Codebook appears, if I don't err, to be an antiquitated topic rarely touched upon in discussions of modern cryptography. Couldn't codebook nonetheless be of high practical utility and even be extremely advantageous in certain critical fields of modern secret communications? Consider e.g. the hypothetical situation where a manager has to securely correspond with his representative who is negotiating with the customer in a bid under rival competitions. It seems that the essential instructions and responses could be realistically formulated as sentences and phrases of an appropriately designed codebook having entries of, say, 256 in number. That is, one needs only to suitably employ a highly limited number of code words in entirely neutral appearing text messages or to otherwise transmit somehow a small number of bits denoting the indices of the involved sentences or phrases in the codebook in certain appropriate way (e.g. the scheme http://s13.zetaboards.com/Crypto/topic/6939954/1/). It may be remarked that the codebook need not be static but can preferrably be dynamic, i.e. the mapping between the codewords and their meanings is dependent on a session key, being a random permutation (Fisher and Yates algorithm or alternatives) determined by it. The codewords are preferrably to be chosen to be certain commonly employed words, e.g. surnames etc. in order to ease the composition of a fairly natural covertext. In case it is desired to have integrity check, one could employ two disjoint sets of codewords that are independently permuted w.r.p.t. the set of meanings so that each meaning (or only those meanings that are classified as important) will be trasmitted via the coexistence of at least one keyword from each set.
Thus seen, top secret communication isn't necessarily a hard problem, or is it? Since no encrypted stuffs are involved, the scheme avoids from the outset issues like enforcing delivery of encryption keys or implanting of "official" backdoors (cf. news of a current debate in US on Apple's phones.)