Longtermism is a cancer.
Speculating is fine but you can’t use pure speculation to justify things in the real world. That’s the problem.
Truth is that we have no idea about anything much past our own slice of time. In 2000 years we may be extinct, reverted to hunter gatherers, settled across the solar system like in The Expanse, or extinct or kept in zoos by superintelligent posthuman AIs that have expanded out into the cosmos. Or none of the above. Who knows.
But we seem to have been unable to outrun wars, greed and outright evil. Those you can definitely count on for the longer term unless we start to actively work against them.
You'd think WWII would teach us that lesson forever but for some strange reason we are immune to absorbing this and we have made that same mistake at least another few 100 times since then. And one of these days one of those idiots is going to kill us all. And for some reason it is always men.
No we actually did not manage to prevent it. We almost caused it but sheer luck in several cases derailed Armageddon. It could have just as easily happened, and probably more than just as easily (figure about 60:40 with us living in the '40' branch of the multiverse). Pretty good odds though. But over the longer term: not such good odds.
Root hardening is op. If a capability increases the risk of societal decline (nukes e.g.) you can increase societies abilities to recover from that by creating a video archive of your civilization techniques aka YouTube. And by increasing the ability to approach scenarios again and again, the likelihood of stable traversal to a scenario with more capabilities increases enormously. The root aka now is hardened against a negative outcome that is likely.
Longtermism with good models and no biases is a plus.