There are lots of reasons to come out with design revisions. Same with driver revisions, yes?
Are all subsequent driver releases the result of the current drivers being broken?
Revisions can be the result of a need to fix broken hardware, the original Phenom had a broken TLB that required a new hardware revision (
B3 stepping) to correct. Likewise with the Intel
6-series chipset recall that required a new stepping to correct the problem.
But some revisions are done for cost-reduction reasons, either directly in having a lower-to-produce cost or indirectly by way of enabling higher yields so there is more revenue generated from the same production costs.
I'm not aware of anything being broken with the GTX590, ergo a revision to the GTX590 would presumably be undertaken by Nvidia as a cost-reduction and/or yield enhancement activity.
That is clearly nothing more than speculation on my part (regarding the intent of the revision) but suffice to say I think we can all agree that "revision" does not always mean "replacing something that was broken".
One would hope the new revised GTX590 offers something more to the end-customer, lowered power-consumption or quieter fan operations maybe, but saying the customer's who own one already should receive a replacement seems a bit over-zealous doesn't it given that the original GTX590 performs as specified?
I owned a power-hungry B3-stepping Kentsfield, when the much lower power consuming G0 stepping came out I bought five more, I certainly did not feel like Intel cheated me by not replacing my B3-stepping chip at the time.
If I had bought a GTX590 I would not expect it to be replaced unless it failed to perform to spec.