I think you're missing Unimpressive's point that if something can be run virtualized we don't cease to refer to it as an operating system. If Linux had been (perversely, ahistorically) developed as jslinux first before the bare-metal interfaces were implemented would it now be considered not-an-OS? Conversely would the act of porting it have blessed its status as an OS without regard to the presence of the same semantics and facilities in the browser version? I think if you're making this argument you're running against the tide of common _and_ academic usage, plenty of research OS projects have been tested on hardware simulators. Perhaps the easiest way to simplify this is to ask whether you can present any rule other than "browsers don't count" to distinguish the virtualized Linux case from the jslinux case.

The point being made is not that a browser is an OS, but that something using the facilities of a browser to implement all the operations and environment of an OS is... an OS. This seems almost like the duck typing versus strong typing approach - one view is that it's the behaviors that define the identity of something, the other is that it's the inheritance & implementation[].

[] Alternatively since it's April Fools, I'll argue that you're interrogating this from the wrong angle and presenting an essentialist view of OS-hood.