Confused by an SSL related blog post

I subscribe to the Planet SysAdmin, an aggregation of Sys Admin related blogs via RSS (http://planetsysadmin.com/). Today, I came across a post that really confused me on what the author was trying to accomplish or explain.

http://utcc.utoronto.ca/~cks/space/blog/web/RevocableSSLNames

I'm hoping that one of you that is smarter than me gets it and maybe can explain it to me. Using his example of online banking, I look at my bank and it looks like the cert is probably handled by a load balancer; or maybe explained in his second case, an HTTPS Proxy server. He discusses revoking the cert by changing the host name.

A) How would that help when I have a bookmark to https://onlinebanking.example.com?
B) and wouldn't you revoke the compromised cert with the certificate issuer anyways?

The only thing I could think of is he's discussing SSL on both ends of his proxy (web client to proxy, proxy to application server). But in that scenario, the only thing that is apparent to the end user is the SSL cert to the proxy which would have to be revoked by the certificate issuer if compromised.

Or, I'm missing something entirely...