Originally posted by: STaSh
Not every end-user wants to willingly prostrate themselves before their corporate masters
And not every end-user is a paranoid tin-foil hat wearing obsessive compulsive. Most end-users just want to use their computers.
Which means that they wouldn't have need of a host-based software firewall in the first place, just buy a router and be done with it. However, even most non-technical people, when presented with the information that MS is potentially logging every media file that they watch on their PCs, and their printer driver is monitoring the relative levels of ink used in everything they print, they tend to get offended regardless. They don't just hold up their hands and say, "oh well - that's life with an MS-controlled PC".
Originally posted by: STaSh
Except that the only type of unknown traffic an outbound firewall can reliably alert you to is unimportant (see above). As I have stated time and again, an outbound firewall on a host cannot be considered reliable for detecting viruses, worms, spyware, etc.
Not 100%, no - I agree with you there. But to make the counter-assumption, that it is 0% useful, would require that the malware running on the machine would be 100% perfectly stealthy. Given that most of it isn't, and in fact, some of it isn't even 3rd-party malware, but spyware components pre-installed as part of the OS itself - most of it is detectable. (Referring to Win XP here, mostly.)
Originally posted by: STaSh
Read my post again. Yes, the SC is spoofable, but only with administrative access. A software firewall is spoofable in the same case..with admin access. So they are both of dubious value when running as admin, which is the case for anything running on your machine when you are admin. So what are we left with? We then need to determine their utility when running with least privilege. Since the risk of self-propagating worms, viruses, etc is greatly diminished when running LUA, the usefulness of an outbound scanning firewall is negligible (other than for paranoia, which I've already addressed). In the same LUA scenario, the SC does offer value to the user, namely by telling them if they are potentially out of date with critical updates by not having AU enabled, and by telling them if their inbound scanning firewall is disabled.
Uhm, that totally ignores the possibility that said malware, might be able to obtain SYSTEM or Admin privs automatically, via one of the various many "local priviledge escalation vulnerabilities", as have been found and documented in the past, and I'm sure that there are still a few lurking in the shadows of the codebase somewhere.
Thus, in the presence of malware running on the local host, it could potentially have "root" privs regardless, and therefore *any* software-based defense mechanisms running on the host could also be suspect. In fact, it's a bit ironic, that just because of this small possibility, you are willing to disregard the utility of said software firewalls in the general case; that's a far more paranoid assessment of the situation than I present. Who is the real wearer of the tinfoil hat here? Just because spoofing is a technical possibility, doesn't mean that it will happen in 100% of the cases, and thus render software-based defense mechanisms
completely useless. There's a fairly wide gap between what is theoretically possible, and what is realistically likely to happen.
Not to mention, running as LUA only protects against user-assisted trojans/viruses, it doesn't protect at all against network-borne worms, that directly attack exploitable holes in the networking components, which run as services with some variant of SYSTEM privs. So LUA doesn't do diddly there. So that's not a valid component of an argument either.