It really wouldn't make it more useful as it would still only show the hops from those specific servers to an end destination - which may or may not share any hops with my route to said end-destination. The only real use I could see would be as a "is this site up" service, or a tool to check propagation of a domain name in other parts of the world - but, that's where utilizing nameserver lookups would be quicker and less resource intensive than a traceroute lookup.
Multiple trace route sources for what I do is very helpful, for example:
A client reports an issue that some of their website visitors are having packet loss/reachability issues, however the network I am working on has 3 transit providers, which means 3 different ways into the network.
I can certainly do a trace route from my machine or routers to their clients (if i know their IP) but that's only showing me the outbound route, I am still none the wiser as to which route they are trying to get in via.
By using multiple off-net trace routes I can probably enter my core network over each of the 3 transit providers. Looking at these results I should be able to see if there are any obvious issues with incoming traffic.
It's certainly not going to work every time, however it will give me a better view as to what's going on.
