The scientific revolution can be seen as an innovation within the linguistic system currently adopted in ordinary languages. Extraction of quantitative features from observations via suitable measuring devices means that the words of science are numbers, and the connecting syntax is a set of mathematical rules. Once a general law is available (as, e.g., Newton gravitation) all consequences can be worked out in a purely deductive way. This characteristics of modern physics displays two orders of drawbacks, namely, Gödel undecidability of deductive procedures, and intractability of computer modelings of complex situations.The way out of such a crisis consists in an adaptive strategy, that is, in a frequent readjustment of the rules suggested by the observed events. This way, the language has no longer fixed rules, hence it appears as semantically open. Semantic openness implies a re-evaluation of the notion of "truth" as recognition of the essential role of some external features in orienting the cognitive procedures, against "certitude", i.e., self consistency within a deductive procedure.