I don’t think the study follows that statement.
Most of the issues are not actually the crypto itself, it’s the implementation.
If there is a “simpler to implement” crypto algorithm which is theoretically less safe than others. (sha2 vs sha3 for a trite example) then it seems that it might be more secure to go with the lesser secure but more simple algorithm put pay more attention to the implementation.
Similar to raft vs paxos.
Given how much work it is to process Yet Another File Format or Just Another Over The Air Protocol or What The Hell Is This Serialization Format……
I think Everybody should roll their own crypto.
Sure a nation state can break my crypto….. and find out about my deep dark love of trees… but it will take them a bit of time, not much, true, but actual real time.
Which they wont have if they have another 999999 crypto formats to go to get to the stuff that matters.