Why simply not always 3D?

We know very well that the Earth is a 3D, nearly spherical object and, therefore, all the dynamic processes inside our planet are inherently three dimensional. Therefore, it is very logical to assume that realistic geodynamic modelling should always be done in 3D. Also, if you talk to geoscientists studying various natural geological objects, you are frequently told that such objects can only be modelled in 3D. This is a normal expectation since they are perfectly aware of the spatial 2D variability of geological structures on the Earth's surface and they thus know that a similar variability also exists in depth. Therefore, 3D modelling appears to be the natural choice for ‘observers’. What about ‘modellers’? Why don't they always use 3D modelling? What's wrong with it? The ‘uncensored’ truth about 3D modelling is the following:

• 3D thermomechanical modelling is quite easy from a methodological point of view – it is fairly straightforward to formulate and discretise the governing equations in a 3D Cartesian geometry for both simple viscous and more realistic visco-elasto-plastic rheologies (especially by using the same relatively simple finite-differences and marker-in-cell techniques that we extensively discussed in this book).

• […]