Performativity of science refers to the phenomenon that the dissemination of scientific conceptualisations can sometimes affect their target systems or referents. A widely held view in the literature is that scientists ought not to deliberately deploy performative models or theories with the aim of eliciting desirable changes in their target systems. This paper has three aims. First, I cast and defend this received view as a worry about autonomy-infringing paternalism and, to that end, develop a taxonomy of the harms it can impose. Second, I consider various approaches to this worry within the extant literature and argue that these offer only unsatisfactory responses. Third, I propose two positive claims. Manipulation of target systems is (a) not inherently paternalist and can be unproblematic, and is (b) sometimes paternalist, but whenever such paternalism is inescapable, it has got to be justifiable. I generalise an example of modelling international climate change coordination to develop this point.

科学表演性是指科学概念的传播有时会影响其目标系统或参照物的现象。文献中一个普遍持有的观点是，科学家不应该故意部署表演模型或理论，目的是在他们的目标系统中引发理想的变化。本文有三个目标。首先，我将这种公认的观点视为对侵犯自主权的家长式作风的担忧进行辩护，并为此目的，对它可能造成的伤害进行了分类。其次，我在现有文献中考虑了解决这种担忧的各种方法，并认为这些方法只提供了不令人满意的回答。第三，我提出两个积极的主张。对目标系统的纵 （a） 本质上不是家长式的，也可能没有问题，并且 （b） 有时是家长式的，但只要这种家长式是不可避免的，它就必须是合理的。为了发展这一点，我概括了一个模拟国际气候变化协调的例子。