The deficiency model of translation holds that the translation will be inferior to the original. A good translation is one that preserves as much as possible, minimizing the losses, but the idea of "preservation" is itself part of this model. We could call it the "preservation / loss" model, maybe. According to this theory a good translation would preserve a good portion of the original, maybe 90%.
Suppose we judged film as translations of written texts, their scripts. So then a good movie would be one in which the brilliance of the screenplay was preserved, not lost. That would be silly, though people do judge a movie as "not faithful to the book." So the f*** what? Who cares? The movie is its own thing, so why does anyone care about fidelity to some novel of the same title?
A new car loses value the minute it is driven off the new car lot. So we might say a translation is automatically less valuable because it is a translation. Say it is a 20,000 dollar work of art; the translation can only be worth 18k even if brilliant.
So the model is that the original is a sacred text of some sort. Because this model only works if we invest the original with a magical, inalterable character. Borges claims that Roy Campbell's "When all my house was hushed" is better than the line from Saint John of the Cross (of which it is a translation): "estando ya mi casa sosegada." He also says that "ars longa, vita breve" is inferior to Chaucer's "The lyf so short, the craft so long to lerne." In that word "so," he finds "the very music of wistfulness."
The corollary of this, though, is that we should hold translations to a higher, not a lower standard. A translation can be inferior, equal in value, or superior to the original. Given that, it is a perfectly valid critique to say that the translation doesn't match up. Not because it hasn't preserved some value in the original, but because, on its own terms, it isn't good enough. The same way a movie can be bad, not because it fails to measure up to the promise of a screenplay, but simply because it isn't good.
It is vile lie to say that a translation is "a good poem in English in its own right," when it's not. Logically, a bad poem cannot be a good translation of a good poem.
Now this assumes that what we are looking for is not the preservation of the pre-existing characteristics of the original, but rather the creation of a text that has equal value or excellence. So my translation might be an excellent poem, but in ways vastly different from the way in which the original was a good poem. In other words, 'when all my house was hushed" does not derive its value from San Juan, or from conserving something that was there in San Juan.
I believe Wyatt and Surrey translating Petrarch felt this way too.
I believe that way people think about translation is deeply fallacious.