In some cases he's optimistic IMO, in others sensationalist. Biometric auth: yes, and thank God. SSL: TLS is not much better. SSH with pubkey auth, on the other hand, is the only thing that works. HTTPS is not as trustworthy as people think. (Note - most SSH hackings I've heard of have been due to misuse of password authentication. And the stuff against OpenSSH is a disgusting example of FUD IMO. Properly used, it is far more trustworthy than HTTPS, and the daemon itself has AFAIK had less frequent holes than e.g. OpenSSL. Which, BTW, is a whole different piece of software.) Pubkey encryption: "when" -> more like "if." Quantum computing is nontrivial. (Repeat after me: The Singularity Will Not Happen.) IPSec: well yeah, except that HTTPS MITM proxies exist. With automatically trusted certs, even. Firewalls: TCP/IP stack vulnerabilities are not common, but have happened. Also it's handy to be able to drop stuff at a packet level. I don't think kernel based, packet filtering firewalls are going to vanish - they're useful to have in case of e.g. DoS attack. Antivirus: I wish, but don't underestimate the power of money. Also Grimes is waaay too optimistic about whitelisting. Antispam: hard to say. From what I've personally seen, spam filters are doing much better than they used to. Anti-DoS: firewalls, anyone? Also I don't believe that "Put them in prison!" is a legitimate solution. The punishment should suit the crime IMO. Event logs: I could make some interesting analogies here with what the NSA is doing. But yeah, IMO this is not doomed, it's a number crunching issue. I'm betting we develop better expert systems for dealing with such logs. Anonymity and privacy: the situation is grave, but I would not write it off as "doomed", if only because we shouldn't go down without a fight. ... All IMO, anyway.
I don't see AV's and firewalls going anywhere. Yes they are not perfect, but still too important. Especially outbound firewalls are very useful, almost all malware relies on outbound connections.
