Sybil attacks are like DoS attacks. They are never going away fully. All webservers are vulnerable to DoS attacks and yet in practise it is pretty rare for major websites to go down because of them despite their simplicity.
But there are strategies to mitigate sybil attacks. Requiring hashcash style proof of work to participate in the network, increase state requirements to participate in the network, basing node id's on something the attacker cannot control (e.g. IP addressss), are common strategies often mentioned.
In the context of bit torrent, i guess it doesn't super matter that much. If your goal is to spy on the overlay network, while the entire point is to publicize the tracker data so what is the point of spying on public data? If the goal is to more actively attack the network - having bit torrent be a dual system with traditional trackers (and other ways to find peers) also, means that an attack causes degredation but is not a killing blow to bit torrent. Which in the cost/benefit analysis reduces the benefits, causing attackers to wonder if the cost is worth it.
Like denial of service attacks, the best defence is to make the attack expensive enough that it just isn't worth it.
Sybil attacks are not the only category of vulnerabilities found.
As your argument indirectly points out, the problems are quite fundamental in nature and therefore won't be resolved by a random developer but maybe by a bunch of mathematicians. ( I know that the Tribler people worked on these a lot )
What you call strategic decisions are in fact choices made on a operational level. In case of your context example, this is something the adversary is very well aware of.
Some decades ago, the did not only poison various DHT-like systems but even went as far as uploading jailbait torrents to the Piratebay, ordered hits on people via the US state department and got people extradited from non treaty countries in exchange for FTA concessions.
Striking a killing blow is not in their strategic interest but might become a operational choice at some point in the future.
In this scenario, the cost/benefit analysis is a no-brainer and that leaves us exposed and controlled.
