Re: [Pan-users] One-off colorization of attributions.

[Duncan]

On Tue 14 Jan 2003 00:11, Eric Ortega posted as excerpted below:
> Duncan, I appreciate your response, but I know this.  It still drives
> me batty at times, though, and I've been a frequent Usenet-type-person
> for over ten years now.
>
> Just because it makes logical sense doesn't mean it makes visual sense.
>
> > A second level quote will be attributed by the guy at the first level
> > quote, so the attribution should be colored in the correct color for the
> > person that wrote the attribution, and it is.  Again, it's all entirely
> > and perfectly logical.
>
> Perfectly logical and confusing nonetheless.  Just because I start counting
> at zero all day long doesn't mean I want to do it at the dinnertable, too,
> IYKWIM.
>
> Perhaps the complications, to me, come about from having color involved.
> I'm not used to this "feature" and maybe I will do better by disabling it.
>
> What I was thinking might be more useful, though, is to colorize the
> "name" of the owner of the attributed text the same as the color of
> their text.  In this way we will only marginally violate the colorization
> of who actually wrote what, but we will make it much easier to go back
> through the multiple responses and easily pick out the quote owners.
>
> It's like that cognition test where you need to look at the word "green"
> when it's actually printed in a red ink ... the dissonance between the
> color and the word causes problems.  The same thing happens when looking
> back through eight levels of nested responses and trying to keep in mind
> that the text I'm reading is brown but it's not actually the attribution
> that's brown, they wrote that too ... it's the attribution before that that
> counts, etc. and so forth.  It's just messy.

I understand where you are coming from, certainly.  However, I believe even if 
PAN COULD be made to pick out attribution lines and color them by this 
method, it would confuse as many folks as it helped, because then the people 
who use the OTHER type of logic would get all confused.  I get a bit confused 
sometimes now, I admit, but I expect it's be far worse under your 
hypothetical, and the fact that PAN was the only one doing it would make it 
far worse.

Then there is the problem of coding it.  You better be sure it's coded right, 
because if it isn't, and it only gets half the attributions, while half 
remain as is, it would make things FAR WORSE!  I think you got it right when 
you said that's a job for Santa Claus!  <g>

Even assuming for now that one could code such a thing to work effectively 
enough of the time to be practical, one has to worry about the complexity it 
would add to the code.  Complex code is error-prone, bug-prone code, and even 
if it COULD be coded to deal with 99.9% of attribution lines, and COULD be 
demonstratably shown not to INCREASE confusion for as many as it helped, the 
increase in complexity and consequential bugginess risk simply couldn't be 
worth it, IMO.  Thus, perhaps unfortunately, I believe the proposal is likely 
to remain as mythical in practice as your proposed developer to take 
responsibility.

-- 
Duncan
"They that can give up essential liberty to obtain a little
temporary safety, deserve neither liberty nor safety." --
Benjamin Franklin