Abstract: China aims to achieve full coverage of social work stations in towns (streets) by the end of the "14th Five-Year Plan", incorporating these stations into key work evaluations and grassroots performance assessments. Based on participatory observations of the government-purchased social work station evaluation project in D district, Z city, H province, and an analytical perspective of the evaluation blind spot hypothesis, the study finds that current evaluations of social work station projects still suffer from blind spot, including the neglect of the one-sidedness of textual materials, the underestimation of the ambiguity of expert evaluations, and the disregard for accountability of the purchaser. These blind spots to some extent damage the comprehensiveness, accuracy, and fairness of the evaluations. The objective causes of evaluation blind spots are due to the temporariness of evaluation time and the fragmentation of evaluation space, while the subjective causes are due to the imbalanced cooperative relationship where the government is strong and social organizations are weak. Using blockchain concepts and technology to construct a dynamic evaluation platform for social work station projects and a cloud evaluation model, and then using technology to empower the reconstruction of the evaluation timeline and to adjust the imbalanced government-social relationship, is an effective method to reduce evaluation blind spots.