Interpreting messages: Every artefact can provide sev-
eral kinds of messages, which it understands by itself
and can react onto them automatically in a reasonable
manner, like triggering an action or starting a negotia-
tion with some other artefact about a certain resource
(e.g. allocate time in a test environment). These di-
rect interactions are similar to interactions found be-
tween software agents and require a dened message or
knowledge interchange format, e.g. KQML [9]. For ex-
ample, the owners of an artefact can post information
about a planned down-time on the wall and inform
thereby dependent components, which can switch to
alternatives or react on the down-time automatically.