I thought it was because the stealth tech was based on faceting. This deflects the majority of radar energy directed at it so it doesn't return to the source to be detected.
That's simply not true.
Round surfaces scatter in all directions, flat surface scatter only in 1 direction of your choosing. For low RCS (Radar Cross Section) paneled surfaces are infinitely better, regardless of computer simulation capabilities.
Of course it's true. Modern stealth aircraft have curves a plenty. This is an extremely confident reply given that's it's to somebody who's referencing the book by the guy who built the thing.
But with the low-poly approach you end up with sharp edges which diffract the waves in all direction…
The fact that no more recent stealth plane have the same shape should be a good indication that this was in fact due to the technical limitation of the time (And btw, who are you to contradict the actual designer of the plane on that topic ?!)
On the other hand consider a flat mirror (at a long distance) or a faceted object with a small number of facets and sharp edges: it's unlikely that any of the facets is correctly aligned to reflect the light to your eyes.
A shape with fewer angles ("low-poly") has a lower probability of giving a visible reflection.
Modern stealth appears to work the same way: surfaces and edges are aligned along a minimal number of angles to reduce the effective number of "facets". See the shape of the B-2 for example, the leading and trailing edges are at the same angles.
It's like saying that 3d graphics lighting can only be done by rasterized rendering. Of course, now that we have enough GPU processing power, we can also use ray tracing.
In time, all light calculations/shadow calculations will be done with ray tracing, but for now, that's unimaginable because ray tracing runs slowly on current hardware.