There is no need to connect to a real world identity, but there is good reason to have a concept of 'social standing' and trustworthiness on something like Steem.
Firstly Steem involves financial transactions. While we remove the element of trusting a third party, we can't always remove the need to trust a counterparty. Since we need trust in such transactions, we need a basis for that trust. Reputation (being known for your traits) allows us a way to judge your trustworthiness in order to facilitate those transactions. The potential loss of social standing also means that there is a cost associated with cheating. It is far from perfect, but it is better than a blind gamble on the trustworthiness of your counterparty.
Aside from that, there's also the fact that trust and authority are sometimes a legitimate part of debate and discussion. A professor of astrophysics can be a legitimate authority on the subject of astrophysics. In a discussion on the topic, you could verify every single claim that they make, but it would be quite time consuming to do so. It is reasonable to lean on someone's legitimate authority on a topic, as long as a) they really are an expert and b) you have a good basis for trust, because even a legitimate authority can lie, and there must be some cost/risk associated with doing so.
Welcome to Steem!
Your example and point is valid, my point was more relevant to the field of politics and economics, where the public believe that politicians know what they are doing based on the premise that to get into that role they should know what they are doing. I think in regards to the sciences a researchers reputation of course matters but what matters more is peer review, so should an anonymous person make a discovery, peer review will quickly validate it, should it be right.