So my question is: does modern Perl have some "global super-mode use utf8"? So it would be like use utf8 - once placed at the beginning. But the effect would be to kill in the program any memories of any encoding but Unicode. So literally would mean "in this program there is nothing but Unicode. You get Unicode, you output Unicode. There is nothing in this world but Unicode." Something like this.

Such a general switch is likely to break things. But first, let me clarify: Unicode is not utf8. Unicode is a concept that assigns a number ("code point") to every character considered worthy by the Unicode consortium. UTF-8 is a recipe to map strings built from those characters to sequences of octets - and interpret sequences of octets as a Unicode string.

This distinction is important because it makes clear that the concept of "Unicode", and of its encodings, only applies to text. Binary data in a Perl program are not Unicode strings, and binary data in files or read with LWP are no valid UTF-8 (most of the times).

Your code JSON->new->utf8(0)->decode($response->content) looks just wrong. You should investigate how the correct invocation somehow mangles things. Step by step:

• $response->content from LWP provides octets. The web server must specify the content type and encoding, and this is the encoding which must be used to interpret these octets, regardless from any settings in the program.
• The combination JSON->new->utf8(0)->decode(...) expects a Unicode string, and you're feeding bytes. So, these bytes are fed into your JSON structure.
• This would explain why you need to run $unicode_literal = decode('utf-8', $data->{result}) on your JSON fields.