Two major pictures of repositories:
Harvesting or federated model. Education is drifting towards the federated model – the trusted source – and this breaks the open model. It should be the case that anyone can create the repository, and anyone can access the repository. A federated repository system will compose of only a few major repositories – I am in favour of many small repositories where the search is done somewhere else.
A need for a mechanism to express rights – we need GPO licences stating this is open source for example. We need mechanisms for showing that our work may be freely shared. This should be in the metadata – a pointer describing the rights – for example how creative commons works. We need an equivalent for learning resources. This keeps the size down, and allows you to change the resource rights at some point, unlike including the rights within the metadata.
Aggregation that brings in second and third party metadata – information from the creator (first party), but also evaluation and review from users – individuals and organisations. Self organisation and recommendation systems. Many small repositories but also many aggregators – so that search and aggregation can benefit from the fine granular approach, and the quality of your search increases.