Beyond the marketing view... here are the (incomplete) changes:

  https://github.com/swiftlang/swift/blob/main/CHANGELOG.md

Here are the evolution proposals that landed in 6.3:

  https://www.swift.org/swift-evolution/#?search=6.3

Overall 6.3 ended up mostly about integration: stdlib, c/c++ (often driven by swift-java interop needs), and builds.

SPM (swift package manager) is slowly reaching out to take over xcode build features (e.g., resources most recently), and they have been inching and hacking their way to a new swift-build engine (now experimental) and "prebuilt" modules for the compiler/IDE macro support. Things Go Wrong all the time when trying to replace parts while driving, and there's very little visibility or control over the interactions between SPM and Xcode.

Conversely to integrations, endogenous language progress seems relatively quiet because there's a lot of behind-the-scenes work extending the initial blush of lifetime controls to iterators, stdlib types, etc. (never mind parameter packs). They're in year 4+ of building out the function-coloring of concurrency and isolation domains (6.2 had an "easy" mode that caused lots of problems), so adding year 2+ dataflow coloring of lifetimes is doubly complicating.

Naturally, diverse support for exciting new features results in quite a spread in the third-party library ecosystem of support for various language features. And now with AI entrepreneurs rewriting and porting any project that can score hits and kudos, the usual open-source authorities are losing their steam.

Finally, there's the baseline platform complexity of multiple OS and devices, physical and simulator debugging support, cloud CI...

So: that makes four dimensions of complexity. As a Swift developer you'd have to be very careful to stay on a working path as things shift underneath you.

All for the love of it?