It was an accurate representation of where testing was at the time, which is what most of us thought of. It was never "final silicon", of course not. It was a snapshot of where things were as of last June, and you can extrapolate just how much things can change with a little over a year left before launch. A big part of that was thanks to Albert Penello when he came in and gave insight into how the 4+ year journey on making a new console takes place.
What "most people" who spread a lot FUD were saying was that a "9.2 tflops console" would be a "best case scenario", since maintaining 2.0 GHz GPU frequencies wasn't plausible. The PS5 was supposed to be a "~8 tflops console" at best...
The only thing right about any of the stuff was that both consoles did end up having double digit tflops. Matt was the only one accurately predicting the difference between them (he said about a ~15 % difference) and continously noted how close these consoles are/were, both before and after GitHub.
There's no doubt GitHub got things right, but a lot of things wrong. No RT, no RDNA2, no 2.23 GHz clockspeeds, no 10.28 tflops... It was, aside from the CU counts - still - mostly wrong. Matt is the only one who's been consistent throughout this, even going as far as predicting MS'll have a slight edge in tflops.
My reaction to the variable frequencies on the PS5 was, well... Let's just say, I really doubted Cerny. I'm taking more of a "wait and see"-approach now.
Matt has, yet again, hinted at most games being identical, with
some catering to the advantages of one or the other (slightly better GPU, twice as fast SSD), which has me really excited for the exclusives on each - I can't wait to see more of both!