This is the most stupid thing I've heard. Everything follows physics, even if you can't have perfect reference you can still design counter systems. Do you think these AA and ABM all have adversaries's physical reference during development? No, they don't.
You don't need perfect reference to design counter systems.
How would J-20's designers not know fifth gen requirements when they are designing a fifth gen aircraft? Are you even serious? Of course they know the requirements, how else would they start designing?
I think you don't quite understand how stealth work.
Stealth is not being undetectable, rather, stealth work in hiding its signature with the background, which radar can detect it, but cannot pick up as a combat aircraft, or as the old saying goes, the best place to hide a tree is a forest.
The topic at hand have nothing to do with physics, because the way to detect stealth aircraft is the same, it have to do with how you can process the signal, which is the problem for mathematic. Radar can detect all aircraft, but how much it return to the radar is another matter altogether.
For example, to simplify, let's say there are 5 band of radar return in all flying object.
Band 1 : Small Bird (such as Sparrow)
Band 2 : Large Bird (such as Canadian Geese)
Band 3 : Small Aircraft (such as 2 seater cessna)
Band 4 : Military Aircraft (such as F-35)
Band 5 : Large Aircraft (such as B-52 or Boeing 747)
Now, stealth work by hiding your signature return and make your radar think its is a different band other than Band 4. Say a F-22 flying signature is equal to a small bird, which mean you have to tune your radar to locate every small bird in order to pick up a F-22
Problem is, at this stage, unless you are working in LM, you DO NOT KNOW what kind of Radar image F-22 return on a radar screen when they are in full stealth mode. Couple with the fact that there are 1600 different speices of bird and around 2 to 300 different type of aircraft, not know what the F-22 looks like in a radar screen would mean you cannot pick it up when you actually see it.
Radar design to pick up noise, and process it, but stealth design to fool it, if you do not know what is the parameter of what F-22 is like, or what other stealth aircraft is like, how can you say it is more suited than F-22?
Also, what you know in your parameter when you design a thing is one thing, what you ACTUALLY made is another. If you care to study what F-22 was to USAF as a Fifth Generation Aircraft Platform, you would find the two are VERY DIFFERENT. Even YF-22 and F-22 are two different aircraft, how can you say you "Know" what is the Parameter of 5th Generation when you design the first one yourself?
Also, whatever Chinese have in their menu is their own way to design 5th Gen Aircraft, what US did is different, and what is the rest of the world did is another different set of parameter. How can you say since Chinese design a 5th Gen Aircraft, then Chinese would UNDERSTAND THE WHOLE 5TH GEN AIRCRAFT SPECTRUM ACROSS THE WORLD? They aren't uniform, you know that, right?
"Advancements in digital electronics are strongly linked to Moore's law" (Wikipedia)
You're giving me all these newer and recent processors that already reaching the limits of Moore's Law. Moore's Law is about exponential growth. Exponential growth has limits. We are at that limit, however if you look back at the past (as in 5+ years before), you can see that Moore's Law still strongly links to advancement and performance of electronics.
How about an example from ~10 years ago?
1st Gen Clarkfield i7 (i7 870) would beat 2nd gen (sandy Bridge) i5 2300 in benchmark
https://www.cpubenchmark.net/cpu.php?cpu=Intel+Core+i7-870+@+2.93GHz
https://www.cpubenchmark.net/cpu.php?cpu=Intel+Core+i5-2300+@+2.80GHz
I can give you any example in any year as long as there is a record for it on any benchmark site.
Again, Moore's Law only define the technology of making die would advance, not the die itself is a guarantee better than in the newer model.