Rasters are boring

To put it simply--absolutely not. Well, maybe in QGIS a little bit, but rasters have potential far beyond the needs of an average GIS analysis. First of all, rasters do not need to be in two-dimensional space. There are 3D rasters called voxels, which can be analyzed in their volume or cut to slices, visualized in a whole, in slices, or as isosurfaces for various values (Appendix 1.1).

Furthermore, cells don't have to be squares. It is  common practice to have different resolutions in different dimensions. Rasters with rectangular cells are supported by QGIS, and many other open source GIS clients. Rasters don't even need to have four sides. The distortions (we can call it sampling bias in some cases, mostly in statistics) caused by four-sided raster cells can be minimized with hexagons, regular shapes with the most sides capable of a complete coverage:

Okay, but can rasters only store a single thematic? No, rasters can have multiple bands, and we can even combine them to create RGB visualizations. Finally, as we contradicted almost every rule of the raster data model, do individual cells need to coincide (have the same resolution)? Well, technically, yes, but guess what? There are studies about a multi-resolution image format, which can store rasters with different sizes in the same layer. It's now only a matter of professional and business interest to create that format.