Going to move this discussion back to the J-20 thread.
The same logic you’re applying to the ventral fins also apply to every other edge surface though. If you’re primarily concerned about the canards because of backscatter effects, absorbers in the scatter hotspots, edge treatments, and materials that can control or alter the specular angle of the backscatter should be adequate to greatly reduce if not eliminate that problem. In that sense the canards not being planar to the wing should also help that aspect.
Even if you can reduce the RCS of canards by 90%, it could still make a real difference on the overall RCS of the aircraft when we are talking about 0.001 m^2. Not being planform aligned with wing is definitely a concern. Also, I think when we are talking about managing the scattering of RF, you are not managing it evenly across all the angles. So with canards, you can manage it really great from the front, but maybe not so good from the back. Which might not be a problem when we consider the other issues from the rear.
Noted, but the underside of the F-35 looks like a mess to me. I say looks like a mess; I'm sure it's designers know better and all those bumps and bulges don't damage the RCS. I don't think we can just eyeball the J-20 and say that these "deviations" from our simplified layman understanding of RCS minimization are actual problems. Especially since they seem relatively simple to fix; for example, the J-20 engine cowling could certainly have finer serrations and hug the engine closer if that were required. Perhaps it's to do with the fact that the WS-10C is still an interim engine that the cowling isn't optimized for?
Also, how much of these differences are militarily significant? Suppose a radar could detect an F-22/F-35 in certain conditions at a distance of 100km. If that same radar under identical conditions could detect a J-20 at 105km, is that extra 5km really going to make much of a difference?
I don't think that's the right way to look at it. In fact, having higher RCS on the rear might not even be an issue if you can operate it in a way that keeps enemy aircraft in front of you until you are ready to speed away.
Think about it this way. Let's say F-35 is a magnitude better than J-20 in frontal RCS. With same radar technology, the range to track F-35 would be 1/1.78 of J-20. That sounds not good until you think about J-20 having a larger radar/EO tracker and EW suite than F-35, which would eat into some of that advantage. On top of that, these things are all networked. Having more aircraft in theater would give better situation awareness to both J-20/F-35 and also allow them to pick up more RF. China could also have anti-stealth radar in the theater which would point them to the general vicinity of F-35. which would allow J-20 to search a much narrower band and have longer tracking range. These are never like your RCS is x times my and your radar is y times my, so I can always track at this z multiple greater than you can track me. Depends on the tactics, the surrounding aircraft, the missiles, the sensor fusion/networking capabilities. All that is to say don't be concerned even if j-20 is 0.001 m^2 from front and F-35 is 0.0001 m^2 against X band radar. There are a lot of other factors in play.