I think what the Pro ASP.NET MVC book means is that, by referring to a route by name you have now created a dependency on a particular route definition, rather than relying strictly on the action and controller that will be called.

The route is what determines the shape of the URL. If you create a link by using a route name, you are literally saying "I want the URL to be this shape," rather than saying "I want the URL to invoke this functionality" and letting the route engine decide which route is most appropriate.

Whether or not this makes sense in your particular application ultimately depends on your needs. Using a named route creates a level of indirection which allows you to change both the shape of the URL, and the controller/method that gets called, by merely changing the entry in the route table.

As to the coupling aspect, using a named route does tightly-couple the links to that specific route because no other route would be eligible, and because it requires the view to have knowledge of that specific route. Whether that is a problem or not, again, depends on your needs. If that coupling is a desirable feature, it doesn't really matter whether it is "tight" or not according to someone else's opinion.

I could be waaaaaay off, but...

It seems that a model is necessary to use the library, you just don't care which one, right? Further, the library doesn't care how the model gets its data, or where the data comes from, it just wants an object that meets the criteria of the interface.

If the library doesn't need the connection object, then make that the responsibility of the model.

interface MyFooModelInterface 
{
    public function getItem($id);
    public function addItem($item);
    ...
}

And here is a simple class example:

class MyLib
{
    protected $model;

    public function __construct(MyFooModelInterface $model)
    {
        $this->model = $model;
    }

    public function getItemById($id)
    {
        return $this->model->getItem($id);
    }
}

And a Model example:

class MyModel implements MyFooModelInterface
{
    protected $conn;

    public function __construct(\PDO $conn)
    {
         $this->conn = $conn;
    }

    public function getItem($id)
    {
        ...
    }
    public function addItem($item)
    {
        ...
    }
}

And here is an example of using it:

// the model is responsible for handling the data retrieval, not the library
$model = new Model($conn);

$lib = new MyLib($model);
return $lib->getItem($id);

Now, if I decide that I want a file-based solution, I can easily swap the model. To test the model, the connection obj is injected, so that's easy. To test the library, the model is also injected, so, again, that's easy.

In this case, the model is only responsible for getting the data, while the library is responsible for handling and processing the data. Neither cares what or how the other one does it.

EDIT

the library with a factory method to create the model if none was given...

class MyLib
{
    protected $model;

    public function __construct(MyFooModelInterface $model = null, $conn = null)
    {
        // if we have a model, use it
        if (!is_null($model)) { 
            $this->model = $model; 
        }

        // if we don't have a model, but have a connection obj, make a model
        if (is_null($model) && !is_null($conn)) { 
            $this->model = $this->factoryModel($conn); 
        }

        // if we don't have either, quit and go home
        if (is_null($model) && is_null($conn)) {
            throw new Exception('need moar inputs');
        }
    }

    protected factoryModel($conn)
    {
        $this->model = new MyModel($conn);
    }

    public function getItemById($id)
    {
        return $this->model->getItem($id);
    }
}