Enzymes are likely to belong to one of these common structures. We have collected l for all of the 1000 sequences which tells us how the degree of neutrality of aptamers/ribozymes relates to that of the common structure. Real ribozymes have rather low degree of neutrality because these molecules have been produced by artificial directional selection. Such a decrease in robustness was shown. However, only 48.2% of the considered 305 real sequences have lower l than the median for the random sequences. And 9.1% of the real sequences fall into the upmost decile, i.e. they have a higher l than 90% of random sequences; and 2.2% of the real sequences have higher l than 95% of the random sequences. All in all the distribution of neutralities is not different from the distribution obtained for the random sequences. This is remarkable considering the fact that these ribozymes had been subject to intense directional selection for the required functionality. Although robustness and evolvability are not necessarily in conflict, it is legitimate to ask whether stabilizing selection could increase the robustness of these populations further, as demonstrated in the theory of Remdesivir neutral networks. We have thus exerted stabilizing selection on different molecules that already had a rather high l with population size 500 through 5000 generations. We show the highest degree of neutrality for structure-preserving variants. It is apparent that stabilizing selection can guide robustness to the top 25% or even 5% of the distribution obtained for random sequences. Thus, we can expect that ribozymes in primordial riboorganisms were even more error-resistant than ribozymes evolved in vitro, as they were subject to many generations of stabilizing selection. We have found that the number of 1-step neutral mutants, for short sequences, is an excellent predictor of the error threshold. Other characteristics of structure are not as highly correlated with the error threshold. Maintenance of RNA secondary structure is a good predictor of maintenance of enzymatic activity, but especially around the active site the actual nucleotides presents are also important. In this investigation we have not considered critical sites in our fitness landscape, which would lower the degree of neutrality of sequences. Considering critical sites would most probably not affect the correlation of error threshold with the degree of neutrality. The possibility of estimating the error threshold by available and easily computable characteristic of RNA sequences allows us to assess the replicability of aptamers and ribozymes. We have shown that functional phenotypes are mutationally robust above chance level and that, in effect, most known ribozymes could be replicated.