Some time ago, I read a text about building random tables using different sets of rolls. At the time, it was made little impression on me. That small impression has grown, though, and birthed an expectation of tables that include a lot of entries, with only a couple of rolls.
First, I started with a 2D6xD6 table type. 2D6 roll on one axis and a D6 on the other. 66 entries. Cool.
Then came the 2D6^2 tables. 2D6 roll for each axis…121 entries. Yass!
The way the bell curves on each axis interact means some results have a very small chance of happening–a 2 & 2, for example–which means there’s space on the table for a highly unlikely event to be rolled. It’s far easier to lay out results as common, uncommon, rare, and very rare on such a table. It’s also easy to lay out more possible results than on most tables, with finer control of chances, than even with D100 tables.
The use of 2D6 rolls also keeps the tables a workable size, I think. One could always try 3D6 for each axis, say, and the percentage chance of the corner cases being rolled would be so miniscule as to not be worth laying out (unless using a computer). Printing the tables out would also demand more space than is likely available in a spread, without the entries being too small to be legible (for older readers, anyway).
So, I’ll be availing myself of both 2D6x6 and 2D6^2 tables.