QUOTE
> 98 runs very well with an alternate browser.
And it runs very well with IE6, and in some cases (like booking air travel) you need to use IE6.
It runs fine until it uses up your resources and forces you to reboot. The only pages I've had any problems with are those using ActiveX. Outside of MSN, they've been very few.
QUOTE
> leaving bad decisions by the user as its worst vulnerability.
The characterization that the user is usually responsible for web-based malware intrusion is wrong. There is DNS poisoning and server-farm hijacking that can result in infection that has nothing to do with the judgement of the user.
Yes, there are attack vectors that don't involve user interaction, but they are not responsible for the majority of the compromised PCs. The vast majority of infections are caused by something the user chose to install, open or click on. The next most common sources are weak Internet Explorer settings (and users who run it with those out of the box settings) and browser exploits, specifically IE6. Yes, there's other ways to infect a system, but the user and IE6 are responsible for most of the problem.
QUOTE
The use of older versions of NAV (Norton Anti Virus) - say, the 2001 and 2002 versions, make a suitable AV solution for win-98.
A five year old AV, Norton of all things, then tell me that I'm the one being foolish? That AV uses more disk space and resources than my entire security package combined. The last virus I've had was one allowed by the exact AV you suggest, and it was up to date. Every infection and system compromise I've experienced happened while using Norton products. Never again will anything Norton be used on any PC I own or maintain.
QUOTE
Win-98 doesn't need protecting from unsolicited incoming requests. In other words, win-98 doesn't need the protection of a firewall UNLESS you've enabled file-sharing on TCP/IP.
I mentioned that when I said to close the NETBIOS ports. Aside from that, 98 itself might not need firewall protection but the installed apps do.
QUOTE
As for un-authorized out-going attempts, such an attempt would have to require that your system is already infected with something, and most likely that something will have already turned off your software firewall anyways, so again the usefullness of a software firewall is pretty close to zero.
Most of that malware doesn't run on 9X systems. On a PC with a default-deny policy enforced in the manner I suggested, malware doesn't run, period. If it can't run, it can't infect you.
QUOTE
Win-98 has vulnerabilities because of the integration with IE (and those vulnerabilities don't go away just because you're using another browser like Firefox). But while the vulnerability exists, the exploit code probably won't function properly on win-98 like it was designed to do.
Then you do admit that IE is responsible for most of 98s vulnerabilities. In all fairness, most of the recent exploits involving IE6 do nothing to a 98box. With the next one, who knows? Getting rid of IE gets rid of most of the vulnerabilities in a 9X box. If a user wants or "needs" IE, they should at least tighten up its settings. Ideally, they should run it thru Proxomitron and filter out the unwanted content.
QUOTE
The big con with that argument is that M$ and IT people are really talking about desktop or login security (the ability to turn on a computer and access it's files or use it to access the network connected to it). They say that Win-9x is insecure, but that's what they really mean. It means they can't control who is using the computer.
Out of the box, 98 did little to control who uses it. That can be largely fixed too. The NT systems are better at user control, but that came with a whole new set of vulnerabilities to external attacks and the ability to hide files and infective processes from the OS and the user, the rootkit. A very bad trade.
QUOTE (eidenk)
QUOTE (herbalist)
As for the leaktests, using them to compare and promote one firewall over another is a gross disservice to users. The entire concept is flawed.
How flawed is that :
It starts with how these tests are treated. The users chooses to start them and expects the software to contain them. Their responses to the security prompts are influenced by the fact that they know it's a test, such as allowing the initial process but blocking the hook it tries to set, then thinking they passed the test. The test never ran. If that "test" was a piece of malware bundled into an install, they'd probably have allowed the hook as part of the normal install process and be owned by someone. Responses to alerts don't reflect real life behavior because the user knows it's a test. Using the pcaudit2 leaktest as an example, most users will allow the process then block the hook. All that does is test if their firewall or separate HIPS will block a hook when they tell it to. On most systems, if the hook is allowed, the system will fail the test. The site leads the user to believe that they failed the test because their firewall allows the hooks. Their firewalls actually fail because loopback connections aren't properly controlled, poorly configured. The result is an emphasis on hook control features, aka HIPS over properly written firewall rules and system configuration. The user needs a better firewall instead of learning to configure the one they have. That is flawed. Look at his results. Only the newest "Pro" versions do well. My system passes that test and most of the ones that apply to 9X with an old firewall, Kerio 2.1.5 and SSM shut down. I don't agree with his reasoning that application control should be part of the firewall. IMO, both traffic and application control are important enough that they should be controlled separately to reduce the risk of common vulnerabilities taking down both at once. On a combined package, one piece of vulnerable code is enough to make that possible.
Rick