guidelines for rewriting questions

[Roger Dingledine <arma@mit.edu>]

Ok, so I was going to post some hints for how we should go about
rewriting some of the "importants". This isn't going to be as good
as it should be because I don't have enough free time right now to
think about it, but it will be a start.

a) We need to avoid asking about specific products or standards. We
want to focus on *functionality*, not whether the user has heard
of a given widget or name. In some cases we should reword the
questions so they talk about the functionality involved, and in
other cases we should throw the question out because it's too
specific or it's covered sufficiently from other questions. 

This affects questions like:
* running servers (mail, httpd, samba, telnetd, ftpd, etc) 
* security certification (C2)
* PnP support in hardware 
* I2O support 
* Unicode support

b) We want to examine the actions involved in the issue, and
figure out *why* they would want to do something, not what it
is they might want to do. This would change "being able to undelete
files" to "being able to restore information that has been recently
deleted".

c) We want to avoid asking several questions at once. Every word counts.
For instance, "secure banking and commercial transactions" is really
asking
1) "do you want to be able to do commercial transactions (purchasing)?"
2) "do you want those transactions to be secure?"
3) "do you want to be able to interact with your bank online?"
4) "do you want those transactions to be secure?"

We want to decide what each question is really asking, figure out
which one we think is most important, and try to word it perfectly.
We want to especially avoid asking the user questions like 2 and 4
above, because the question turns into "do you realize that you
should want those transactions to be secure?" We can't really say
much no matter which way they answer it, other than "whoever's
teaching the public about security should do a better job", and that
has little to do with our job right now.

We have been focusing a lot on specific functionalities (virus
scan, defrag, configuring for isp dialup). It would be nice if we
could figure out some of the basic issues we're trying to ask, from
the more complicated issues, and ask both:
"How important is general ease of use of software?"
"How important is general ease of use of interface?"
"How important is it for the software to be powerful?"
Fundamental issues like this should get asked too, probably towards
the top of the survey.

I propose the following format for modifying these questions, so we
can keep track of things:
* being able to restore information that has been recently deleted
  [was: being able to undelete files]
  [was: microsoft "unerase.exe" program]

Whenever we modify a question, put its entire history with it.
That way we'll know at a glance what we were thinking at each stage
of development for the question, and things will generally be more
orderly.

--Roger