"Something is technically wrong" is the understatement of the year. It's ridiculous. As Michael Arrington said, if Twitter could keep the site live for 24 consecutive hours that would be news.
It's interesting. I'm reading Getting Real, the e-book from 37Signals (the guys who made Ruby on Rails on which Twitter is built). In it they're quite adamant about launching your product/site quickly and with minimal features - worry about scalability later. The whole time I've been reading it I've had the thought, "Oh yeah? What about Twitter?" rolling around in the back of my head.
I'm the first to admit I know zero about building web apps (I consider myself a consumer geek, not a back-end geek). But I do know that from a user standpoint, if your product crashes this frequently with no explanation as to why or when it will be back up, something's gotta change.
I don't know if the problem lies with Rails' inability to scale or Twitter's. But Twitter needs to figure it out and it certainly seems like they've had enough time to do so.