Abstract: With the world’s largest number of researchers and second largest investment in scientific research, China has achieved remarkable results in both quantity and quality of innovations. However, the US-China trade war and the ZTE incident have shown that China's scientific prowess is not as dazzling as what statistics show, which has triggered criticism and controversy from all walks of life on China's "low level of science and technology" and "uselessness of scientific and technological achievements". Why is there a considerable discrepancy between statistics and reality? In order to answer this question, first, we extract the paradox of economy of scale in scientific research, an significant issue with Chinese characteristics, by comparing the data among OECD countries. Then, we propose a theoretical framework based on the innovation value chain and the typical characteristics of scientific research. And under this framework, we present a multi-level research on S&T talents and innovation performance at micro, meso plus macro levels, and comprehensively work through the multi-dimensional research results on the relationship between S&T talents and innovation performance, economic performance, social performance plus international performance, putting forward the following research propositions: (1) There may be an inverted U-shaped relationship between the scale of S&T talents and innovation quantity. (2) There may be a positive relationship between the scale of S&T talents and innovation quality. (3) There may be an inverted U-shaped relationship between the scale of S&T talents and economic performance. (4) There may be a positive relationship between the scale of S&T talents and social performance. (5) There may be an inverted U-shaped relationship between the scale of S&T talents and international performance. On this basis, this article points out the available research has four shortcomings: insufficient exploration of talent heterogeneity and structure, insufficient attention to tacit performance, incomplete consideration of outcome and subject differences, unclear mechanism and boundary conditions. For this reason, we suggest that future research should be carried out from the following aspects: (1) Fully consider the diversity and multi-tier of input-output indicators and strengthen micro-level research. Based on the subdivision of S&T talents in different regions, subjects, types, structures and levels, we should study the heterogeneous relationship between the accumulation of S&T talents and different types of performance, comprehensively analyze the differentiated results among different indicators, and explore the mechanism and boundary conditions of the complex relationship to provide solid theoretical evidence for more precise formulation of S&T talents support policies and the realization of the optimal allocation of scientific research resources. (2) Enrich the tacit performance indicators and explore the synergistic effect plus transformation mechanism among different performances. (3) Conduct crossover study on multiple types, levels, and systems to clarify the functions and positions of different subjects. (4) Discern the acting mechanism and boundary conditions among S&T innovation inputs and outputs in the condition of hierarchical classification. Finally, we consider that it is not convincing to criticize China's "low level of science and technology" and "uselessness of scientific and technological achievements" merely based on economic performance indicators without considering the type of outcomes and performance structure.