Staying close to the cutting edge is easy when it's an incremental change. It's a very different thing when it means throwing out all of your your code.
Adopting Swift doesn't mean throwing out all of your code. You can have Swift classes in a mostly Objective-C codebase and Objective-C classes in a mostly Swift codebase as you wish.
I could be wrong, but I really, truly don't see anything about the new language that makes me want to start rewriting those billions of lines of existing code in a new language.
Who is talking about doing that? The conversation is about Swift being the dominant language for iOS development. If most new code is written in Swift, then it's dominant regardless of the fact that there's lots of legacy Objective-C code out there.
Apple is positioning this not as a replacement for Objective-C, but as a replacement for the Ruby/Python/Perl bridges
That's not even close to accurate. Read the Apple material, watch the WWDC videos, talk to the Apple engineers. This is not a replacement for a scripting bridge, it's intended to be the first choice for typical developers.
tell me why in the world you think that Apple won't take even longer to replace a non-temporary language
Once more, the conversation is about whether or not Swift will be the dominant language, not whether it will be the only supported language. Nobody is arguing that Apple are going to remove Objective-C support tomorrow.
Preliminary third-party analysis of Swift shows that for many simple operations, it is more than an order of magnitude slower than Objective-C. Assuming their testing methodology does not prove to be invalid for some reason
It already has for the most part, where have you been? Most of the benchmarks out there were run with beta tools, had different compiler switches, or other beginner level mistakes. Yes, there are some areas where Swift is slower, but most application developers aren't going to be significantly affected by that. Letting vague aspersions about performance dictate your language choice is nuts. Most of the time it doesn't matter and when it does, which language is faster depends on the exact thing you are doing.
if you're thinking about writing a major app in Swift, you should probably think twice
This is just FUD. Most languages are fast enough for typical use cases, and if you think you are going to be in the minority affected by performance issues, you really should consider Swift rather than dismiss it because depending on your use case, it could be faster for your situation.
start with Objective-C. That will let you get started working with real-world code now
Ah yes, my running streak of conversations involving the phrase "real-world" meaning "things I don't irrationally dislike" remains unbroken.
There is absolutely nothing that is not "real-world" about Swift.
The vast majority of what you learnâ"the frameworks themselvesâ"won't change if you later decide to switch to Swift. Only the syntax changes.
If you think that the only difference between Objective-C and Swift is syntax, then you really haven't given Swift more than a passing glance. It's a very different language.