According to this scheme, practically all cellular structures and functions came later, and only RNA replication was present at the beginning. This is clearly a replication-first hypothesis, and in fact it is the more updated version of the replication paradigm, but it does not follow directly from the experimental data. It is important therefore to check if it is really possible to say that the RNA world was a world of replicators.

The replication catastrophes
The idea that life started with RNA replicators implies that the primitive Earth became populated by RNAs that could either replicate themselves or could catalyze the replication of other RNAs. Up until now it has not yet been proved that RNAs can self-replicate, but it has been shown that some of them do have a week replicase activity, and can therefore replicate other RNAs, thus forming a system that is, collectively, capable of replication. This allows us to conclude that some primitive RNAs could behave, to a certain extent, as the replicases used by Spiegelman and Eigen in their experiments, and that similar experiments were performed by nature, on a far larger scale, some four billion years ago.
The environmental conditions of the primitive Earth were surely different from those of Spiegelman’s and Eigen’s test-tubes, but this can be regarded as a secondary complication, and in a first approximation it can be ignored. What we cannot ignore, instead, is the fact that any replication process is inevitably affected by errors, and it is therefore imperative to understand the consequences that such errors have for the very survival of a replicating system. This is a crucial problem for all replication-first theories, because it has been proved that any self-replicating system can tolerate replication errors only below a critical threshold. Above such a threshold, the system is overwhelmed by a runaway error catastrophe, and is inexorably condemned to collapse. This is a fundamental problem, and in order to address it we need first to quantify the critical threshold.