The plan is evergreen + IE11. We’re still working on a properly configured build that produces an IE11-compatible output but it will come.
We’re actually not using proxies by default. The observation mechanism is still largely the same as in v1.
We do have an opt-in to use proxies which adds some “automagic”-ness to a few things like computed observation (in v1 you need to explicitly declare dependencies, in v2 dependencies are auto discovered by proxies), and the default observation mechanism can be configured to use proxies instead of getters/setters. However, this is disabled by default because it performs worse on mutations.