As Mitchell pointed out below, there are severe theoretical issues. The theory predicts 22 new particles (and doesn't even specify their masses thus making detection difficult) and fails to account for the properties of many existing ones. Most importantly, the theory has no chirality, and chirality is extremely important for fundamental particles. This error is severe and no way to avoid it has been found. E8 does crop up in string theory in other settings but it's unrelated to Lisi's work.

A deeper problem is that the theory doesn't actually solve a lot of problems, even if it were true. You mentioned Not Even Wrong; here is an excerpt from Peter Woit on the E8 theory:

One idea Garrett is fond of that has generally left me cold is the idea of unification via a large simple Lie algebra like E8. While there may be some sort of ultimate truth to this, the problem is that, just as for GUTs and for superstring models, all you’re doing when you do this is changing the unification problem into the problem of what breaks the large symmetry. This change in the problem adds some new structure to it, but just doesn’t seem to help very much, with the bottom line being that you get few if any testable predictions out of it (one exception is with the simplest GUTs, where you do get a prediction, proton decay, which turns out to be wrong, falsifying the models).

I personally consider the publicity over the theory as simply a failure mode of science journalism. The theory got some tentative endorsement from some physicists (before they could look at it more deeply) and thus it was prematurely promoted by journalists despite not being distinguishable from the thousands of other theories-of-everything that crop up and never get any attention.