There are a few problems with errors-as-nodes in a syntax tree. An abstract syntax tree is a hierarchical representation of the program in the language's gramma

There are a few problems with errors-as-nodes in a syntax tree. An abstract syntax tree is a hierarchical representation of the program in the language's grammar - and errors are not members of the grammar (they're everything else!) There is also the problem that an error returned by a language server is a class of a "diagnostic" that includes syntax errors, semantic errors, warnings, lints, etc, associated with a span in the source code. It's much easier to think of diagnostics as a separate data structure that gets filled up during lexical/semantic analysis and associated with spans in the full syntax tree (you can even store them there as fields, but they don't necessarily have children). Then it's obvious how the structure gets created and fed back to the user. And finally, the whole point of an AST is to be a valid canonical representation of a program so the compiler query it drives doesn't have to do additional input validation. So it just makes the queries/compiler passes easier to write.