There is always going to be signal loss from the raw data of the actual experiment -> the scientists quantizing and interpreting the data -> formatting and pres

There is always going to be signal loss from the raw data of the actual experiment -> the scientists quantizing and interpreting the data -> formatting and presenting salient parts of the data to get published -> popular science writers condensing the publication down to a few paragraphs -> general public interpreting pop science journalism -> word of mouth from your friends who read a pop science piece. It's the age old problem of going back to the primary source. What exactly is the primary source? The raw data? The original publication? You can even go so far and claim the original experiment wasn't done correctly and you don't trust the experimenter. It's an infinite regression that just isn't very practical. The key I think is more openness in scientific publishing and data. No more anonymous peer reviewers. Make negative data just as important as positive results...etc