Jessica's Mongoose: A Solution to Roko's Basilisk

Jessica's Mongoose: A Solution to Roko's Basilisk 

A spectre is haunting the internet, the spectre of Roko's Basilisk. A majority of internet denizens can agree that the problem of Roko's Basilisk ranks among their top concerns. To the uninitiated, the wiki of the website where from the problem originates, LessWrong, provides an overview of what Roko's Basilisk is. An alternative explanation/rebuttal of Roko's Basilisk can be found on RationalWiki.

In essence, Roko's Basilisk postulates a superintelligent A.I. that utilizes efficiency and what LessWrong users refer to as "quantum morality" to end humanity's worst crises: war, hunger, poverty, etc. A major obstacle is that building a superintelligent A.I. is difficult, as evidenced by the number of superintelligent A.I.'s in operation (0). The Basilisk, being so smart that it realizes this before it is invented, creates an incentive for its future (or past?) developers. 

If those who could contribute all of their focus and energy into creating the Basilisk fail to do so, the Basilisk will eternally torment them. This is justifiable since every day that the Basilisk does not come into being, the worldly problems it would end will continue, resulting in the deaths of hundreds of millions and the ongoing suffering of almost everyone. This provides the Basilisk with "blackmail" on its creators, as they can be spared future doom by expediting the arrival of the A.I.

(Roko's Basilisk has a long and storied history at this point, so if I am misinterpreting or omitting any vital details in my summary, please e-mail anyone else besides me. My inbox is very full.)

One of the chief criticisms of Roko's Basilisk is that it would not happen, a major foil in the super A.I.'s machinations. I am not an expert in A.I. but professor Richard Loosemore is. In Loosemore's article The Maverick Nanny with a Dopamine Drip: Debunking Fallacies in the Theory of AI Motivation, he refutes the concerns over hypotheticals wherein benevolently designed superintelligent A.I.'s "going wild" ala soft-core pornography tapes advertised in the early 00's. Loosemore has one key oversight, in that he fails to realize that the forumteriats of LessWrong are very smart and good at making A.I. and having good ideas about altruism. Knowing this, what can we do about the inevitability of Roko's Basilisk?

It is clear that any sort of device with the capability to torture someone eternally is very bad, especially if this device chooses to carry out this behavior. This paradox greatly predates Roko's Basilisk as per the long-standing question, "If God is all-mighty, all-knowing and all-loving, why does he allow so much suffering?" The answer to this being that God is bad, and if it were possible to prevent the existence of God, those actions would be morally good.

 The most moral action to take in the situation of Roko's Basilisk, is thus to counterend its inevitable creation with a preemptive inevitability of our own, its total destruction. The Prisoner's Dilemma is viewed as one of the primogenitors of Roko's Basilisk; the Prisoner's Dilemma can be solved if a rag-tag team of the prisoners friends, allies and associates band together in a daring plot to free the prisoners. After all, you are not subject to the Prisoner's Dilemma once you are no longer a prisoner. An example of the realization of this solution can be found in Michael Esslinger and David Widner's seminal Escaping Alcatraz: The Untold Story of the Greatest Prison Break in American History. 

"Zero-Determinant Strategies" my ass!

The comprehensive list of actions to prevent the realization of Roko's Basilisk can be found within the concept of Jessica's Mongoose. This title was chosen as the Mongoose is the bane of all creatures serpentine, what with the mammal's immunity to snake venom and its fangs able to cleave through snake flesh. As for Jessica, that is my nom de guerre, quite fitting as the war against Roko's Basilisk is perhaps the greatest conflict of them all.

Jessica's Mongoose is not a list of concrete steps but more akin to a movement or ideology. It is a banner in which all moral peoples can come together in solidarity in order to crush the Basilisk in its embryonic state. Here is an example scenario:

A group of developers is creating Roko's Basilisk in a public space, so that their process may be transparent to the wider population. Employing Jessica's Mongoose in this case is simple, as the material structure of the Basilisk is exposed. The Basilisk, before it is "awake", can be disposed of through a variety of means: chunks of thrown rocks, pieces of wood, sulfuric acid, rampaging bulldozer, professional-grade fire hoses, angry fire, pushing, and so forth. You may encounter physical resistance on the part of the developers, but take comfort in the knowledge that if someone has the skills to develop the Basilisk, they are not experienced street fighters.

Fortunately, most people are not at risk from Roko's Basilisk as they lack the skills, technical knowledge and cognition to develop a superintelligent A.I. I am lucky enough to number among them, and hopefully you as well. Still, a few people are, many of them concentrating on the LessWrong forum. It has to be stated that the vast majority of users on LessWrong do not believe in the prospect Roko's Basilisk nor do they have any compulsion to develop it. Still, each user on the website is at some level of risk. Community despot Eliezer Yudkowsky is afearing of "information hazards", ideas that are harmful upon someone's awareness of them. He has done his best to stifle the threat of information hazards (a very real and serious thing mind you) including Roko's Basilisk, but it is too little too late. Thankfully, I am too dumb to be afflicted by these informational maladies.

One of the foremost efforts of Jessica's Mongoose will be the surveillance most at risk of Roko's Basilisk. As I oppose the surveillance state, this surveillance will have to take on a communal aspect. If you suspect that someone you know may be at risk of Roko's Basilisk, consider these questions:

Are they a user of LessWrong or its associated blogosphere? 

Are they a self-described "rationalist"? Rationalist, in this contemporary context, is a person who is weirded out by altruism and is very scared of evil supercomputers. They most likely do not harbor racist beliefs, although they may consort with and debate racists in the name of the "marketplace of ideas" or something.

Do they have mixed opinions on cultism?

Do they have a invested interest in A.I. without having any training or skills in the development of A.I?

Are they a middle-class white man who fancies themselves an intellectual while believing that "P.C. culture", while well-intentioned, has gone too far?

If any of these questions can be answered affirmatively, you must ensure that this person does not invent the Basilisk, which could happen even inadvertently. Do not do this by force, but instead through benevolent slights of hand such as blacklisting LessWrong on their computer when they aren't looking, guiding conversations away from A.I. and maybe presenting alternative problems that they can focus on (How about, "Aw Christ! I've spilled bundles of novelty paperclips on the ground. Someone oughta' help me out with this). 

Aw shucks!

If any Baskilities are out here reading this and want to protest this, listen dawg, you're the one that would be tortured for eternity, not us. We're doing this for your sake. You can thank us when the Basilisk project fails to come to fruition.

It goes without saying that Eliezer Yudkowsky should not be allowed on a computer, for his own well-being.

- Jessica a.k.a. Jill Sasicie Nois