There is a strong possibility of a name conflict within the key space of a resource with the four reserved keys, [id, type, href, links]. I'd like to float the idea of prefixing these keys with an underscore (or other character). Thus we would have the reserved keys of [_id, _type, _href, _links].

The primary benefit is that it opens up the possibility of using a key with one of the currently reserved key names. I think it is quite likely for some existing datasets to be using the href and type keys and this will save developers the need to rename these existing keys. A secondary (and minor) benefit is that it's easy to pick out the special keys visually.

The only downside I see is the adding of an extra character to each key, but with compression that is mostly a non issue. It's also possible there could still be conflicts, but its probability is probably a couple of orders of magnitude less.

An example:

{
  "posts": {
    "id": "1",
    "title": "Rails is Omakase",
    "post_href": "/rails_is_omakase"
  }
}

would become:

{
  "posts": {
    "_id": "1",
    "title": "Rails is Omakase",
    "href": "/rails_is_omakase"
  }
}

or

{
  "posts": {
    "_id": "1",
    "_type": "posts",
    "_href": "http://example.com/posts/1",
    "title": "Rails is Omakase",
    "href": "/rails_is_omakase",
    "_links": {
      "author": "9",
      "comments": [ "4", "5" ]
     }
  }
}

Thoughts?