I was hoping this thread would benefit from your wisdom, bookey - you were the person I had in mind when I made the comment about "others ... in a better position to offer properly informed comment".
SRVCC is painful and difficult, it is better to push 4G coverage to be greater than that of 2G and don't provide any level of fall back....
Every time I've read something about SRVCC, it's accompanied by commentary that amounts to "you really don't want to go there". It sounds so difficult to get right that I can see the advantages of building a 4G network that makes it unnecessary.
2G network has another 20 years left in it, the spectrum used may decrease but a lot of M2M type services rely on this, solid base layer and have no need for high bandwidth services.
Expect the 3G network to cease before that of 2G, 4G is a much better data layer and a much better blend of spectrum.
There is consideration to use HD-Voice on the 2G stack but it would be more spectrally efficient to move 10Mhz of spectrum from 2G to 4G to shift the capacity.
My comments about 2G were from the perspective of phone and mobile broadband users. I long for the day when I can get better than a 2G signal everywhere I use my devices. As things stand, there is no network that currently offers better than 2G service in all the locations I visit regularly, so I work round this by carrying a 4G phone on one network and a 4G Wi-Fi hotspot on another.
Once the networks reach the point of being able to service phone and mobile broadband devices without needing 2G for fill in coverage and for circuit switched fallback in the absence of VoLTE, they can then run a data and SMS only 2G service in minimal bandwidth for M2M applications such as alarms and smart meters. M2M hardware is price sensitive and these devices only need to exchange small amounts of data, so GPRS or EDGE is more than sufficient. This sort of 2G voice switch off can happen gradually, with 2G voice service remaining in areas of coverage difficulty until the networks are content than the 4G coverage is sufficient.
2G technology will also live on in other contexts. I'm not sure whether Network Rail have completed their GSM-R network yet, which is replacing their legacy voice and data networks.
I can see the attraction of getting rid of the 3G air interface once 4G coverage and device support reach a suitable point. 3G is an "in between" technology that even in the DC-HSPA configuration doesn't offer the data capacity phone and mobile broadband users will increasingly demand, nor the coverage that 4G should eventually offer. 3G also has no advantages for M2M, which doesn't need the greater data bandwidth.
Ultimately, reusing the spectrum currently dedicated to the 3G air interface for 4G makes sense, not least as the majority of devices should support LTE-A carrier aggregation (allowing them to operate on multiple bands at once) by the time it becomes feasible to withdraw 3G.
Consumers and networks alike should derive maximum benefit from legacy free 4G service for phone and mobile broadband devices.
It is possible to discern a pathway towards the future in all this. Legacy air interfaces can only be discontinued once all users can move to newer technologies. There can be no 2G voice or 3G switch off whilst these air interfaces are still needed for circuit switched fallback, so VoLTE is an important element of moving towards a legacy free future.