I'm one of the authors of this piece and I'd like to make clear that it responds mainly to specific suggestions that end-to-end encrypted chat tools could be modified in very small ways, without impacting the overall communications protocol or requiring the tools to be substantially redesigned.

Since these tools are not in general obfuscated to hide their activities from the users and don't perform extra work to conceal things like how many recipients there are for each message or session, there are lots of possible avenues for noticing anomalies.

We aren't arguing that there is no way to create communications tools that prevent users from detecting if their communications are being intercepted. Instead, we're arguing that those tools wouldn't be simple, straightforward variants on existing end-to-end encrypted messaging systems.