The 2X performance over 3080 Ti is almost definitely taking RT/DLSS 3.0 into consideration, not raw gaming performance. So you can expect to see that kind of performance boost only in limited scenarios. I expect the 4080 16GB will in reality be 25-40% better than the 3080 10GB and the 12GB version probably 15-25% faster in raw performance. All the bells and whistles are nice, but Nvidia is relying heavily here on enhancements from new tensor cores and rt cores. I can say this because we have a pretty good understanding already that the raw performance of the 4090 is 90% faster than the 3090, but it also has >50% more cores and faster clocks. The 4080 16 GB has only <12% more cores than the 3080 and the 12 GB version actually has <12% less cores than the 3080. The 4080 just is not going to be 2X the performance of the 3080 ti or even 3080 for that matter if the 4090 required 50% more cores to get 90% better raw performance. It is nice to see cards that should finally be able to handle real time RT and make it a practical enhancement, but, its unrealistic to think you'll get a return on that $1200 GPUs feature set for quite some time.
Edit: Actually its very evident from Nvidia's webpage that DLSS 3.0 and RT "Overdrive" are exactly the reason they are claiming 2X-4X faster than the 3080 Ti. This was even done in DLSS performance mode likely to maximize the performance uplift percentage.
But there is one more problem I have with your post. How many "it's just $100-$200 more for X more performance" should we be okay with? The 1080 Ti, the best gaming GPU you could buy at the time was only $600-$700. The 4090 here is starting at $1600 now and the 4080 (not Ti) is now $1200 (or starting at $900 if you want to pretend the 12GB version is not really a 4070). I know there has been quite a bit of inflation since then, but should a high-end gaming GPU really be more than $700-800? Apparently Nvidia thinks so, you now have to shell out at least $900 to get into the "high-end" club.
I agree there is a discussion to be had regarding price of GPUs. But, like anything, it comes down to value. First, we've had some pretty nasty inflation over the past 18 months. That can't help keep cost down. Second, my understanding is that Nvidia cards are pricey to make and profits are lower. When looking at the 1080Ti, a great card, its MSRP was $699, the same as the regular 3080. Given that the 3080 is 75-80% faster (in benchmarks) I'd say that makes the 3080 a pretty good value, if you assume that the 1080 was priced appropriately.
Now consider the 40 series. You have a new GPU that cost $899. That's about a 28% price increase. Will you get 28% more performance out of it? I guess we will have to wait and see. But, if you get 50% or more performance at that cost, that would seem like a reasonable value. Again, assuming that the 3080 was priced appropriately.
I think a lot of people are upset because they are used to getting more performance for the same price-point as the previous generation. Maybe that's not possible any more.
I agree that the benchmarks leveraged DLSS and ray tracing. That's one of the areas where Nvidia tends to prevail over AMD. And, higher resolutions (4K) as well. The question is at what point do higher FPS become a moot point? Do you really need 300, 400 or 500 fps in games? Maybe, but I'm thinking there is diminishing value there. Maybe there is a 40 series that will deliver 100-200 fps in most games, with DLSS and ray-tracing turned on at 4K. The question is what should that cost?
For me, the Nvidia announcement didn't really impress enough to wait for a 40 series card. I'll likely go 3080 w/12G or wait to see what AMD brings out next week. I'd love to find a card for $500-600 w/3080Ti performance, but I'm not hopeful for that happening short of buying a used card.