A project from around the year 2000, which now has published an RFC numbered RFC 5058. Still there are broken links on http://www.xcast.jp and the main open source implementation has disappeared from the IBM website.
Among the few things I found is this slide presentation from the IETF meeting in march 2000. The following picture suggests, that XCAST provides an explicit top-down list of recipients to deliver data to, then does indeed do a spanning tree distribution which qualifies as multicast.
The only available download is an IPv6 implementation for BSD from the year 2001.
The most important point in XCAST seems to be to provide the same kind of API to existing IP Multicast applications as IP Multicast. In what way XCAST can actually be useful needs further research, but if the project already died off a year later, that isn't exactly motivating...
Slide 18 suggests there are more similar protocols out there, but that doesn't mean they are available.
words of wisdom from RFC 5058
In "meshed unicast" or "multi-unicasting", the application keeps track of the participants' unicast addresses and sends a unicast to each of those addresses. For reasons described in Section 3, multi- unicasting (rather than multicast) is the prevalent solution in use today. It's a simple matter to replace multi-unicast code with Xcast code. All that the developer has to do is replace a loop that sends a unicast to each of the participants by a single "xcast_send" that sends the data to the participants. Thus it's easy to incorporate Xcast into real conferencing applications.
Of course, we prefer to call "xcast_send()" castmsg().