction originates in mathematics, a discipline in which certain problems have a known and proven solution. Such problems are then referred to as closed problems.
In a business context, this type of problem can be resolved relatively easily within a company or an IT project provided that:
- the problem has been identified,
- there is knowledge of the existence of a solution, and
- there are sufficient resources and the will to implement this solution.
These three points are important and, sometimes, the inability to recognise that something is indeed a problem becomes… a problem in itself.
Open Problems
Conversely, open problems are those for which no solution is known. For some, it has even been proven that no solution is possible.
The Utility of the Distinction in IT Projects within Companies
Most IT projects today face closed problems, and the technical solutions have already been invented. Often, these solutions are not known by all members of the team. A lack of awareness of the solutions is another problem in itself.
A few examples of closed problems:
- Completely and automatically recreating an environment, from development to production, on a cloud provider’s infrastructure. The solution is to use infrastructure-as-code tools such as Terraform or Pulumi.
- Preventing the deployment into production of an application that exhibits explicitly described anomalies. The solution is to automate the detection of these anomalies, to halt code integration and thus ultimately the deployment, following a defect-stopping approach (build pipelines, in a CI/CD methodology).
- Avoiding the multi-year development of applications that may fail to deliver sufficient value to their end users. The solution is to adopt an agile software development approach, whereby new versions of software are presented to users as frequently as possible in order to validate or invalidate certain hypotheses and adjust course in subsequent iterations.
Some problems are open, and it has not yet been demonstrated that a solution is possible. The development of general artificial intelligence (AGI), for example, is an open problem.