[OpenID] human-memorable namespaces (was RE: Identity concepts explained by cartoon ducks)

Drummond Reed drummond.reed at cordance.net
Sat Jan 6 08:16:33 UTC 2007


Jens,

I had a good talk with Eve Maler from Sun about this the other day about
your point about human-memorable namespaces. Part of the reason that AOL
users are stuck with names like "fredsmith226" is how constrained DNS syntax
is. One of the major changes from XRI 1.0 to XRI 2.0 architecture was
shifting the XRI delegation character for reassignable identifiers (i-names)
away from dot to star.

Why? Besides the fact that under the new URI spec (RFC 3986), dot is no
longer a reserved character, the big reason was that we realized its power
as a logical separator. For example, many organizations have standardized on
dots as logical separators in the username portion of an email address. For
example, my email is drummond.reed at cordance.tld (tld there to kill the spam
bots -- it's really cordance.net). Unlike the "cordance.net" string, where
the dot is a DNS delimiter, in the "drummond.reed" string the dot is not a
machine-meaningful delimiter, only a human-meaningful delimiter. In other
words, it's only there to make the name easier for people to remember
because it divides a single syntactic namespace (the username string) into
smaller logical units for people to process.

Apply that same concept to a global namespace and it actually becomes rather
easy for every human on the planet to have a relatively easy-to-remember,
easy-to-say, easy-to-transcribe address consisting of at most three logical
segments.

With regards to English language names, for example, we talk about this as
the "John Smith" problem. There are probably at least 100K people with the
name "John Smith". With a single logical segment you could have only two
i-names:

	=john
	=smith

With up to two logical segments, you can have up to four -- the two above
plus:

	=john.smith
	=smith.john

But with three logical segments, you not only can expand the size of the
semantically-meaningful namespace to include all middle names...

	=john.alford.smith
	=john.buford.smith
	=john.carson.smith

...etc., but when you run out of middle names (or don't have one, or don't
want to use one), you can add any memorable word in the English language to
create an semantically meaningful name:

	=john.smith.montana
	=montana.john.smith
	=john.smith.boston
	=john.smith.boston
	=boston.john.smith
	=john.smith.red
	=red.john.smith
	=john.smith.mustang
	=mustang.john.smith

Of course that's just in English. The XRI namespace is Unicode, so the same
principle applies to all Unicode languages (as supported by XRI registry
services). Dot (and dash, also supported in both DNS and XRI syntax) are
universal characters that can be used as logical separators in most (but not
quite all) languages.

I just wanted to make this observation since the perception that
human-memorable global namespace can't scale doesn't take into account
richer namespaces designed for this purpose.

=Drummond 


-----Original Message-----
From: general-bounces at openid.net [mailto:general-bounces at openid.net] On
Behalf Of Jens Alfke
Sent: Friday, January 05, 2007 6:02 PM
To: openid-general
Subject: [OpenID] Identity concepts explained by cartoon ducks

In the future, even comic strips will make important points about  
online identity.
At WhatTheDuck, that future is today:

http://web.mac.com/aaronandpatty/iWeb/What%20the%20Duck/Comic%20Strips/75180
11B-1F35-45D5-9A08-30C18FC87B91.html

The ducks have discovered that a global namespace does not ensure a  
human-meaningful name ... the same corollary of Zooko's Triangle that  
I brought up the other day:


> The ... marketing stuff points out how simple and clear they are.  
> [But] After a few million have been snapped up, the latecomers won't  
> be able to get anything nearly that nice: just like AOL users,  
> they'll be stuck with "=fredsmith226365" or "=ilikecheeeeez", and  
> the memorability advantage is completely lost.

Note that duck #1 has decamped for the alternate approach of an 11- 
digit non-meaningful "key", which is paradoxically easier to remember  
than the "friendly" alphabetic name. In fact, he's probably already  
associated a short "petname" with the key on his cellphone, so he  
won't have to remember the key at all.

As you can see, I'm really quaffing this Zooko/Petname kool-aid :-)

--Jens

PS: If you dig funny animals explaining CS, you must read "Why's  
(Poignant) Guide To Ruby" < http://poignantguide.net>.
_______________________________________________
general mailing list
general at openid.net
http://openid.net/mailman/listinfo/general




More information about the general mailing list