The term predictive is deceptive.
In the Disciplined Agile (DA) tool kit we use the term traditional or serial, rather than predictive or waterfall, to refer to the classic/linear ways of working. We feel that predictive is deceptive, more on this in a minute, and waterfall to be insulting (albeit still in common use within the IT community). Furthermore, we're starting to move away from using traditional as we're now seeing a generation of practitioners who feel that some of the older agile approaches, in particular Scrum, are traditional ways of working.
There are several reasons for why we feel the term "predictive" to be deceptive:
- "Predictive" implies predictable. Predictive is defined as "relating to the ability to predict" whereas predictable is "something that happens in a way or at a time that you know about before it happens." Something that is predictable is a sure thing, yet something that is predictive is not. This is an important difference, particularly given that we know that projects aren't completely predictable - otherwise we wouldn't need risk management.
- "Predictive" approaches to IT projects are a poor choice in most cases. Years ago I led a study for Dr. Dobb's Journal that investigated the effectiveness of different approaches (agile, lean, iterative, ad hoc, and traditional) to IT projects. We found that traditional strategies were less effective in practice than agile and lean approaches, and we weren't the only ones to have found this. We also investigated what was initially predicted at the beginning of the project and what actually happened by the end of the project, and once again traditional approaches didn't do as well as agile & lean. BUT, I must stress that the study focus was on IT projects only, not on projects in general.
- "Predictive" approaches to intangible projects are likely a poor choice DDJ found, in several studies in fact, that "predictive" strategies were less predictable in practice than agile/lean approaches in IT. I highly suspect that this is true of intangible projects in general although do not have hard data to back up that claim. We need to investigate this.
- "Predictive" approaches to tangible projects are likely a good choice, but I suspect we can do better. I suspect that "predictive" approaches are more appropriate for tangible projects, such as building houses or buildings, than agile/lean approaches. I also believe that a hybrid approach combining the best from traditional, agile, and lean strategies is likely better than traditional alone. Having said this, as with the previous point, I don't know of any research that has compared the various project management paradigms for tangible projects, so this too is something we need to investigate.
In short, we know that "predictive" is a deceptive term for a large category of projects and suspect this to be true for other project types. As a result the only use of the term predictive in DA is to tell you that we don't use it.