But there is; gold standard for human rated missions is 98% reliability rate. This means testing every single component, and making sure that Product(Px) > 0.98 where Px is the probability failure of component x. This drives the cost up exponentially. E.g. if the rocket has Px = 0.99 and the crew capsule has Pcapsule = 0.99, then Psystem = 0.98, barely clearing the threshold.
Falcon 9 had 3.5 failures out of 185 launches, giving it a Pfalcon9 = 0.98; PfalconHeavy then is,at most, 0.98^3 = 0.84, way too low to make it human-rateabale.
The difference between Procket 0.84 and P0.99 is exponential cost.
Falcon Heavy has never had a launch failure. It's 4/4
If you're arguing it's had a failure because it shares heritage then SLS was 133/135 before the launch which is worse and so shouldn't fly crew until it does way more than just one flight.
Second, I am talking about design goals, not track records. You don't evaluate the safety of a large system by running it a lot of times and see what happens - you design the thing towards a goal of safety. So this involves designing each component to the specific failure rate, and testing individually, then then testing together, documenting every step, to provide traceability of every change on every component.
The comparison to historical data on Falcon is not entirely valid, because we don't know the design goals of Falcon, etc. Just because Falcon 9 is human rated, doesn't mean Falcon Heavy is human rated, etc.
The quiet secret is that CCP can bypass a ton of requirements that NASA can’t. That is, until they do something like lose an astronaut and then I suspect they will get much more oversight.