That is, if using Newton's method one could start close to the solution but still not converge, while their method seems much more well-behaved.
I guess that might be a more attractive property than "just" being faster. However not an expert so not sure if other, cheaper alternatives also has this property.
That said, the whole point of their method is to use higher-order derivatives. Admittedly I don't do optimization problems that often, but when I do my function is seldom analytical, and so getting good higher-order derivatives are usually expensive at best and problematic at worst.
They do touch on this in the section about future work, wondering if a quasi-Newton method[2] could be created in a similar vein, using only sparse sampling of the higher-order derivatives.