At 07:59 AM 04/08/2002 -0600, Anonymous wrote: >"Any attacker who can control 100,000 machines is a major force on the >internet, while someone with a million or more is currently unstoppable: >able to launch massively diffuse DDOS attacks, perform needle in a >hayfield searches, and commit all sorts of other mayhem. We already >understand how worms could be used to gain control of so many machines. >Yet the recent revelation that Brilliant Digital Media has bundled a >small trojan with KaZaA has underscored another means by which an >attacker could gain control of so many machines: poorly secured >automatic updaters. If an attacker can distribute his own code as an >update, he can take control of millions of machines. " > >http://www.cs.berkeley.edu/~nweaver/0wn2.html > >So, now, how hard would it be to use this mechanism to upload PGPNet >with opportunistic encryption enabled to millions of hosts ?
Do you mean "How hard would it be to crack into Brilliant Digital's servers before some other SKR1P7 K1DD13Z take it over"? Or do you mean "Is that easier than cracking into Microsoft or Adobe or M0Zilla or some other quasi-reputable company's distribution system?"? Actually using it to upload PGPNet would probably be pretty hard - it's no longer just Phil's ~200KB of badly-written MSDOS code, it's now 5-10MB of bloatware (:-), and you can't distribute a few million copies of a few megabytes to unsuspecting users without somebody noticing. Also, leaving aside the "opportunistic encryption" issues, which depend on having working secure inverse DNS for the FreeS/WAN flavor, you can't depend on tunnels working through firewalls or NAT or other arbitrary connections out there, so a lot of recipients wouldn't really get to have it working for them, but it might break quite visibly - especially for people who already have VPNs, and therefore usually have corporate IT support or corporate security departments who'll notice it. Better to just build a nice small ipsec client into a flashy MP3 player :-)
