Hybrid planning - the optimal environment for an AI¶

Project planning is a sub-area of project management in which an AI can demonstrate its capabilities particularly impressively. This is especially true when project management creates an optimal environment for an AI with so-called hybrid project planning. At Can Do, we understand this as a combination of waterfall, rolling and agile planning.

Rolling planning¶

Can we assume that the further into the future events are, the more difficult they are to predict? Conditionally, the answer is yes. Recurring events like Ramadan, solar eclipses or Christmas can be predicted fairly accurately. But when will the administrator be finished with the installation? They usually don’t know themself.

A look at the calendar helps. The planning follows exactly this principle: events that lie in the distant future are difficult to predict or are currently unknown. The result: the calendar is full with many appointments planned exactly for the next few days and maybe a few weeks. Then the entries become fewer and fewer and the specific times less and less frequent. The whole thing is interrupted by precisely predictable dates, such as public holidays, and controllable fixed dates, such as holidays.

The above principle can be extended by a derivation of accuracy: the further an event lies in the future, the fewer details can be planned.

From this (greatly simplified) the planning procedure of rolling planning can be derived: I plan precisely or finely in the short term and imprecisely or roughly in the long term.

Applied to project management, this means that work packages are planned in detail in the near future (a few days or weeks) - but still imprecisely, since they are based on the planner's knowledge and assumptions. The further the planning horizon, the more work is grouped together (trades, phases, epics, etc.). This grouping is then broken down into more detailed work when it is "closer".

Thus, a plan is rough at the beginning, i.e. with few planning objects, and becomes more detailed over time. For example, at the beginning of the project the plan consists of 10 planning objects, at the end of the project it consists of 2,000.

In rolling planning, a "wave of detail" is constantly pushed ahead until the project goal is reached. This approach is realistic and corresponds to the human way of thinking. Practice shows: planning does not become more realistic the more detailed it is. Nevertheless, they do exist, the huge detailed plans that plan a project partly to the day and for years. However, they are not based on reasonable assumptions, but often arise from the planner's fear of forgetting something. Or they are simply wishful thinking. Always remember: Just because something is written in detail in the project plan does not guarantee that it will happen.

Agile planning¶

Agile planning is an extreme variant of the method described above. It is particularly popular in the IT industry. It arose from the realisation that most IT projects in the past were completed late, became more expensive and / or did not meet expectations. Therefore, the planning was adapted to the fact that such projects are precisely afflicted with a certain degree of unpredictability.

When a project and the work involved is unpredictable, it is logical to "drive by sight". You work from day to day or from Sprint to Sprint. You finish when you finish - whenever. This method is not as bad as it may sound, but it needs to be critically reflected upon. There are definitely IT projects (and also other projects, e.g. in construction) that could be successfully planned and executed. And when there is justified criticism of projects, we all tend to attribute this to human error on the part of others. This is superficial and wrong: often it is quite different influencing factors that cause a project or scheme to fail. And usually it is also clear to many of those involved from the beginning that the project cannot and will not go as planned. But economic constraints, lack of preparation time and the influence of unqualified people make really good and realistic planning obsolete from the start.

Of course, technical experts could theoretically predict a project like Stuttgart 21 or Berlin Airport exactly - at least 80 to 90 percent. But that would mean an unjustifiable amount of planning. Moreover, planning would have to be done with certain tolerances, which contradicts other decision-makers and the classic budget planning of state-owned or commercial companies. Unfortunately, no contract is awarded on the basis of a (realistic) offer "between eight and ten million". Rather, the company with the cheapest offer is awarded the contract.

It makes perfect sense to plan only what you can predict. After all, the plan for a project should represent the likely future, not just wishful thinking. However, this future can also include not knowing something and simply having to try it out to see if it works.

This task can be solved faster in IT than in other projects. This is what distinguishes an agile, cloud-based IT industry from most other industries in 'old' industrialised countries such as Germany. The perfectionism sought in industries such as construction, automotive or machinery is fundamentally anchored because such problems cannot be solved there comparatively quickly, cheaply and easily. This is why agile working has developed and established itself in a sensible way in the IT industry in particular - but not in other industries.

As sensible as it may be: agile work and agile project planning are sometimes not possible, even if the management postulates this to show its own modernity.

Ideal combination: hybrid planning¶

But if neither rolling nor agile planning can serve as a standard tool for all project planning cases - what then? The hybrid planning approach could offer a way out of this dilemma. This combines two planning techniques:

1. At the rough level, classical, i.e. rolling, planning is used.
2. Agile planning is used at the detailed level.

Combined with the ability to plan imprecisely, even at a coarse level, we come very close to realistic and probabilistic planning that matches people's level of knowledge.

This planning method sounds complicated at first, but it is not at all. In classical long-term planning with its milestones, phases and budgets, an - imprecise - wishful thinking is formulated from a rough perspective. The details, during the actual implementation, are planned long term (within the framework of Sprints). The framework of the rough planning is not left behind. Only the human factor is taken into account on different, quasi-vertical and horizontal planning levels. At the strategic level, wishes that are not unrealistic are described vaguely.

Example: "Project end in the 4^th quarter". At the operational level, which has a high level of detail, the work is planned for only two weeks (one Sprint) and / or somewhat longer (2 to 3 Sprints). This way, the different perspectives of the "strategic" managers and the "detail-oriented" project staff come together and complement each other in a goal-oriented way. Agile work is dynamic and constantly adapts to new insights and circumstances - everything is in flux. Strategic planning, on the other hand, looks at the 'big picture', has a long time horizon and is characterised by vagueness. The hybrid planning technique takes the best of both worlds and thus does justice to the thinking and actions of all actors. It is applicable to most, but of course not all project challenges.