Part of its process (Im merging and simplifying here) involves having many peoples computers store a copy of the information. When something needs to happen theses computers do a series of computations. The process only completes when these computers reach consensus many computers agree that the final result, the output of the work, is valid. If someone is acting up, they are identified by their inability to concede.
I think it can be even simpler.
When the information is updated, the updating computer must spend money before the others will consider accepting the update. If the update is valid, then the update is accepted and the updating computer is paid for its work. If it is not valid, then the update is rejected and the updating computer is not paid. This process ensures that computers are motivated to submit only valid updates. When there are competing updates, then the update that costs the most is accepted. This ensures that undoing an update is very costly.
BTW, I don't think that consensus needs to be explained since it is implied, and more importantly, because it is actually more a result of the process than a part of it.