Both are fairly interesting, from different puzzling perspectives. The Sudoku Code comes from the tradition of puzzles leading to metapuzzles in puzzle hunts like the MIT Mystery Hunt. There have been a few Sudoku variations in past hunts (783658 and One of a Kind.

In The Sudoku Code, sets of two puzzles work together to provide part of a long message. Knowing the message before or after the set you are working on allows one to add extra information to the solving of both number places.

Snakes on a Sudoku is based on a more established variant on number place where irregular regions replace the standard square regions. You can find it as Geometry Number Place in Ed Pegg, Jr.'s article on Sudoku Variations. Snakes on a Sudoku adds a constraint for the regions that they be "snakelike"; the region can't branch or touch itself.

Geometry number place can create a very different solving experience; regions account for more and less of each row or column, disrupting the balance of a standard number place.

Whether either is "more interesting" depends on your attitude toward solving number-place puzzles. I'm glad to finally have some variants after the glut of number places, regardless of how they get marketed.

Thanks Tablesaw, this is great additional information. I've seen the different region-based Sudoku, but can you explain a little more about the snakelike region constraint?

In terms of dividing the grid into "snakes," I've given the basic definition already: no branching, no touching. If it helps, you can add "no bunching." "Snakes" appear in some other logic puzzles as well. Dave Tuller and Michael Rios's Challenge Your Brain features a section called "Snaky Tiles" where the challenge is to divide a grid into these "snaky" segments.

It's easiest to explain by example. Look again at the geometry number place I linked to. The region in the top right corner branches, there's a square where the region creates a T intersection. The same is true for the region in the bottom right. The region in the top left either "bunches" or touches itself. Either way, you can see the bulge where four squares touch each other. In this grid, the only "snaky" region is the one in the bottom left.

I haven't solved enough of this variant to know how different it is from geometry variants generally (I don't intend to buy SoaS until I'm finished with The Sudoku Code). But looking at a few of the grids, it appears that the number of reagions in a row or column varies widely, with a few having only one two regions, and several having more than four. I would imagine that this would more often disrupt the solving by forcing the solver to consider more sources of information for certain cells.