Re: modeling wallets

On Sat, May 16, 2015 at 6:00 PM, Melvin Carvalho <melvincarvalho@gmail.com>
wrote:

> We're thinking along the same lines, I think.  Where I may differ slightly
> is as to the container / content boundaries.  I'm particularly careful to
> use an HTTP document as a container of content, and not a concept in
> itself.  This is because I can then monitor things like the request state
> of the document, when it was last accessed, the etag etc. without confusing
> it with the data.  I suspect this is going to be the magic sauce when this
> work goes to production, at least, that's my bet.
>

Yes, the line between the data and its representation should be clear
(REST!). Monitoring/analytics could easily be gathered from the API
exposing those documents.

The first challenge is going from a user to a wallet.  Of course the user
> may have more than one wallet.  So what do you think the predicate should
> be called?
>
> 1. Alice  wallet  <URI>
> 2. Alice hasWallet <URI>
> 3. Something else
>

In order to encourage privacy protection, I think the predicate should be
unidirectional, i.e. allow the subject to describe its relationship with
the object (e.g. Bearer tokens, Bitcoin private keys), but not the
opposite. In that sense, I would also go with (1), not with (2) since
ownership can be privacy unfriendly.