Note that the person asking for the review is not the author of the paper, but the editor of the journal or conference. They have several hundred submissions to keep track of at a time. They need to assign two or three reviewers to each paper, and then find alternative reviewers when those decline. They need to collate all the reviews for a paper, use them to decide whether to accept/reject/accept-with-modifications the paper, forward the reviews to the author (except the parts that the reviewer has marked confidential), having stripped the reviewer's identity. If the authors upload revisions, all the reviewers (but not anyone who declined to review in the first round) should receive the revisions, and so on.