ratiocinator
Lightly seared on the reality grill.
Although short strings of RNA can replicate themselves without enzyme assistants, longer strings need a retinue of helpers, and specifying them requires a very long sequence—longer than could be replicated with high-enough fidelity until those very enzymes were already present. We seem to face paradox once again, in a vicious circle succinctly described by John Maynard Smith: "One cannot have accurate replication without a length of RNA of, say, 2000 base pairs, and one cannot have that much RNA without accurate replication" (Maynard Smith 1979, p 445.)
Dennett is agreeing with John Maynard Smith concerning the paradox.
Note my emphasis. Dennett is not agreeing that there is a (real) paradox. Once you have something that replicates with inheritance and variation, then you have natural selection.
The next paragraph from Dennett's book:
One of the leading researchers on this period of evolutionary history is Manfred Eigen. In his elegant little book, Steps Towards Life (1992)—a good place to continue your exploration of these ideas—he shows how the macros gradually built up what he calls the "molecular tool-kit" that living cells use to re-create themselves, while also building around themselves the sorts of structures that became, in due course, the protective membranes of the first prokaryotic cells. This long period of precellular evolution has left no fossil traces, but it has left plenty of clues of its history in the "texts" that have been transmitted to us through its descendants, including, of course, the viruses that swarm around us today. By studying the actual surviving texts, the specific sequences of A, C, G, and T in the DNA of higher organisms and the A, C, G, and U of their RNA counterparts, researchers can deduce a great deal about the actual identity of the earliest self-replicating texts, using refined versions of the same techniques the philologists used to reconstruct the words that Plato actually wrote. Some sequences in our own DNA are truly ancient, even traceable (via translation back into the earlier RNA language) to sequences that were composed in the earliest days of macro evolution!