Even if less than one in 10000 new hypotheses is worth anything, missing out on one that is correct is a disaster. That's why it is important to create a very efficient method for testing all hypotheses. That's where such an Internet platform comes in. Since many hypotheses can make somewhat different and even unique predictions on the outcomes of the same experiment, one experiment can test many hypotheses. That's where free linking comes in. If an user who knows an experiment finds a hypothesis that makes falsifiable predictions for which the experiment is relevant, he/she simply adds a link from the hypothesis page to the experiment page! Don't ask the hypothesis for credentials, don't care if you support or kill a hypothesis, just link to whatever experiments you know! Of course, such a website must demand the falsifiable predictions to be described concretely and understandable to whoever have the practical knowledge to do the experiment. A failed engineering project can be theoretical scientific progress by falsification!