It isn't that easy, unfortunately. If it was that easy Satoshi didn't need to develop Proof of Work as the double spend problem would have been already solved (a chain of blocks would then be enough, they would fulfill the same function than your cosigning algorithm).

Nodes connect and disconnect all the time. If the attacker has spawned enough nodes, then it's more likely that a node connecting while the sybil attack is going on, will connect to one of the attackers nodes, who will "infect" it with the double spend transaction. The nodes who had connected to the attacker would thus also be blocked by the honest nodes, even if they had no bad intention.

The idea that a few honest nodes can revert a sybil attack only works if all nodes (or at least an extreme large majority) are online 24/7. You can't rely on this. It's possible that the honest nodes can repel a sybil attack once, or twice. But a well prepared attacker would eventually be able to steal funds. A cryptocurrency with such low safety wouldn't be able to attract value.

Your concept thus needs an additional mechanism to prevent the sybil attack. I don't know how different it is exactly from Obyte, but you could look there for inspirations.

Such topics were discussed a lot around 2014-16 when the first "working" PoS algorithms were developed (the original Peercoin algorithm was flawed just because the stakers could grind through different configurations, so basically it was "obfuscated proof of work"). IOTA is an example of a coin that tried to rely on a "provably random" transaction selection function, but they weren't able to shut down a centralized coordinator node until now. And their plans to shut it down depend on a PoS system not unlike Ethereum's. Currently if I'm not wrong they're in a transition phase with selected validators.