« Linux upstages Netware | Main | Don't say you've got it unless you've checked first »
August 17, 2003
How can I trust the Internet?
I have just finished reading "Rethinking the design of the Internet: The end to end arguments vs. the brave new world." This paper revisits the original End-to-end paper written by Saltzer, Reed, and Clark in 1984, which was fundamental to the design of the Internet (especially the Internet Protocol and the Transport Control Protocol--TCP).
Though Rethinking addresses many issues which threaten the end-to-end principle, the most serious one is that of trust.
Most of the discussion and development regarding trust on the Internet has been on technologies such as PKI for authentication and encryption.
In theory, PKI and similar technologies are the right direction. But there is a more serious threat to trust, that is undermining the general confidence in the Internet as a viable long-term communications medium.
As discussed in Rethinking, the first implementation of trust on the Internet was with the TCP checksum. The receiving host can validate the trust relationship by checking the checksum to verify what was sent. This is a fairly simple but scalable trust mechanism. TCP and the underlying Internet Protocol have allowed millions of computers and hundreds of millions of users to develop spontaneous associations with enough trust to interact effectively.
Trust on the Internet is much more complicated and diverse today. We have SSL, double-opt-in spam, RBL's, route registries, PGP, encrypted and authenticated IM and email, and more.
But in spite of all these trust mechanisms, trust on the Internet is more threatened now than ever. The miscreant community has repeatedly undermined public confidence in the most basic and pervasive implementation of trust: the Microsoft operating system.
Every time another Microsoft vulnerability is exposed and exploited, tens of millions of people lose confidence in their ability to productively use the Internet. Yes, users of other operating systems have the same experience, but those incidents happen far less often, to far fewer people. Relative to the impact of Microsoft, they cause an insignificant amount of damage (and more likely provide some hope for those Microsoft customers looking for alternatives).
As valuable as the Internet is, home users, small business owners, and others who can't afford full-time staff to watch for these incidents, defend against them, and clean up the mess afterwards--they are the ones who with each loss of data, customer information, tax records, billable hours, from whatever latest virus or worm is circulating now to exploit one of the many Microsoft vulnerabilities, question the value of being on the Internet, and whether they're better off being less-connected if only to be less-vulnerable.
I use Windows (desktop and laptop), and I take great caution in keeping it updated and protected from the network. It's a fair amount of work, that I doubt the average computer user could keep up with. In spite of that, during the most recent RPC worm, I have been scared that I have overlooked some patch and my machine is still vulnerable (as happened to many sysadmins with the SQLslammer worm).
I get much better sleep knowing that my most important machine runs Linux. Linux has certainly had it's security issues, but I personally feel the development model behind Linux, it's transparency to the general public, and it's track record inspire confidence that it is an operating system to be trusted. I know that having been a user of Linux for only a few years less than I've used Windows, I feel Linux merits my trust far more than Windows does.
Maybe Linux and OSS are the counter-balance to Microsoft, giving users confidence enough to stay on the big, bad Internet.
Posted by pete at August 17, 2003 12:08 PM
Trackback Pings
TrackBack URL for this entry:
Comments
Patching Windows is "a fair amount of work"? Windows Update (the web page, not the little thing in the system tray) is about as easy as up2date. I have both XP and Redhat 9 on my desktop and get about the same number of patches for each system.
The reason people don't patch is that patching systems are built for programers. They ask a bunch of questions that only a programmer/sysadmin would know how to answer.
Posted by: Allen at August 19, 2003 09:35 AM