Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!husc6!lloyd!kent
From: kent@lloyd.camex.uucp (Kent Borg)
Newsgroups: comp.sys.mac.programmer
Subject: Re: Checking for MF (was Re: Need some MF help)
Message-ID: <370@lloyd.camex.uucp>
Date: 12 Apr 89 00:24:09 GMT
References: <1179@internal.Apple.COM> <2749@pegasus.ATT.COM> <1200@internal.Apple.COM> <1558@husc6.harvard.edu> <38341@bbn.COM>
Reply-To: kent@lloyd.UUCP (Kent Borg)
Organization: Camex, Inc., Boston, Mass USA
Lines: 82

In article <38341@bbn.COM> levin@BBN.COM (Joel B Levin) writes:

[In talking about detecting whether MultiFinder is running, whether
there are icons along the right edge of the screen, whether the user
is running a Finder replacement...]

>Let the user set these values and their defaults on the basis of needs
>or wants; don't presuppose what they should be based on external
>circumstances which you might not be able to get right anyhow.
>
>	/JBL
>==
>UUCP:     {backbone}!bbn!levin		POTS: (617) 873-3463
>INTERNET: levin@bbn.com

Two comments.

1) Finder replacements seem more-and-more delicate to me.  I hope MDTS
is working closely with any brave souls who are doing this.

2) ``Let the user set these values...''  No, No, and No.  

When the Mac user interface was first being designed there were many
tough questions: ``Should we go with A or B?''.  Engineers have been
faced with tough questions for years, and engineers are usually payed
lots of money to find good answers.  People accept that--nothing new.
Now move to the field of user interfaces and people start popping up
with the suggestion: ``Let the user decide!!'', ``I know, we'll make
it user configurable!!'', etc.  No, that's poor engineering.

	It is not legitimate to pass engineering decisions off
	to the end user.  

(I do dub thee ``Kent's Law of User Interfaces'' ~boing~)

If you have an engineering problem (``How do I know whether there are
icons I might be covering up?''), then come up with a well engineered
solution (even ask MDTS for help), do *not* leave it for the user to
fix.  That is not a legitimate solution with other fields of
engineering, it is not legitimate with the Macintosh (IBM users might
get a macho buzz out of configuring their own software and
hardware--but this is comp.sys.MAC.programmer).

Wait!  Before you all kill me for locking out the user, I certainly
think there should be lots of room for end-user customizing.  I'll
even say ``as much as possible''.  But think first.  I can adjust the
seat in my car--but I don't need to configure the door latches.  I can
set how brown I want my toast--but I don't need to reshape the heating
elements.

	= If a user adjustable feature is a cop-out for making an
	  engineering descision, it is a mistake.

	= If the user adjustable feature will make your gizmo
	  dangerous, it is a mistake.  (``Car Crash!  ID=-1, Wheels
	  fell off.  Sorry, you should have configured to run with lug
	  nuts `on'. '')

The bottom line is that you should design a complete, well thought
out, and useful product.  The user should not have to finish building
it for you.  Whatever customization you can add within those bounds is
great, but even then, stop and look:

	= What will customization `X' offer?  
	= Is there a way we can give that to user's *without* forcing
	  the user to play with options--sometimes you can get some
	  great ideas this way.

Remember, the Macintosh is simple to use, but getting it that way was
not simple.  Think very hard about user interface decisions you make.
Be wary of little, incremental, one-off changes which make the larger
design worse.  They are one of the causes of `featuritis'.  

Partial exception: You can sometimes reasonably pass engineering
decisions off to other engineers if they are your market and you are
building an engineering tool--but even then don't use this as an
excuse.  Even engineers can sometimes spot that.

Kent Borg
kent@lloyd.uucp
or
...!hscfvax!lloyd!kent