I’m not sure what you mean about persistence of the character strings: I consider these comparable to written words. The language may go away/be forgotten, but a pattern of characters is immutable. Of course their meaning, not so much….

All of the above thinking can be applied to URLs or URNs as well. And any meaning that has been assigned to any of these entitites (e.g., URIs in the semantic web) can be ‘expropriated’ by the community, if it should misuse, misunderstand, and/or misapply a given URI. (Even when the original owner explicitly states a conflicting meaning, it turns out.)

So your key point for me is, “Who do you trust to maintain a system? So when we are talking about persistence are we really talking about trust?” Yes, trust, and also technical evaluation — will the network and social environment adopt and support that technical model? Many sincere companies fail to reward an investment, not because they aren’t trustworthy, but because they did not produce a successful product or outcome. So it will go with URIs, in various times and ways.

So I conclude that any identifier that isn’t resolvable in all browsers (e.g., dois and urns) will always suffer from limited adoption, for social reasons, and the additional ‘persistence’ obtained by the URIs “designed for persistence” can be roughly matched by social and technical mechanisms overlaid on plain old URLs. The choice of URI matters, but no URI mechanism perfectly avoids the pitfalls. So yes, use what works today, and seems likely to work tomorrow.

Next essay: Transparency?

Lee