in defence of [some] mnemonic URIs [was: Re: on "Make readable URIs": don't rely on users to remember/type URIs]

At 9:52 AM -0600 2/3/05, Dan Connolly wrote:
>I disagree with the main point of this draft QA tip, and
>I have a number of comments on the details as well.
>
>   Draft - Make readable URIs
>   http://www.w3.org/QA/Tips/readable-uri
>
>It seems quite counter to the principles of web architecture:
>
>[[
>It is tempting to guess the nature of a resource by inspection of a URI
>that identifies it. However, the Web is designed so that agents
>communicate resource information state through representations, not
>identifiers. In general, one cannot determine the type of a resource
>representation by inspecting a URI for that resource. For example, the
>".html" at the end of "http://example.com/page.html" provides no
>guarantee that representations of the identified resource will be served
>with the Internet media type "text/html". The publisher is free to
>allocate identifiers and define how they are served.
>]]
>   -- http://www.w3.org/TR/webarch/#uri-opacity

Fortunately the Web as practiced today is more "you get what you
expect" than the "you get what you get" policy that is the formal
policy of HTTP and is suggested in this quote.

>Let's please do *not* encourage users to remember nor type URIs.

That's gold-plated bad advice. It's anti-blind. It is indeed
discouraging that after all these years you, Dan, haven't learned any
more HCI than to say this. Or internalized the intrinsically
man-machine nature of the Web system.

Maybe we should rewrite the tip to make it clearer than not all URIs
can be, or should try to be, mnemonic. But some tip in this vein has
merit -- should be in the collection of tips. The best Web practice
with appropriate use of mnemonic URIs is better practice than the
best Web practice without any. And it's easy for the point-and-click
generation to overlook this.  That's why we have quality tips.

The user's recallable context for the current state of a Web dialog
is an asset in making the Web work well [1]. We should capitalize on
this capability, even as we work not to rely on it too much.  There's
a both-and of good practices, here.  "Many ways to win" wins.

The architecture document is correct.  At the bottom line, it asserts
the following as good practice:

<quote cite=
"http://www.w3.org/TR/webarch/#uri-opacity">

>Agents making use of URIs SHOULD NOT attempt to infer properties of
>the referenced resource.

</quote>

That statement addresses what automated processes should and should not do.
Not people.

People will recall or not recall URIs as a function of their
mnemonicity. Quite independently from anything the W3C can say or do.
And the objectives of both user and publisher will be better served
by an appropriate, measured use of mnemonic URIs in publishing a web
presence.

Not every URI can be mnemonic. In fact, counting all URIs at equal
weight, the majority won't be. But few sites would not benefit by a
stable hierarchical structure which is mirrored in mnemonic paths in
URIs.

Increasing the word-of-mouth flow of URIs into typing adds value to
the Web. Non-trivial value at the very least for those whose primary
display medium is audio.

People will recall URIs, and not page TITLEs, because of their role
in direct navigation through the address bar (or whatever you wish to
call it). You remember what you can act on, for what you can do with
it.

The clear value of mnemonic URIs is demonstrated by the land rush in
domain name registrations. It's a fact. Don't try to play King
Canute. Note the [to me annoying] ascendance of alpha mnemonics in
propagandizing business phone numbers. They're harder to type but
easier to remember. Takes more time, suffers fewer failures of
recall. Net win.

Global direct navigation, the ability to start browsing from a
user-entered URI is one of the democratizing features of the Web.
Don't trash it. Compare with W3C statements about 'deep linking' as
valuable. If it's worth offering these 'resources' to agents, it's
worth making them controllable as language by people.

Cut and paste works fine with arbitrary, opaque URIs. Of course, it
is error-prone in speech-only screen reader delivery contexts. Real
email, not ideal email, has a propensity to mangle the passing of
URIs among friends.  Things that mean something and spell right
from hearing have an edge in arrival-at-destination.

Word-of-mouth referrals will more likely succeed if there is a
memorable, typable value of a URI that gets the user within two or
three clicks of paydirt. These add to the likelihood that your
page/site/content will reach more people, over and above the people
you reach by inbound links.

[paydirt = the onset of net positive value added, in the user's value
system, for engaging in this thread of browsing.]

On the other hand, I have to quickly say that one has to recognize
the limitations of hierarchical structure for web presences. They
should have some, but it can't be all things to all browsing users.

The graph of active associations in hypertext allows people to arrive
at paydirt via many paths. Hierarchies get less and less effective in
communicating what is there beyond a depth around three levels. A
tuple-space index is more likely to connect with the user's intent.

Although squeezed between link-following and full-text search, site
hierarchy, reinforced by mnemonic-path URIs, still has a contribution
to make to Web user-friendliness and hence effectiveness.

Well-built websites, such as those live today, run by the high-value
businesses on the web, offer the user multiple ways to win. If one of
these is a text-friendly hierarchical site map that works like a
DAISY Book [2] Table of Navigation, that is a value-added feature for
the blind visitor. And many others as well. The upper reaches of this
hierarchical view of the offering, subject to its [likely] stability,
is well done to be reflected in a stable path tree using mnemonic
path segments.

Maybe we should rewrite the tip to make it clearer than not all URIs
can be, or should try to be, mnemonic. But it is still true that the
best Web practice with appropriate use of some mnemonic URIs is
better practice than the best Web practice without any.

Al

[1] http://www.trace.wisc.edu/docs/ud4grid/

alt: http://www.google.com/search?q=grok-locked+loop

[2] http://www.loc.gov/nls/z3986/

alt: http://www.google.com/search?q=DAISY+talking+book

>I don't think this is established as best practice:
>
>[[
>Use directories instead of ?aid=342h987f2
>
>Why?
>
>Today it is common for persons to use different computers and browsers,
>which do not readily synchronize bookmarks with each other. If your page
>use URIs like "http://www.site.com/fishing/" instead of
>"http://www.site.com/?aid=342h987f2", it is more likely that users will
>be able to remember the URI for later reference. Thus they will avoid
>excessive use of search engines, and won't need to find the link on your
>possibly over-crowded first page.
>]]
>   -- http://www.w3.org/QA/Tips/readable-uri
>
>In what way is use of search engines "excessive"?
>Using search engines and navigating from one page
>to another is to be encouraged, not discouraged.
>
>
>So much for my main point; on to relatively smaller comments...
>
>[[
>Most (all?) web servers will serve
>"http://www.site.com/fishing/index.shtml" (or whichever extension) if
>you type the address "www.site.com/fishing" in the address field of the
>browser.
>]]
>
>Not all. That's a server-side convention, not an HTTP protocol
>feature/constraint.
>
>I don't know what "This should work even if you submit variables in the
>URI." means at all.
>
>"Upgrade your web server as needed" seems counter productive. Give
>links specific documentation for the most popular 2 or 3 servers.
>
>
>--
>Dan Connolly, W3C http://www.w3.org/People/Connolly/
>D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E

Received on Thursday, 3 February 2005 18:45:09 UTC