The implementation should adjust to the spec, not vica versa :) AsciiDoc and AsciiDoctor should both be fixed.
There is no spec...yet.
The main issue with the constrained formatting in AsciiDoc is that it isn't constrained enough. I'd like to move towards rules that require constrained formatting (like *, _ and +) to be at the boundaries of a word character, with some exceptions of trailing punctuation. This should make them less greedy for things like this.
Having said that, Google is still a chump for selecting the + character, which is already loaded with meaning in IT.
Thanks for the comments. I disagree with the last comment "google is a chump for selecting the + char", because a lower level text (such as a link) which is properly escaped for a higher level text (such as adoc) should never disrupt the higher level text (such as adoc). Similarly, adoc text written in an SQL statement's field, properly SQL escaped, should not disrupt the SQL statement.
Or a freemarker template noting an SQL statement, which in turn notes an adoc statement which in turn notes a google plus link, should not be disrupted anywhere :) But the plus might be escaped up in level 2, 3 & 4 (and it's level 2 escapes might be escaped too in level 3 & 4).