Abstract: Seismicity-based earthquake forecasting/prediction techniques progressed rapidly in the last decade. Some were even claimed to be promising in predicting the occurrences of future large earthquakes. Yet, apparently ambiguous algorithms and lack of theory about seismicity precursor signals cast doubts on these techniques. Two fundamental issues stand out: is there any relationship between these apparently ambiguous algorithms, and is there any conceptual model that can unify these independently proposed algorithms? In this paper we suggest a unified framework based on the theory of critical transitions to link two seismicity-based predicting techniques, i.e., the loading-unloading response ratio (LURR) and pattern informatics (PI) algorithms. In terms of critical transitions, the loss of lithospheric resilience suggested by the LURR technique is theoretically associated with high seismicity variances portrayed in a PI hotspot map. This study thus implies the existence of generic seismicity precursors that could be also revealed by the recently proposed approaches of detecting early warning signals in the theory of critical transitions.