Agreed. The CLI of gpg is horrible. There are some semi-acceptible GUI variants, not least Enigmail, and a good UI is is definitely going to be required if you are going to get general acceptance.
But the main reasons it continues to not get used are
0) Math* is hard!
1) The rise of webmail
2) Inverse network effects
* encryption being a subset of math.
0) It's hard to explain to people that they need encryption, how it works, what it is. People think email is secure! The "envelope" iconography is very misleading - email is more like a postcard, delivered by a random selection of disreputable postmen.
1) Webmail makes it much harder to do encrypted mail because to make it secure you'd have to install browser plugins. None of the webmail providers want to make one, because it will destroy their revenue stream of monetizing the analysis of your mail traffic.
2) If you want to actually use (G)PG(P) your recipient also has to grok it, install software to use it, and you have to exchange keys. This is a massive hurdle to overcome for all but the most dedicated cryptonerds. Until there is a majority of people who want to use encrypted mail, that will carry on being the case.
There are projects attempting to overcome some of these hurdles ; you have the likes of keybase.io that takes some of the sting out of key exchange (and verification).
But!
Until encryption comes with the communications software you are using out of the box, is enabled by default, interoperates with everything properly, and forces you to configure it to even use it, the vast mass people won't use it. And this is well known by the SIGINT agencies who view people actually using encryption AT ALL as a red flag that they should look closer at.