Windows 95: Doomed by Its Own Success

In the sense of security that was in place at the time, Microsoft would have said that security was a business/network issue, and that it was Windows NT that needed to be secure. Windows 95 was not designed to be secure, in the sense that it was not designed to be a part of a managed network.

This turned out to be beside the point, which was that the biggest security problems are not in how a system is properly used but how the holes in its design and implementation are abused by outsiders.

But imagine that Microsoft had a crystal ball and wanted to design Windows 9x to be more secure.

Very, very few people at the time were greatly concerned with addressing vulnerabilities in their programs to overflows in the stack, even fewer in the heap, or bugs that could result in a privilege escalation. (Of course, there is no such thing as privilege escalation in Windows 9x, since everyone has ultimate privilege.)

Addressing these matters at the beginning would have been really hard, and the problems would have begun with the incredulity with which Microsoft would be greeted in claiming that the Internet presented a nightmare for computer security.

Undoubtedly such analysis would be viewed as a scam to sell security software. And Win32, the API, and other guts of the operating system, had been under development for many years.

And of course, Microsoft was focused, as was everyone else at the time, with building the applications rather than making them secure. The really hard-core gold rush didnt begin for a couple of years, but it was definitely there at the time, and security was a distraction from it.

Nobody wanted to believe in the security problems. They would much rather believe in nightmare Y2K scenarios that, after all, were useful for selling new hardware and software. And most of this new hardware and software was made possible by Windows 95 and its descendents. And so were the problems that sat there, just under the surface of the programs.

The first time I remember security getting serious ink (we still used ink back then) was in the ActiveX vs. Java arguments, and more generally about Java.

Once again, the whole thing turned out to be largely beside the point. ActiveX has not actually been a major security problem and even if we accept that Java would be a very secure one, it was way ahead of its time in terms of system performance.

Click here to read more about ActiveX and security from Larry Seltzer.

To this day its not really considered acceptable in terms of performance for large applications. (Ask yourself why Sun doesnt use it for StarOffice.)

I wrote the first hands-on review of the Win32 SDK based on what would become Windows NT in my hotel room at the Win32 PDC in San Francisco in July 1992.

It was based on work that had begun years before. Would Microsoft have been reasonable in imagining IRCbots and phishing scams back then?

Clearly they could have foreseen things being worse than they were, and they might have foreseen many of the problems that developed.

But 10 years ago the PC industry had a lot of growth ahead of it, and Microsofts talents lay in driving that growth, not in prophylactic measures against theoretical problems, the addressing of which would only slow down product development.

I suspect that even if they could have seen how the next 10 years would turn out, the wouldnt have done much different, security-wise.

Security Center Editor Larry Seltzer has worked in and written about the computer industry since 1983. He can be reached at larryseltzer@ziffdavis.com.

Check out eWEEK.coms for the latest security news, reviews and analysis. And for insights on security coverage around the Web, take a look at eWEEK.com Security Center Editor Larry Seltzers Weblog.