Comments on Device Independence Principles 020628 from ACCESS

Dear Mr Yamakami,

I'm forwarding your mail to the mailing-list.
If you want to interact freely with the mailing-list I would recommend
subscribing to it.

Best Regards
Stephane Boyera
 --
Stephane Boyera		stephane@w3.org
INRIA/W3C			+33 (0) 4 92 38 78 34 
BP 93				fax: +33 (0) 4 92 38 78 22
F-06902 Sophia Antipolis Cedex,		  
France


-----Original Message-----
From: T. Yamakami [mailto:yam@access.co.jp] 
Sent: Monday, July 29, 2002 10:33 AM
To: www-di@w3.org
Subject: Comments on Device Independence Principles 020628 from ACCESS




Hello, DI experts:

 This is Yam, ACCESS, Japan.

 Attached please find the comment to DI principle draft dated September
2001.  I would like to express my thanks to Dr. Gimson who kindly
advised me how to post a comment.

 Regards,

-- Yam

===============================================================
Comments on Device Independence Principles 020628 from ACCESS
(yam@access.co.jp) 1. Summary:  We think the DI-2 to DI-6 needs further
considerations to be embedded into the Device Independence Principles.
Even we agree the high level goal of the Device Independence Principles,
there are three major problems in the current set of Device Independence
Principles.
 (a) It is too web page identifier-centric.
 (b) It is too tightly coupled with implementation-dependent or policy
decision issues.
 (c) The protection for end users should be clearly emphasized.

 In conclusion, we propose to split the current set into the two parts,
one for universal access guidelines and the other for content adaptation
principles to clearly separate the individual distinguished rights and
interests.

2. Rationale:
 The device independence principles draft describes the high level
guiding sprit of the device independence works in W3C. The goal of the
easily accessible web is an agreeable one, however, the current DIPs
needs a significant revision. The fundamental issue to be resolved to
augment the web usability is that the device independence principles has
two aspects, one from the end-user requirement driven, and the other
from the value-added server requirement driven. These two aspects may be
peacefully met in the long run, however, the significant discrepancy
among them should be well recognized.

When we discussed the issues in Delivery Context Workshop in March, we
witnessed the significant confusion among the various technical terms
surrounding Device Independence. For example, these terms include
dynamic content, single authoring, universal access, delivery context,
context dependent web service, multi-modal access, and content
adaptation. The author agrees with the DIP-1. However, as discussed
below, DIP-2 to DIP-7 are too web identifier-centric to consider lessons
from the reality of wireless Internet.

The fundamental issues in the current device independent principles are
as follows:
(a) the principles assume that the dynamic content adaptation with
device capabilities is good for end users, which is questionable.
(b) the principles are too tightly coupled with implementation or design
policy details.

The following use cases describe the conflicts that seem quite difficult
to be resolved in the current practices.

Use case 1:
 This principle needs more attention for  the end user protection.
Considering the fact the start point of this device independence is the
wide variety of emerging new information appliances, the end user needs
more protection for control of users and accountability of the current
content and current preference that the content server accepts.
Considering the limited capability in the devices, which leads to the
limited meta level controllability, the importance of the end user
capability of choices and rights to know the content adaptation process
cannot be underestimated.  In the past, we can see some of the promising
technologies fail due to the lack of the attention to the end users. For
example, push technologies assume that the pushed content is meaningful
for end users and additional value of timeliness could be accepted in
end users in any sense. The fact is the assumption made at the server
side without considering the resulting communication cost that the
content is valuable to end users were wrong in many cases. When someone
out of control of the process pays the cost, the technology should be
very careful about the end results. It is important to learn from the
past experience in the Internet. One of the reasons of the Japanese
explosive emergence of mobile Internet using handsets lies in the fact
that the wireless carriers extensively protect the end users by
excluding any end-user unaware information exchange.  In the environment
in which we would like to explore the device independence, end users are
so powerless that they sometimes cannot check the mark up language. The
importance of end user protection cannot be underestimated to empower
the technologies in this domain.

Use case 2:
 Whether the web page identifier is constructed by
"http://www.foo_bar.com/a.php", "http://www.foo_bar.com/page1.html" or
"http://www.foo_bar.com/a.cgi?page=1" is completely author's choice.
Each device has a content size limit, which leads to the split of the
same content to the different presentations. The current principles
implicitly require that such split should be URL-transparent manner.
Because the implementation to accept like
"http://www.foo_bar.com/a_particular_content_on_some_special_device_page
1.html" is unacceptable in the current principles. Whether it is done in
such a manner or the most frequent used device can enjoy the fixed
format to cope with the traffic is completely under content author's
decision. The current principles have easy conflict with this design
policy decisions.  Even considering the fact that the DIP-2 clearly
states the interpretation of the web page identifiers are for further
works, the identifier issue directly conflicts with the content design
policies which is out of scope of W3C.

Use case 3:
 Considering the poor protect ability of information appliances, at
least one of the wireless carriers in Japan intentionally does not
expose the URL on mobile handsets.  The current principle has definitely
conflict with this policy.  We can argue the legitimacy of the hidden
URL, however, it is a policy issue and it is not adequate that the basic
principles have easy conflicts in the widely accepted practice.

Use case 4:
 The current principles have the implicit assumption that the using the
best of the device capability is good beyond doubt. It should be
arguable. It is the content provider side logic that does not pay for
every octet transferred.
  The end user should have the significant control of the capability
choices because the richer set of content is more expensive. When the
control is under the server control, it easily leads to the richer
content because the communication cost sensitivity does not exist at the
server side.

Use case 5:
 It is a common practice to redirect the URL access to the
language-dependent page in Internet. For example, an
"http://www.foo_bar.com/index.html" can be redirected to
"http://ww.foo_bar.com/index_fr.html" for French,
"http://www.foo_bar.com/index_jp.html" for Japanese, et al.  It is
questionable that language can be constructed in this way and that the
multi-modality is constructed in a completely different strict DIP-based
manner.

Use case 6:
 One of the reasons to stick to the single URL is to empower people with
the ability to share URL-based knowledge. However, in the assumption of
content adaptability, it is little use for single URL because the end
users have no method to know the how the content adaptation is done
among different type of devices. It is even misleading. Without the
user's capability to know the  adaptation process and which type of
information is used during the adaptation, it could be even harmful.

3. Discussions:
 We don't disagree with the usefulness of the content adaptation. It
could stand as one of the independent techniques in the world. However,
the current principles starts from the universal access for end users
and shifted to the total monopoly of the power by server technologies.
This twist could harm the total value chain and easily corrupt the
diffusion process.  We think it is appropriate to split the principles
into two parts, one for end-user, and the other for value-added service
at the server side.  For example, for the pure end user accessibility
viewpoint, there is no need to accept all the capabilities. It is a
valid solution to define a minimum guaranteed capability set, which is
counter-evolution, therefore usually not acceptable for Internet
standards. To protect the end user interests and suppress the temptation
to control everything on the server side could be the best way for the
wide acceptance of the newly introduced device independence paradigm.




References:
[1] Gimson, R. Ed., "Device Independence Principles", W3C Working Draft
18 September 2001, available on
http://www.w3.org/TR/2001/WD-di-princ-20010918/.

Received on Monday, 29 July 2002 05:02:08 UTC