Let me start by saying it's not going to happen, no way: an editing feature won't even make it into Sippey's or Dorsey's cutting-room, let alone onto the cutting-room floor, for the simple reason that it isn't simple, and the whole aim of Twitter product strategy is ruthless editing of features toward this end of simplicity. Powerful primitives, but few verbs, and fewer nouns.
And let me continue by pointing out that it is obviously a bad idea, because it misses the whole point and power of Twitter. Twitter is ephemeral, it is lighthearted, only The Now has prominence.
You can say Twitter is like a conversation, and it is except in the ways in which it isn't. Yes, utterances are brief, additive, and ordered in time, they sometimes have a call-and-response (@reply) nature, context is often essential to understand the meaning of an individual utterance, older utterances have less prominence (in the fading of memory, below the fold).
But how is Twitter not like a conversation? Utterances live forever, they are durable, inscribed into twitter's various data stores and search indexes and map-reduce clusters. (Unless you delete them of course--then they only sort-of live forever).
To cut to the chase: we're dealing with a new medium. It has some aspects of conversation, and some of broadcast, some aspects of live television, and some of a(n edited) newspaper. But it is none of these things all at the same time.
Telling the news on Twitter is different than telling the news in a magazine or newspaper. I realize journalists have a difficult job these days. The way mistakes are made and disseminated and the way they are corrected, is utterly different on Twitter than at a magazine like Wired or a newspaper like the New York Times. This places unfamiliar demands on journalists and novel demands on consumers of news. And the bigger burden is on the consumers, which I imagine makes the journalists especially cross. Because if we consumers want to have a real-time account of events--and we do, it really makes us a better informed citizenry--we have to understand how to deal better with ambiguity.
Consumers don't just have to be "skeptical" or "critical thinkers" of breaking information: but they themselves have to operate as do journalists, by e.g., waiting for at least two independent sources as confirmation, and even then realize a piece of news only has some higher probability of being true. Tweets about older events have a lower threshold for warrant than breaking news, for obvious reasons. The price of timeliness is eternal vigilance.
I can understand the temptation to want to edit some (perceived) egregious fallacy you accidentally helped perpetuate, but that's not how things work on Twitter. Delete the tweet, tweet a correction, or write an elaborate apology on your blog. It will harm your reputation to make a careless error, but on the other hand the audience should know to expect corrections when who-they-follow switch to the breaking-news game. And the audience wants breaking news, warts and all.
This doesn't even mention the obvious (and nontrivial) engineering challenges of moving from being able to assume that posts are immutable (and hence once they're written they're written and can be propagated) to actually having to worry about which version of a post you're passing around.