This emphasis on the libretto as a separate entity with its own literary style perhaps leads Smith to underrate the importance of the nuts-and-bolts craft of libretto writing -- structure, pace, suitability of the words for music -- and to underrate librettists who didn't have any particular style but were able craftsmen. One example would be Lorenzo Da Ponte, a decent poet and bon vivant who wrote three superb librettos for Mozart, Le Nozze Di Figaro, Don Giovanni and Cosi Fan Tutte; they are quite a bit better than his work for other composers, leading to the reasonable conclusion that Mozart demanded, and got, more out of Da Ponte than other composers could. Still, it wasn't Mozart writing the words. Similarly, Verdi browbeat and bullied his most frequent librettist, Francisco Piave, until he had just the texts he wanted; but the fact that Piave wasn't the most distinguished of writers doesn't take away from the fact that, acting on Verdi's demands, he came up with some really excellent librettos (libretti?) like Rigoletto and Traviata. These are librettos that capture the essential qualities of the plays they're based on while condensing the plots; they make sure that the arias and duets arise logically from the situation instead of feeling stuck in; they give a general and strong impression of who each character is and what he or she wants, so the composer can build on that in his music.
This isn't in itself the stuff of great art, but it is difficult to do well. And it has to be done well, because an important rule of opera is that almost no opera succeeds with a bad libretto. The librettos that are often cited as bad are really anything but. Il Trovatore has a somewhat insane plot, it's true, but it is based on a then-popular stage play which had the same plot. In other words, the reason Trovatore's plot strikes some of us as ridiculous is not that Salvatore Cammarano's libretto is badly written (it is, in fact, very well written, cleverly structured and with some good lyrics) but that it uses a type of plot that has gone out of fashion; the original plays are no longer performed, but the operas remain.
To see the importance of a good librettist, look what happened to some of the great opera composers after they lost their best librettists. Puccini's three biggest hits, Boheme, Tosca and Butterfly, were all written by the team of Giuseppe Giacosa and Luigi Illica. (They weren't really a team: Illica was a fine dramatist but a lousy poet, and Giacosa was an excellent poet without much dramatic sense, so Puccini's publisher put them together: Illica created the structure and wrote a sort of prose version of the libretto, and Giacosa turned the outline into rhymed verse.) Giacosa died after that, Illica concentrated on other projects, and Puccini's next opera, The Girl of the Golden West, was hobbled by a pointless and rambling libretto with some geniunely stupid lines ("Whiskey per tutti!").
One more thing: I think that Smith perhaps overrates Wagner, whom he calls the greatest librettist ever. Wagner was certainly a talented writer, one of the few great composers who was also a first-rate librettist (Berlioz was another). His librettos dealt with complicated issues that few librettists had ever dared to tackle; the monologue for Wotan in Die Walkure is a rumination on fate, morality and power where most opera librettists would just write a few lines about what a bad day the character is having. But by writing such long, rambling passages in his librettos, Wagner started the bad practice of over-writing in librettos, a practice that helps to explain why so many 20th-century opera librettos have been so bad. Traditionally, opera librettos had a lot of words in recitative but short, simple words for arias. By the 19th century, even the recitatives were short and concise (Verdi in particular insisted on this). The point was to have simple words expressive of one particular thought or emotion, that the composer could then expand upon. These words, by the French librettists Meilhac and Halevy, don't look like much:
Toreador, en garde,
Et songe en combattant
Qu'un oeil noir te regarde
Et que l'amour t'attend.
But when Bizet sets it to music in Carmen, repeating a lot of words in order to make them fit his tune, it works. Wagner didn't like to repeat words, though, and he didn't like to have the characters express one emotion at a time; the point of the leitmotif system he developed was to follow the changing moods and feelings of the characters, and that in turn meant they had to have more words, with more changes of mood. It worked for Wagner, because the words were of a piece with the kind of music he was writing. But a lot of post-Wagner librettists adopted his verbosity even if it wasn't necessarily appropriate to the kind of music the composer would be writing. The libretto for Schoenberg's Moses and Aaron, which Schoenberg wrote himself, strikes me as having this problem; many of the monologues and choruses are very wordy and go on longer than they need to make their point -- not a great match with twelve-tone music, whose raison d'etre is to be more concise and less repetitive than traditional music. (The best and least wordy parts of the libretto are the parts for Moses; Schoenberg couldn't fall into the temptation to write too many words for him, because Moses is supposed to be someone who can't and doesn't say much.)
The last point I'll make is that the best librettists are the ones who are the best at tailoring their work to the composer's strengths (at least where the libretto is written for a particular composer; many librettos were written first and then shopped around, and even set by more than one composer; Metastasio's librettos in the 18th century were set by dozens of composers). A good poet can be a bad librettist because the poet may be concerned with writing words that sound good on their own, without regard to whether they will sound good when sung. A good playwright can be a bad librettist because the structure he creates may provide insufficient space for arias, duets, or other moments where the composer can do what he does best. That's why, even though composers do it all, the success of a librettist is ultimately judged by how well he allows the composer to do. One of the most famous librettists of all time, Arrigo Boito, was a poet and a playwright and a composer in his own right, and given to over-verbosity; his librettos teem with archaic words and obscure, pretentious allusions that must annoy the average Italian audience. But when writing librettos for other people, he made sure to build the libretto around the composer's strengths, rather than making it a showcase for his own strengths. His libretto for Verdi's Otello carefully finds room for Verdi to do all his specialties -- a storm scene, a drinking song, a love duet, a huge ensemble scene -- while making all these things seem dramatically inevitable. That's why it's Verdi's Otello, not Boito's; but Boito's achievement lies partly in creating a libretto that allows the composer to do his best work.