Yeah, it's annoying. But theoretically it's a more integrated experience, it's used by all Gmail users - no need for people to get Skype, which is good for people who don't like installing programs - and it's pretty impressive for an online implementation.
Or they could use an open standard protocol, as could Skype, which would be good for people who don't like vendor-locked-in proprietary technologies.
(Which seems to be very few people, unfortunately. By the way, I spent some fun fun fun minutes today trying to recover my broken Outlook NK2 file. A proprietary secret format to store a list of email addresses with names? Hello?!)
"And in the spirit of open communications, we designed this feature using Internet standards such as XMPP, RTP, and H.264, which means that third-party applications and networks can choose to interoperate with Gmail voice and video chat."
What this blog post doesnt say, is that google completely ignored the perfectly valid current Jingle draft for RTP video (XEP-0167) and instead invented their own (and doing very weird things like signalling audio+video streams together, although they're different streams).
They also use a variant of H.264 that basically nobody else uses (H264/SVC), for which there are no free codecs available, and provide no fallback to some other codec for interoperability reasons.
Google might have the best intentions, but until they adress these issues (well, too late for the signalling, but the codec problem is more important), they can't seriously talk about the spirit of open communications.