Yeh I've seen that as I use TPU but it's heavily skewed towards MS/Xbox as it's obviously a PC site and majority NA-based traffic.
IGN's is the best barometer in terms of demographic as it's console-centric and full of casuals. It's the biggest gaming website by far still.
I think the split will be 2.5:1 this gen or even more heavily skewed to PS5 and thats ok as MS can make money from Gamepass with a much smaller install base.
There's no way to prove that TPU is "heavily skewed" one way or the other any more than it would be feasible to discount the IGN poll as being "Sony biased" for some reason. Both polls are legitimate barometers of where things stand as is the AndroidAuthority (perhaps that one is skewed to Sony because, erm, Sony uses a Linux OS variant and makes Android OS phones? Joking. But this is as silly as your dismissing the TPU poll for similar reasons). The sample sizes of all (3) of those polls are far too large to dismiss any of them as "outliers". Likewise with the techspot poll which had about 35k respondents at last check before it was pulled. It showed similar results to the TPU poll, with a 50.5% XseX vs 49.5% PS5 split.
There was also a large Twitter poll (with 10s of thousands of responses IIRC) in March which showed an advantage for XseX. By your metric, that is about as "casual" an audience as it gets. I'll see if I can dig a t up and post it here.
On a side note, polls by far have their greatest predictive value when we do serial polling and look at trend lines. Far moreso than aggregate polling, for example, because individual polls are merely snapshots frozen in time. But when we look at serial polls we begin to see a reliable picture with strong predictive value. So, the value of combining the polls we have at our disposal - each of them basically large "snapshots" - is limited...but still far greater than anything else we have at our disposal.
I'm not saying you're wrong, by the way, about the possibility of a 2.5:1 eventual sales ratio. What I am saying is the same thing I've been saying all along in this thread - that any tiny GAF poll on this topic is less than useless by virtue of being conducted in a heavily biased, non random sampled population of extremely small size. The 60%-10% results here are ridiculous vs the much larger polls which even in the IGN "worst case" show far more friendly splits to XseX.
In the end, I think sales may end up being more like 2:1 globally (67-33%) and around 1.5:1 (60-40%) in the states. Both of us can speculate of course. The difference is what we allow as guidance to inform our speculation. I choose to eliminate bias.