This is a good article but the author severely underestimates how difficult and expensive it is to acquire 1/3 of the entire stake of a decentralized network. In many cases it is essentially impossible. The cost grows exponentially since the attacker is forced to buy tokens in order to perform this attack. Since the tokens are limited in supply, the price will have started growing exponentially by the time the attacker manages to acquire even 10% of all tokens; at that point, they would be forced to give up when they realize the insanity of their plan and how much they stand to lose. Also, many PoS variants require forgers to lock or otherwise hold onto the majority of their tokens in order to earn block rewards; so these tokens are essentially out of reach of the attacker... On the other hand, with PoW, the cost of an attacker acquiring 51% of the hashpower is linear; the price of an ASIC miners does not increase (in the long term) as demand increases because the market will produce more ASICs in response to any increase in demand; unlike blockchain tokens, ASIC miners are not a scarce resource. Not to mention that in the case of a PoW attack, unlike with PoS, the attacker has nothing at stake aside from their investment in ASIC miners (unless they rent most of their ASIC miners from big mining pools which requires even less investment).

Also, in the case of certain PoS variants like DPoS (Delegated PoS), there is no difference between active stake or inactive stake. This creates a bias towards incumbent delegates/forgers as it allows them to benefit from old votes of inactive accounts from x years ago to hold onto delegate positions for longer but IMO this a fair trade-off as it provides additional security.