Computer Security : What are we trying to do here?

	Current approaches : The sexy hardening methods (SSP, non-executable 
stack [the favorite NX-bit], ASLR, JVM/CLR models)
	 	Flaws in these models
	That there are flaws in these models demonstrates we're doing 
something wrong; by playing an arms race with attackers we're losing the 
battle before we start.
	If we instead ofcus on removing any possible benefit from exploiting 
a process by granularly dividing privilidge across all users and services 
running on a machine, the exploitable profile of the machine is lessened
	If we perform active and real time auditing and monitoring of both 
privilidged activites and transferrals of privilidges within the system, and 
terminate processes and users that violate this security model, we even 
further enhance our security.

	 File and process ACLs: Why the UNIX model sucks
	 Side by side comparison of the UNIX file permissions system to the 
NT permissions system -- this is why people look at me strange when I tell 
them how NT has a better security model
	The roles of users and groups and group memberships, given a great 
degree of granularity in NT, blows UNIX away.
	Obviously this is unacceptable, so how do we get back on the 
playing field?

	Different peoples approaches to the problem : POSIX.1e and the 
Darwin/OS X (oh.. add that to the list of stuff to look at if I get it 
running on my laptop ;) ) standards -- POSIX.1e moving one way and 
Darwin/OS X / Samba moving more after the NT model of priv enumeration

	File and process ACL comparisons across multiple platforms : 
how SELinux and freeBSD do it with contexts and labels, how the 2.4 cap 
system got left out to dry .. somewhere in here cover the politics behind 
the introduction of LSM into the linux kernel, basically hit all the hard 
and fast points of how you interface, as a user and administrator, with 
the ACL system and the creation of a security policy on a few different 
types of machines ( server, desktop, mail server, file server, web server, 
shell server, combinations of the above)

	Examination of the combination of secure engineering and design 
with system hardening -- RSBAC can combine its RBAC functionality with 
that of PAX's system hardening ..

	Examination of overhead of both hardening and RBAC systems ... 
you're storing extra data on a per file and process basis and doing 
additional checks, this part might be fuzzy.

	RBAC Is Not A Silver Bullet -- Know your systems in and out and 
backwards and forwards. Server administrators should either be the ones 
creating security policies for their machines or be working very, very 
closely with security officers in creating a policy that both locks down 
the system and keeps it usable -- your bottom line sees no difference 
between a server shut down by a draconian policy and the bad guys. someone 
will probably be calling YOU a bad guy ...
	The Similiarities between good security systems and rootkits - What If?
	A rootkit based entirely off of existing security frameworks? 
The potential for misuse of LSM and security "features" in the kernel 
that are exported by default ...