That’s not necessarily true.
For starters, vulnerability to these “attack vectors” is solely related to the consensus mechanism used. It’s funny you should mention 33% given that this (BFT-like consensus) is more commonly used in limited-party mostly-trusted environments (e.g. Consortiums in regulated spaces). This makes the probability of a 33% attack unlikely. Would you consider a database insecure because 5 admins in the company each have an access key? That’s a “20% attack vector”. That’s not even getting into public blockchains, where a 51% attack (Traditional proof of work, proof of stake) doesn’t necessarily follow, again because the consensus can vary, each with its own advantages and disadvantages.
Linear scalability is an open problem. Yet, blockchain technology has only been around for fewer than 10 years. Over the last few decades, we’ve seen cryptographic “scalability” improvements, such as storage of multisignatures, that would seem to imply there is a way to overcome these early problems.
You can’t look at a nascent technology like this with fatalistic eyes, there are many counterexamples of previous technologies that have gone through similar growing pains.
