Is the web sticky enough?

Domain names are, by policy, transferrable

We’re using the internet for a world-wide-conversation, and lately we’ve been wrapping a lot of communications in atoms of information we sometimes call “posts”. People pass around links to posts and sometimes copies of posts, and we have feeds full of posts, tweets and things included. It’s all good. Hopefully the posts have unique IDs so we don’t confuse them with one another and they can be cross-referenced. But what is that ID? How do we choose them? Usually today we use URLs. URLs are rooted in DNS names. Full rights to DNS names are reassignable according to the policy of ICANN. There’s currently no policy to say the future owners of a domain can’t reuse URLs for most anything they want. Whoops?
Illustration

If we want our decentralized world-wide-conversation to work for a very long time, we need to consider strategies for keeping our post IDs unique. If we use URLs, do we care that a malicious future owner may in some ways “overwrite” our posts in history by issuing their own posts with the same ID? For that matter, are some gTLDs more stable roots than others for long-term ownership?

Would we be better off working with an ID namespace for posts that had an explicit policy for the persistence of identity uniqueness at least written down, if not enforceable by contract law? A new gTLD?

Post archives

Here’s one idea that might get around all that. What if there was a giant P2P historical web that could be queried with a date and a URL (or any URI) and it did its best to return whatever post or resource was issued there at that time. Peers could make sure that posts they felt were important stayed in the archive by mirroring them, all regardless of where on the web the information was originally posted. A distributed file system of sorts.

Atomism and Holism

Of course, where we stop writing atomic posts and start communicating in high definition streams, it seems the persistence of ideas gets a lot harder, as one of the inspirations [scripting.com] for this #blogpostfriday post starts to get at.

What would Heraclitus say?

Further reading

Drummond Reed. “The Persistence of Persistence” [www.equalsdrummond.name]

Bob Wyman. “The Persistence of Identity” [www.wyman.us]

5 Responses to “Is the web sticky enough?”

  1. @plaggypig Says:

    Individuals and businesses need to trade names that fit their particular interests and markets. A namespace that didn’t cater for this demand would be a pretty unattractive one.

    I dare say if identifiers weren’t reassignable then the process of registration would probably have to be qualified with checks that applicants are eligible to own any particular name, thereby raising the cost and reducing participation.

    But anyway, couldn’t the problem be mitigated with an audit trail? The governing authority (TLD registry) could log a timestamp whenever a domain is transferred between registrants, and provide a list of these historical changes (via whois).

    I’m loving your blog Mason, this is a great discussion.

  2. Mason Lee Says:

    @plaggypig Thanks. I agree, the re-assignability of memorable names is important for some systems. An audit trail maintained at the DNS level wouldn’t be able to tell you if path-named resources were reassigned. Can you explain more your thinking?

  3. Mason Lee Says:

    Textbook examples of Atom feed elements use GUIDs as IDs. The nice thing about these is you don’t need a central authority to generate them, they are just probably unique. Here is a scheme for generating them and for representing them as URIs:

    http://www.ietf.org/rfc/rfc4122.txt

    RFC 4122 uses 128 bit IDs. Using 256-bits would get you almost enough IDs to assign one to every atom in the universe (Est. 10^80). (Did God use 256-bit IDs and run out of space?)

    How resource IDs fit in with versioning is an interesting question…

  4. Mason Lee Says:

    Just found this briefing paper on the same topic:

    http://www.digitalpreservationeurope.eu/publications/briefs/persistent_identifiers.pdf

  5. Mason Lee Says:

    The Tag URI scheme (RFC4151) looks like a nice method for minting URIs that are globally unique across time. They aren’t network resolvable by nature, though.

    http://www.ietf.org/rfc/rfc4151.txt

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


Follow

Get every new post delivered to your Inbox.

%d bloggers like this: