So here's my 2 cents:
Sandy Bridge, architecture improvement
Ivy Bridge, die shrink
Haswell, architecture improvement
Broadwell, die shrink
Seriously, I don't get what all the buzz is about. Just as Ivy was only ~5% faster than Sandy, we can expect Broad to be ~5% better than Has. If you are building a desktop rig in 2014-15 when Broadwell comes out, you will not be 'behind the times' to get a Haswell chip that still runs in a socket.
There is another interesting thing brought up by the Anandtech podcast last week:
Intel currently only has 3 chip designs on the market, and they get amazingly good yields on their chips. On the low end you have your base model chip which covers all Atom CPUs. On the midrange you have a chip design that covers all Pentiums, i3s, i5s, and consumer i7s. Then on the high end you have the Xeon based i7's and server processors. If you have a Pentium G CPU today, it is very likely you have a fully functioning i7 that has been artificially 'broken' so that it functions as what you purchased.
So the idea is that you buy a motherboard with BGA CPU integrated into the board. Need an upgrade from that i3 that you could afford when you bought it? Send Intel some money, and they send you a code to enter in to upgrade the system to an i5 or i7. Personally, I am not a huge fan of this idea as it makes repairs a major pain, but from an upgrade standpoint there is potential for this to be much more user friendly than mucking about with doing a little bit of surgery with your system mucking about with thermal paste and potentially having to reformat the OS to recognize the changes properly.
Lets not forget the even more concerning stuff that was announced over a year ago: Intel does not want to simply get rid of chip choice. They are huge proponents of moving to a fully SOC style architecture in the future where the CPU, iGPU, NB, SB, Audio feature chip (note their recent dealings with Creative), networking chip, and possibly even RAM all on a single chip on the motherboard. This would mean that Mobo manufacturers would have to focus on this like... the color of the motherboard, or the type of capacitors, and that's about all that they would have to differentiate themselves from their competition. In all seriousness it would mean that the competition would be on BIOS/UEFI POST times and features, and extra gimmicks like over-volting USB ports to charge devices faster, which would be great to see. But there is no way that most mobo manufacturers survive that transition.
So, on the one hand. Yes, I am a little disappointed because as a power user this means less choice for me. On the other hand I see it simply as a sign of the times; I mean honestly, other than for gaming and video editing (2 things I have less and less time to do these days now that I have kiddos), my big rig is used more and more as a home server rather than a client desktop. In 3-5 years from now when it becomes time to upgrade again I expect my desktop to become a dedicated server. whatever I replace it with as a client will probably be a physically small device with a lot of connectivity. Think of something like a dockable tablet capable of standing alone, but when docked can game on a 4K display with ease.
That is the direction things are moving in, and while it is very different, it is not altogether a bad thing.