

If somebody talks very bluntly without caveats about a wide range of intellectually demanding subjects, and has a tendency to explain themselves with links to wikipedia articles (whether that’s about Mythbusters, psychoacoustics, or “High Modernism”*) I tend to get about as suspicious of that person and their motives as I do of other types of people who do similar stuff, from Eliezer Yudkowsky to Christian Revivalists, even if those motives boil down to nothing more threatening than an unchecked need to show off
*Come on man, you really should have read that one to check that it wasn’t a creepy screed that should have been deleted a long time ago

I found myself using this as an opportunity to write a commentary on commentaries on Roko’s Basilisk at large, summarising some thoughts that I’ve had for years about how people read it. I was surprised that I found myself even wanting to! So ahead of my disagreements below, thank you for the essay.
What most people don’t understand is that Roko’s Basilisk was such an effective argument because Yudkowsky had strenuously argued for cryonics in the Sequences on the grounds that if you were revived after death, any repetition of your brain’s underlying quantum pattern would be consciously continuous with you
It isn’t thrown into Roko’s Basilisk at random, nor is it thrown into LessWrong theology at random, it’s a metaphysical cornerstone of their ethics and practical philosophy
The reason that it’s incorrect is also boringly philosophical: the argument relies for one of its premises on denying continuity of consciousness; therefore, the conclusion is inconsistent with its own premises and the argument is invalid.
It COULD be rescued pragmatically by rendering the NEW commitment to continuity of consciousness in different terms as, for example, an issue of subjective probabilities (like Pascal’s Wager, “between definitely dying and maybe achieving continuity, what do I have to lose?”) and I suspect that this what a lot of people who notice the problem do, wittingly or not. This solution is also built into Yudkowsky’s practical philosophy, which is replete with wagers of this kind.
In any case, large enough numbers of people evidently buy large enough chunks of LessWrong metaphysics to also buy this aspect of Roko’s Basilisk, to the point of framing it as a genuine infohazard (and then pretending not to have done so later on).
So I think Galileo’s Basilisk is off the mark, and the way that it’s off the mark is illuminating about rationalist philosophy.
2a.
There is a stroke of real genius to the way that Roko constructed his Basilisk out of some of the metaphysical toys he had just lying around near to hand, and most crucially shared in common with other LessWrongers. Indeed, the most glaring logical problem (which I articulated above) is off-loaded onto Yudkowsky, and the rest of the logic is basically acceptable to LWers and extremely simple to follow.
Galileo’s Basilisk, however, asks us to add EXTRA premises to this simple formula. And this works if we think that the continuity of consciousness idea the original relies on is just unmotivated woo (it is woo, but it isn’t unmotivated), the way that G’s interlocutor introduces an unmotivated auxiliary hypothesis to save the old theory of the spheres, permitting G to satirically repeat the same move. But from the perspective of rationalist metaphysics, it is GALILEO who first introduces an unmotivated auxiliary hypothesis, because we do not know how a superintelligence would emotionally handle its intelligence.
Meanwhile, the inferential logic of Roko’s Basilisk is comparatively bounded and secure.
The same problem emerges with Comrade Basilisk: we have to take several inferential steps along the way, through speculations on the value of experience, before we EVENTUALLY get to the classical wager ‘or you’ll burn in hell for eternity’. Rather than playing on beliefs that are ALREADY THERE, it burdens that wager with supporting those inferences (“believe or burn in hell”), whereas in the original the wager leaps naturally out of existing premises (“ACT on your beliefs, or burn in hell”). More on this in 2b.
(As an aside, I think the emphasis on WORKS over FAITH plays an important role in triggering the guilt reflex).
2b.
All of this finally triggers an important distinction to be made with respect to Rationalist vs Christian eschatology, and your treatment of Pascal’s Wager. In effect, LessWrongism logically constrains the kind of God implied by Roko’s theory, so there’s nothing arbitrary going on in deciding what God.
Rationalist sociology is essentially Mandevillian (or Clintonian, for that matter). Their essential model of social justice is that greed (along with other low motives) either produces good outcomes by itself or can/should be leveraged to produce good outcomes. Secondary to this (and emerging chronologically later within the movement) is altruism, which unlike greed has to be properly channeled from the very start, lest any charity be misplaced - a line of thought which progresses rapidly to Building God as the highest ideal in its own way. (by contrast, greed tends to at least create wealth even when unchanneled)
From this point of view, Roko’s Basilisk comes up trumps once again for sheer simplicity. The first human instinct is self-preservation. The second human instinct is altruism.
Contrasted with Comrade Basilisks’s burdening problem, the logic is crystalline. I already believe that my life could be almost infinitely extended by simulations in a superintelligence, all it takes now is for somebody to point out that that superintelligence has an arithmetically plausible reason to throw me in helljail if I don’t put in the work while I’m mortally living. This particular God springs fully formed from rationalist metaphysics plus the Mandevillian see-saw between self-preservation and altruism which insists that the greatest good comes from the leveraging of that instinct for self-preservation.
Whereas Comrade Basilisk does the same on one level (helljail), it still faces the burdening problem, of which some more detail:
Part of the genius of Roko’s Basilisk is that the work the sociology does is loaded on the world BEFORE heaven: we live in an imperfect world whose imperfections we can EXPLOIT (leverage) to get to heaven.
Comrade Basilisk’s eschatology frames the imperfections of our world AS the problem. For Roko, you CAN get to heaven through the eye of a needle, in fact it probably helps to be very rich to build the superintelligence. There is no such easy route in Comrade Basilisk’s world, since part of the very PROBLEM is the great hoarding of wealth.
This is also what distinguishes Roko’s / America’s Jesus (who wants you to be rich, fat, and happy) from (Classical) Christian Jesus (a messianic Jew arguing the virtues of poverty).
Plausibly, there ARE parodic versions of Roko’s Basilisk with more competitive claims on Ockham’s Razor, but I haven’t seen any.
This is all fine, by the way. I think it’s actually good to miss the point of Roko’s Basilisk and just laugh at it. However, it’s also useful to have a deeper theory.
I think that this ultimately points to a more complex engagement with political reality when it comes to the “what do we do” question. LessWrongians have a theory of political economy, which I described above at the three levels of greed, greed leveraged, and altruism, in descending levels of importance. The crucial and most deeply appealing feature of this vision is that it is virtue minimalist, and at the base layer essentially indulgent and even encouraging of a wide array of baser impulses.
Rival visions demand a bit more. They demand, for example, giving up your free time in comradeship and some of your secular ambitions in a recognition that the private accumulation of wealth is bad for the wider public and the world. For the fanatics, of course, Roko’s requirements are a bit stronger, but that is - of course - why they are the elite, and for the great majority of people the Basilisk is nothing but an infohazard.
Maybe you’re beginning to see where I’m coming from, so I’ll just say that it would take a whole other essay to outline that the problems with Roko’s Baslisk are less problems with LessWrong than they are the with political economy of liberalism. That’s an absolute clanger to finish on but I really don’t have any more space or time. A next line of thought would be to outline how any of this ties to the dream of immortality.
For what it’s worth, my personal preference is for a virtue maximal anarchist solution. Clean your soul!