Based on the discussion in #253 a need for guidance discouraging usage of blank nodes was identified.

Even though it is one of the principles of Linked Data (to identify everything with an IRI), publishers tend to use blank nodes for various types of entities including some very important ones like dcat:Distributions.

I suggest:

1. Not to use blank nodes in any examples in the specification. A similar effect (shortening) can be achieved by a combination of prefixes and relative IRIs.
2. Recommend explicitly to identify everything with an IRI and support it with a link to this Linked Data page. This should be placed somewhere in the beginning of the specification.
3. Explain the disadvantages of using blank nodes - impossible to link, different identifiers with each access