There's nothing in the XForms spec regarding the specific naming of the child node, other than there can only be one:
...The detached copy must consist of content that would be well-formed XML if it existed in a separate document. Note that this restricts the element content of instance to a single child element.
And according to the ODK Spec, the unique form identifier (aka filename) and version are specified as attributes of this node:
<?xml version="1.0"?>
<h:html xmlns="http://www.w3.org/2002/xforms"
xmlns:h="http://www.w3.org/1999/xhtml"
xmlns:jr="http://openrosa.org/javarosa"
xmlns:orx="http://openrosa.org/xforms"
xmlns:odk="http://www.opendatakit.org/xforms"
xmlns:xsd="http://www.w3.org/2001/XMLSchema">
<h:head>
<h:title>My Survey</h:title>
<model>
<instance>
<data id="mysurvey" orx:version="2014083101">
<firstname></firstname>
...
[I read thru 152, and associated issues] So making the child node name be some (hopefully, which in some documented cases it wasnt...) valid XML custom label conglomoration of a filename, timestamp, what-have-you, is IMHO pretty naughty and asking for trouble... The desired unique id and version is right there in the attributes!
Indeed, because there is only ever going to be one child node to begin with, it is in fact largely redundant in trying to uniquely identify elements within the scope of their associated instance XML submission. If the purpose of this exercise is to minimize the prefix needed to uniquely identify exported properties in, say, a flatfile (eg CSV, etc) then the only things you need to uniquely identify a submission property are its node name and the paths of any groups/subgroups it falls under. Further, if, in addition, you want to globally uniquely label each property - ie across all forms/versions - then the necessary information needed to accomplish this is in the child node's attributes (if it exists anywhere at all).
I believe (strongly!) that attempting to exploit the child node name to somehow globally uniquely identify submitted property values is bad, and should probably be punished to the fullest extent of the W3C Specification Violation Law (which amounts to about 30 lashes with a wet noodle these days...)