🚀 go-pugleaf

In article 1ghs93INNl04@iraul1.ira.uka.de, s_titz@ira.uka.de (Olaf Titz) writes:
>In article <WCS.92Dec13203554@rainier.ATT.COM> wcs@anchor.ho.att.com (Bill Stewart +1-908-949-0705) writes:
>>
>>The main job of security is to say "NO".  
>>The main job of Unix is to say "Yes".
>
>O.K., you agree in this point: The main job of Unix is to get the
>user's work done.
>The main job of security is to stop the user from getting his work
>done.

This is a ludicrous statement.

Some of the reasons for implementing security on a computer is to protect the 
information processed and/or stored on it from unauthorized disclosure; protect
the information processed and/or stored on it from unauthorized additions, 
deletions or other changes; and protect the machine from being misused, however
the organization that owns it chooses to define "misuse".

>
>=> "Security" as defined such belongs on the crap heap and not in
>computers that are to be USED, imho.

Your opinion.  Not everyone else in the world shares it.

>
>>The other main job of security is to keep track of who did what to what,
>
>To spy on the users.

Lord have mercy!  Paranoia will destray ya!  Has it occured to you that audit trails 
can be used to figure out what brought a system to it's knees and provide clues as
to fixing the problem?

>
>>	so that if you decide the system *should* have said no, but didn't,
>>	you know what got stolen.
>
>But then it is too late and you have caught a bug, perhaps, but not
>the intruder, according to your 'security' definition.

And isn't this worth knowing?

>
>(And, mentioning that word reminds me of the following: The sometimes
>arbitrary redefinition of an otherwise legitimate user as 'intruder'
>makes me at least raise an eyebrow.)

And I've seen "legitimate" users do some very strange things that instantly
classify them as intruders.

>
>>...
>>to fix if you *know* the identity of the real user, 
>>and that sometimes it's genuinely hard to know who the real user is,
>
>Do you ever know who the 'real' user is...
>
>>especially across a network where you don't control the other end.
>
>....even on your local machine?

True.  And sometimes "real/trusted" users do things that look suspicious, and
turn out to be a legitimate action.  This is a problem that security is trying to
address, admittedly with mixed results.

>
>>One way to resolve the latter problem is to always trust the user;
>
>That from *you*? ;-)
>
>>another way is to never trust the user; a middle way is to work hard
>>and try to give you some control about how much trust you have.
>>
>>(Then of course, there are bugs and misfeatures, and the relative
>>quantities of these in Unix vs. other OS's is beyond the scope of this note :-)
>

<stuff about rogue sys admins deleted - no argument, it *could* happen>

>
>>But it is a lot easier to do security, and everything else,
>
>Security, as defined above, means unusability - for this I don't need
>any OS at all. I switch the machine off.

Not necessarilly.  You don't have to turn the machine off to have a secure system.
Just make it single user and unconnected in any way from another machine.  This
does not imply unusability, mind you.

<more stuff deleted>


---
                __
           /  /   \
          /  /    /
   /     /  /    /
  /_____/  /____/

John de la Garrigue                              ||  Phone:  619/546-6192
Site Manager                                     ||  E-mail: jd@c3ot.saic.com
Science Applications International Corp. (SAIC)  ||  

My views are not necessarily those of my employer, the US government,
or anyone else for that matter.