I kind of agree. Lot's of very expensive stuff without any tangible way to know if it's going to be used effectively or just end up sitting there.
I think a mid-range i7 CPU with 4 cores or even the "low-end" i7 920 overclocked and pushing between 3 and 3.5 GHz, plus 4-6GB of memory will be fine. As for graphics cards, it depends on what you want. I have no problem waiting 6 months to a year for DX11 cards to get cheaper. If SoW comes out in the meantime, my Ati 4890 1GB still supports DX10. I doubt there will be heavy use of DX11 specific functions like tesselation on the initial release version to make me want to get a DX11 card before they drop in price.
Louisv has a point. Hardware is way ahead of software, so why go for top of the line hardware if we can't bring the medium range line of components to their knees? On one hand, people want to get a capable rig with some amount of future proofing, a PC that will last a couple of years without major upgrades. On the other hand, if you go too far with top-end components you might find yourself in a position where something newer comes out and as is often the case, it happens to be incompatible with what you have (different sockets, need for specific components and so on). It's a fine balance between the two.
Just like people with a core 2 quad couldn't simply upgrade to an i7 without buying a new motherboard and DDR3 RAM, there might come a point where you can't upgrade from an i7 to the next CPU without changing most of your components.
If and when this happens i think it's better to be in the middle of the scale of the previous platform, so that you haven't spent an enormous amount on building a PC you never managed to use to its full capacity before it became obsolete.
Absolute performance is a very poor benchmark in my opinion. What more accurately reflects the consumer's needs is having a PC that can run most or all current games at high detail and some of them at maximum, plus a small performance margin or capability for future minor upgrades to make sure you won't need to do a major upgrade sooner than 1-2 years, but not so powerful that it runs everything at 50% load and then gets superseded.
That would just mean the other 50% of processing power that you never used before upgrading to a different CPU is money thrown to the wind. Most importantly it's money that could have gone to an upgrade towards a mid-range component of the next platform, which could do as well as the top-end component of the previous platform for less money. The only catch here is waiting a bit for the new models to penetrate the market and trends to stabilize...a person getting the fastest core 2 quad 6 months before the i7 was released would probably wish he'd gotten a mid-range core 2 quad for less and keep the spare change to go towards an i7 when it hit the shelves.
As an example with abritrary numbers, if an i7 920 is priced at 200-250 Euros, can be safely overclocked to 3.8 GHz or more and runs SoW maxed at 80% load, while the i7 980 is priced at 600-800 Euros and runs SoW at the same detail level (max) at 50% load, it's better to get the 920, keep those 400 Euros and use them after a year or two to buy a complete set of new motheboard, new RAM and a mid-range new CPU from the next series. That is, unless you want to be running photoshop on a second monitor during the dogfights in order to edit the screenshots with your favorite parts of the fight at the same time that they happen
Most importantly, today we have quite a lot of things that can break the line of continuity. If it's not the new CPU requiring a different socket and motherboard or RAM type, it will be the new OS requiring a different video card to make use of the latest DirectX version and so on. So, for people who do upgrade it's better to do a series of small upgrades when the price of the component is low and that usually happens when it's either a low or mid-range model of the best line of products, or when it's the top-end model of the second best line of components.
Of course, others may have a different opinion on this, they may really need monster PCs for other reasons (eg, graphics and video editing professionals) or they simply decided to spend 1500-2000 Euros on a PC and not upgrade for the next 5-6 years no matter what. In that case, when going only for maximum system longevity with minimum upgrades, it does make sense to get the absolute best you can afford. The money spent will end up being the same over the course of time as someone who follows the method of incremental cost effective upgrades, maybe even less.
What throws a spanner in the works in this method however is potential failures. What happens if i break the bank on a monster PC, don't upgrade anything at all and after 4 years it's still going strong but my graphics card gets fried one day? There's nothing compatible with my motherboard to replace it without changing everything else is what usually happens, which leaves me wishing that i'd spent the same amount of money incrementally over the years so as to be up to date with the current hardware standard and have available spare parts to buy
Just food for thought and how most people i know tend to do things. Your mileage may vary and after all, it's your own wallets.
