There are tradeoffs. Most of the time I find terminology has drifted so substantially over time that reading old papers isn't worth the effort. I instead usually look for literature review type of papers that often have good summaries of relevant results in modern terminology.

I have also found old papers to be the best for learning a new concept - they make it as simple as it can be. A colleague of mine wrote a paper that proves that negative numbers exist and would often cite archimedes, Galileo, newton, etc in his bibliographies. I think he loved crediting the first person to come up with whatever idea or concept he was using.