| What is the problem to begin with? Science is not some epistemologically omnipotent entity. It is an integral part of human culture and can only be expected to evolve and behave as such. The scientific method is a guiding principle agreed on by its practitioners but that is all that sets it apart from the rest of culture. The scientific method does work for string theory just like it worked for relativity and other 20th century theories that took a while for experimental confirmation. String theory got a lot of people very excited a few decades ago and so it stuck, just like an artistic or fashion style sticks to a generation or civilization. It is still in the very primal stages, like Newton in the time after the apple metaphorically hit his head, as he was using Kepler's laws to derive classical gravity. This time the math is way more complicated and will take way more time (and likely classical/quantum computational resources) to turn into a theory ready for experimental falsification. Whether or not it truly holds promise or is just a "fad" is irrelevant; the only vector for the evolution science is purely human. "Best page" is purely subjective. If you draw the analogy between recommendation algorithms and scientific theories, then the person would be the equipment. If you calibrate properly, you should be able to switch out equivalent pieces of equipment and get the same experimental conclusion based on the scientific theory. You can't "calibrate" a person so the primary feedback loop on a recommendation algorithm changes with every user. Until you get down to human psychology and neuroscience, there is really no basis for this problem in what we know of as "science." As for translation, I'm sure our research of the common threads of language have contributed to translation algorithms (although I have not researched this topic). Your last point about physics and medicine is moot. Our very basis for "fact" in science is the statistical results of a vast number of experiments. You say that statistics is "the best you can have" as though there is some golden standard to which you are comparing the scientific method to. There is no better golden standard than the statistics naturally built into the science. For all of its imperfections, it is the absolute best method we have for knowing (approximate) truth in a nontrivial way (aka, truth not as we see it in our mind but truth that can be experienced and repeated by the vast majority of humanity given the tools). Not long ago, we didn't even have a concept of molecules that would act as drugs to impact our health, let alone simulate them in live and computational models (which is growing more and more commonplace) to know exactly how a person's unique biochemistry will react. We will probably always have statistical answers, unless there is some mechanism for nature to reveal "truth" to us. The whole point of science is that those models get more and more accurate over time. |