If you ask me, at this point it really comes down to how you want to gamble. The 8500 right now appears to perform about on par with the Ti200, with the Ti200 coming out about $40 cheaper. To move to the GeForce3, you're paying $50 more than the 8500, and the Ti500 will set you back over $100 more. (these are ballpark prices)
The big gamble is: how will the cards perform a year from now?
This is a very relevant question for me because if I'm going to pay $200+ for a card, it needs to last me at least 3 years.
I think there is a very good likelihood that the 8500 will be able to outperform the Ti200 a year from now because it has much more potential and the performance gap is small. Will I gamble that the $40 extra will be worth it a year from now? Absolutely.
It is less likely that the 8500 will outperform a GeForce3 a year from now, but still quite possible. Will I gamble that the performance difference will be worth $50 more for the GeForce3? No, because again the potential for the Radeon to catch or surpass the Ge3 is still very real.
What about the big Ti500? Do I think the Radeon 8500 will be able to outperform this giant? No. Am I going to gamble that the performance difference will be worth $120 a year from now? No way! $120 is a good down-payment on the new card I will inevitably buy in a few years, and the 8500, even if it doesn't oust the Ti500, will do well until then (and it still has the potential to surpass the Ti500, although I honestly don't think it will happen).
So my choice -- the Radeon 8500 -- the best graphics investment on the market today.
The biggest factor here is the fact that the Radeon is not optimized yet and the Nvidia cards are very close. In order to get a substantial performance gain from Nvidia in the future, I will have to buy a new card. From the Radeon, I can expect good performance increases in the future for FREE from new drivers and Dx8.1 optimised software.
I know this was long, but IMO it summarises the whole debate.
Save the children