Honoring Petrov Day on LessWrong, in 2019

Just after midnight last night, 125 LessWrong users received the following email.

Subject Line: Honoring Petrov Day: I am trusting you with the launch codes

Dear {{username}},

Every Petrov Day, we practice not destroying the world. One particular way to do this is to practice the virtue of not taking unilateralist action.

It’s difficult to know who can be trusted, but today I have selected a group of LessWrong users who I think I can rely on in this way. You’ve all been given the opportunity to show yourselves capable and trustworthy.

This Petrov Day, between midnight and midnight PST, if you, {{username}}, enter the launch codes below on LessWrong, the Frontpage will go down for 24 hours.

Personalised launch code: {{codes}}

I hope to see you on the other side of this, with our honor intact.

Yours, Ben Pace & the LessWrong 2.0 Team

P.S. Here is the on-site announcement.

Unilateralist Action

As Nick Bostrom has observed, society is making it cheaper and easier for small groups to end the world. We’re lucky it requires major initiatives to build a nuclear bomb, and that the world can’t be destroyed by putting sand in a microwave.

However, other dangerous technologies are becoming widely available, especially in the domain of artificial intelligence. Only 6 months after OpenAI created the state-of-the-art language-modelling GPT-2, others created similarly powerful versions and released them to the public. They disagreed about the dangers, and, because there was nothing stopping them, moved ahead.

I don’t think this example is at all catastrophic, but I worry what this suggests about the future, when people will still have honest disagreements about the consequences of an action but where those consequences will be much worse.

And honest disagreements will happen. In the 1940s, the great physicist Niels Bohr met President Roosevelt and Prime Minister Churchill, to persuade them to give the instructions for building the atomic bomb to Russia. He wanted to bring in a new world order and establish global peace, and thought this would be necessary—he believed strongly that it would prevent arms race dynamics, if only everyone just shared their science. (Churchill did not allow it.) Our newest technologies technologies do not yet have the bomb’s ability to transform the world in minutes, but I think it’s likely we’ll make powerful discoveries in the coming decades, and that publishing those discoveries will not require the permission of a president.

And then it will only take one person to end the world. Even in a group of well-intentioned people, natural disagreements will mean someone will think that taking a damaging action is actually the correct choice — Nick Bostrom calls this the “unilateralist’s curse”. In a world where dangerous technology is widely available, the greatest risk is unilateralist action.

Not Destroying the World

Stanislav Petrov once chose not to destroy the world.

As a Lieutenant Colonel of the Soviet Army, Petrov manned the system built to detect whether the US government had fired nuclear weapons on Russia. On September 26th, 1983, the system reported multiple such attacks. Petrov’s job was to report this as an attack to his superiors, who would launch a retaliative nuclear response. But instead, contrary to all the evidence the systems were giving him, he called it in as a false alarm. This later turned out to be correct.

(For a more detailed story of how Stanislav Petrov saved the world, see the original LessWrong post by Eliezer, which started the tradition of Petrov Day.)

During the Cold War, many other people had the ability to end the world—presidents, generals, commanders of nuclear subs from many countries, and so on. Fortunately, none of them did. As the number of people with the ability to end the world increases, so too does the standard to which we must hold ourselves. We lived up to our responsibilities in the cold war, but barely. (The Global Catastrophic Risks Institute has compiled an excellent list of 60 close calls.)

Petrov Day

On Petrov Day, we try to live to up to this responsibility—we celebrate by not destroying the world.

Raymond Arnold has suggested many ways of observing Petrov Day. You can discuss it with your friends. You can hold a quiet, dignified ceremony (for example, with the beautiful booklet Jim Babcock created). But you can also play on hard mode: “During said ceremony, unveil a large red button. If anybody presses the button, the ceremony is over. Go home. Do not speak.”

In the comments of Ray’s post, Zvi asked the following question (about a variant where a cake gets destroyed):

I still don’t understand, in the context of the ceremony, what would cause anyone to push the button. Whether or not it would incinerate a cake, which would pretty much make you history’s greatest monster.

To which I replied:

The point isn’t that anyone sane would push the button. It’s that we, as a civilisation, are just going around building buttons (cf. nukes, AGI, etc) and so it’s good practice to put ourselves in the situation where any unilateralist can destroy something we all truly value. When I said the above, I was justifying why it was useful to have a ritual around Petrov Day, not why you would press the button. I can’t think of any good reason to press the button, and would be angry at anyone who did—they’re just decreasing trust and increasing fear of unilateralists. We still should have a ceremony where we all practice the art of sitting together and not pressing the button.

So this year on LessWrong, I thought we’d build ourselves a big red button. Instead of making everyone go home, this button (which you can find over the frontpage map) will shut down the Less Wrong frontpage for 24 hours.

Now, this isn’t a button for anyone. I know there are people with an internet access who will happily press buttons that do bad things. So today, I’ve emailed personalised launch codes to 125 LessWrong users, for us to practice the art of sitting together and not pressing harmful buttons[1]. If any users do submit a set of launch codes, tomorrow I’ll publish their username, and whose launch codes they were.

During Thursday 26th September, we will see whether the people with the codes can be trusted to not, unilaterally, destroy something valuable.

To all here on LessWrong today, I wish you a safe and stable Petrov Day.


[1] I picked the list quickly on Tuesday, mostly leaving out users I don’t really know, and a few people who I thought would press it (e.g. someone who has said in the past that they would). If this goes well we may do it again next year, with an expanded pool or more principled selection criteria. Though I think this is still a representative set—out of the 100+ users with over 1,000 karma who’ve logged in to LessWrong in the past month, the list includes 53% of them.

Added: Follow-Up to Petrov Day, 2019.