Otherwise
machine reading could rescursively read the templates and get the
information from the fields as well.
No it can't. ... not reasonably. The only way to reliably recurse in
anything external is to have a copy of all the templates, and to
implement a full mediaiki parser. The only truly complete parser is
the (rather slow) one in MediaWiki. ...
We could use html tags (eg. <span class='reviewer'>) and parse the html
page instead.