I infer it's not a high occurrence issue by the fact that no one else is complaining about the problem. Either they aren't seeing it or it doesn't bother them (or enough to chime in). I'm not saying give up on it, but it may take a while to fully understand (if ever). I haven't found any smoking gun Microsoft documentation that indicates how 'pinning' determines the appid to use. Yes, it would be best if we could figure out if there's anything we can do to get Windows to add the right appid, but it may be that getting addpm to set the AppID in the shortcut it creates is sufficient to resolve this adequately.
And I have seen it work "by default" (without even running addpm) sometimes. I don't have a good explanation for when it "just works" and when it doesn't. But it seems that adding the appid to the shortcut doesn't hurt and makes it work in cases where it wouldn't otherwise. So I still think we should get addpm to add it. But for those that don't run addpm, we can document the need to set the appid in shortcuts using either win7appid, or the registry tweak to edit it in the properties dialog or advising them to run addpm. At least on emacswiki.
I'm sure we can get addpm to do it. We should make it conditional on what the OS is. It shouldn't try to add the AppID on older versions of Windows that don't support property stores in shortcuts (anything before Windows 7). There might be other issues.
Rob