I completely agree with what you said, in fact, I think after a certain point it's impossible to trace and count how many jets PLAAF has given their information disclosure policy.
BUT, at this moment, we still could estimate a number. We know Wuhu is the first frontline troop with J-20 from CCTV if my memory is correct. that means before other places got J-20, the total number of J-20 is known, which is 20. It's also confirmed by all those serials photographed.
Later we heard from "multiple" sources claim that "Fushun"? in NTC will be the 2nd to have J-20, but until now we still haven't seen any pictures of flying J-20 around that place, given how dense Chinese cities are and how close Fushun base to the surrounding cities, as such a high profile jet, it's very difficult to hide J-20 like this.
Moreover, we have some secondary evidence suggested that the four J-20 with WS-10B in yellow primer at CAC witness in Nov 2019 was still at CAC (at least around April for these newly released pictures).
Combine all these, we could logically estimate the number, but again, deep down, I seriously hope I'm wrong, for example instead of Fushun, the 2nd place to have J-20 is somewhere else.
The problem is that your estimate is literally the "floor" estimate based on the number of airframes we've identified and projected based off serials.
We have 8 serials identified at Dingxin and 4 serials identified at Cangzhou, and 6 serials identified at Wuhu (including up to the 9th aircraft at Wuhu), meaning it's reasonable to assume that the floor estimate of in service aircraft is 8 + 4 + 9 = 21 aircraft.
The problem is we don't know what the true number of J-20s at Dingxin, Cangzhou and Wuhu is, which is why the number of aircraft we are estimating based off serials is a definitive minimum floor, rather than an estimate of the actual number.
If you are suggesting that the serial numbers we have identified are all of the true number of airframes, I would find that to be very surprising.
In fact, as far as the 60 number that pupu states is concerned, that is about where my estimate for what the likely true number of J-20 airframes would have been, using the rule of thumb that the real number of airframes is 2.5-3 times the number of confirmed serials for new aircraft types early in their introduction.