Languages and abstractions aren't really that different. Every time you define a new function, you're in some sense extending the language. In languages with

Languages and abstractions aren't really that different. Every time you define a new function, you're in some sense extending the language. In languages with powerful macro systems, this "language extension" is even more potent. I agree with you that good abstractions (architectures are just large-scale abstractions) are needed, but that doesn't mean the new abstractions won't come in the form of a new language. In some sense, they always will. In the sense you mean, some abstractions fit better into existing languages than others.