I have it on flimsy authority that some of the problems facing Twitter's technical architecture were addressed long ago, and that somehow, the madcap idea was to let the problems linger as test of the community's loyalty, or rather, its "index of need".
By letting the problems fester while keeping the lion's share of what needs repairing in a back pocket, the management at Twitter has divined what few can answer via surveys, i.e., how much do people really want and need this service.
I have often said, (mostly to myself as I have no industry profile or juice), that the problem with most social networks is that they have no mission critical use case; well, now we know, what with all the articles and posts on nothing but Twitter downage, Twitter Architecture, Twitter Monetization, that Twitter does have a mission critical use case - whatever that is to each user.
Keeping the repairs in abeyance is a stroke of genius. A good friend of mine who writes messaging applications for BIG distributed finance systems says that, "This Twitter thing could just be done right as a ground up restart, we carry much more traffic with a far more complex transaction and message profile."