The suggested query indeed appears to be faster. Thank you.
> Did you try if a properly normalized model performs better?
I've tested the below schema, which doesn't appear to perform much better but
has a couple other advantages for my application:
```
create table subscriptions (
email text references users(email),
ucid text,
primary key (email, ucid)
);
explain (analyze, buffers) select cv.* from channel_videos cv, subscriptions s
where cv.ucid = s.ucid and s.email = $1;
```
Is there something else here I'm missing?