I'm fairly big on using non-structural damage indicators to assess a tornado's strength. It's still a long way from exact but it tends to be a bit more predictable than building damage. Ground and crop scouring, and heavy or sturdy and low-built objects being tossed long distances are some of the best indicators in my humble opinion, even if there are still a number of variables. Let's say a tornado sweeps away a house. There are a lot of reasons why, at least in theory, it could be done by an EF2-EF3 tornado. If that same tornado tosses a freight car a quarter mile, or scours 12 inches of soil...I don't think many people will dispute that it's a violent one. It's the reason why I would personally have upgraded the Vilonia, AR, Chapman, KS, and Washington, OK tornadoes to EF5, but left the Rochelle, IL, Tuscaloosa, AL, and Chickasha, OK tornadoes at high-end EF4.
I also definitely agree that the 1950s to the 1970s were definitely much too liberal in terms of tornado ratings. For instance, the NCDC lists 12 F5 tornadoes in the 1950s, but I've only found evidence of five, maybe six at a stretch, of them causing EF5 damage. 11 F5 tornadoes are listed for the 60s, but again, only about six seem to have caused anything that could be considered EF5 damage. They list a whopping 14 F5 tornadoes in the 1970s, but even then I only see 7, maybe 8 at a fairly big stretch, which caused clear EF5-level damage. By the time of the early 1980s, the standards for F5 ratings seem to have become much more reasonable, although it isn't really until the mid to late 1980s that standards for F4 ratings become more in line with what should be expected by EF scale-standards.