I was mostly using "reputable" as a hand wavy way to eliminate the minority of smart contract projects that don't have such protections. As it relates to smart contracts or more broadly crypto, the point is that you can theoretically inspect every aspect of the system, rather than trusting whoever is running it won't do anything bad because they're "reputable". There are still failure modes for this (eg. hidden bugs, intentional or otherwise), but at least you can't pull off what happened at FTX.