Re: Web forms considered harmful (to learning the shape of the question).

From: Jim Pitkow (
Date: Tue, May 18 1999

Date: Mon, 17 May 1999 22:30:22 PDT
To: Al Gilman <>,
From: Jim Pitkow <>
Message-Id: <99May17.223101pdt."363208">
Subject: Re: Web forms considered harmful (to learning the shape of the     question).


  Thanks for the feedback.  I've forwarded them over to Henrik Frystyk
Nielsen whose overseeing the development of the issue/question tracking


At 08:41 AM 5/17/99 , Al Gilman wrote:
>At 06:07 PM 4/13/99 -0700, Jim Pitkow wrote:
>>                                               ...  The WCA is requesting
>>each WG with characterization issues submit a set of questions by May 15,
>>   We are in the process of setting up a Web-based submission form to
>>facilitate letting us know your groups set of questions. 
>Why a Web form?  It's a truly _wrong_ tool for the right job.
>The question you are posing is "Define the domain where you would like our
>assistance accumulating knowledge."  Web forms are a bad fit to this
>question, to a serious degree.
>The semantic class of the desired input is a query; it is a statement of
>the form "we want to know about FOO, where FOO is defined by {assertions}."
> One potentially effective representation for this meta-characterization -
>the articulation of what is and is not known about [foo] - is a schema.
>This defines the question by asserted a web of relationships between known
>things and unknown things.
>An example of an application of Web technology which would be significantly
>more appropriate for this question-gathering exercise would be a template
>document in least common denominator HTML.  The template should be a
>scenario template such as those one would use in a User-Centered
>Requirements process as articulated by Karan Harbison
><>.    The ability to make connections by
>document structure and by citing URIs in the submission is critical to what
>the inputter has to convey - the shape of the question.  
>Web forms are carefully designed for applications where you want to get
>multiple people to respond under the constraints of an unchanging shape of
>the question.  In the present case, it is precisely the shape of the
>question that you _want the respondent to control_.  If one wants to
>constrain inputs in this case, it is a language or schema for structure
>building that one should impose as the constraint, and not a set structure.