First, it did way less than 50 dives. I'm finding it really hard to find any solid numbers, but I've only found reference to only 3 or 4 previous dives to the Titanic, a test dive and a few dives to much shallower depths. That would put the failure rate for dives to this depth closer to 20%.
Second, the space shuttle never carried any paying passengers.
I'm very much of the opinion that the second you start accepting paying passengers, the safety standards should massively go up. Consumers have an expectation that if a company is offering a service to the general public, that service will meet a minimum level of safety. Companies shouldn't be able to hide behind liability wavers (well, not for a ~2% risk and absolutely not for a 20% risk) as even with a waver explictly stating death is an option, they will under estimate it.
Third, NASA engineers actually quantified the risk all the way back in the design stage and had it at ~2%, remarkably accurate in retrospect. NASA decided that was an acceptable level of risk for their mission profile and went ahead (it would have been way safer if the air force hadn't put insane mission profiles on it)
With everything we now know, there is no way in hell that OceanGate had any level of risk calculation. They were denying any risk at all, asserting that their real-time monitoring procedures would catch any failures before they became critical.