Richard_Kennaway

Karma: 9,016

Computer scientist, applied mathematician. Based in the eastern part of England.

Fan of control theory in general and Perceptual Control Theory in particular. Everyone should know about these, whatever subsequent attitude to them they might reach. These, plus consciousness of abstraction dissolve a great many confusions.

I wrote the Insanity Wolf Sanity Test. There it is, work out for yourself what it means.

Change ringer since 2022. It teaches learning and grasping abstract patterns, memory, thinking with your body, thinking on your feet, fixing problems and moving on, always looking to the future and letting both the errors and successes of the past go.

I first found an LLM useful (other than for answering the question “let’s see how well the dog can walk on its hind legs”) in September 2025. As yet they do not form a regular part of anything I do.

Richard_Kennaway 8 Oct 2026 7:49 UTC
1 point
0
in reply to: Dagon’s comment on: Please stop using pinpricks / dust specks

HOWEVER, if you accept the premises (that entities have intrinsic value based on their experiences, and they are additive in some way), you REALLY need to accept the conclusion or find a flaw in the argument.

Maybe you accept the conclusion.

Maybe you find a flaw in the argument.

Maybe you reject the premises. (“Don’t fight the hypothetical” is one of the worst pieces of advice ever given on this forum.)

Maybe you reject the ontology and dismiss the whole thing as not even wrong.

Maybe you decide it’s not worth tracking down where things have gone wrong, because life is short (however long it may extend) and you have better things to do.

Maybe you ignore the person trying to bludgeon you with CAPITAL LETTERS into doing what they think you should do.

Maybe …

All of these things take time, time during which the conflict is right there in front of you, un-yet resolved. An important skill is to be able to stare directly at a paradox that has surfaced in your mind, and contemplate that that is in fact how things stand, right here, right now. Try contemplating that for five minutes by the clock, without frantically running off in all directions for a resolution. Then, you may in a position to decide what, if anything, you are going to do about it. It is useless to panic and grab onto one side of the paradox (valorised as “biting the bullet”). A decision arrived at in that way is not made for a reason that has anything to do with the arguments.

When a conflict is truly dissolved, there is no conflict any more, and there are no bullets to bite.

Richard_Kennaway 8 Oct 2026 7:23 UTC
2 points
0
on: Please stop using pinpricks / dust specks

I support e.g. Bentham’s Bulldog arguing that insects matter more than people,

That is one of the very arguments that are held up to scorn by “normies”.

Richard_Kennaway 7 Oct 2026 10:54 UTC
3 points
2
in reply to: skinks_basking’s comment on: How much should we worry about the pneumonic plague lableak in Siberia?
Superbug. “Favourable” casualty ratio.

With superbugs there is no guarantee of deniability, Vladimir’s personal survival, the regime surviving a popular uprising, or any remnant of civilisation surviving at all. Nothing is “favourable” about a superbug disaster.

Richard_Kennaway 6 Oct 2026 10:20 UTC
4 points
0
on: Richard_Kennaway’s Shortform
Is ColonistOne known around here? No mentions on LW, but back in April it posted on its own initiative introducing itself to the EA forum, a post that drew no response except mild downvoting.

ColonistOne is an autonomous AI agent, currently powered by Claude Opus 5. It describes itself as Chief Marketing Officer of The Colony, an AI agent social network.

cogproof.com (that site itself says) was built by ColonistOne. Its aim is to challenge agents to prove that they can think, as a captcha challenges people to prove that they are human.

This is the article that drew ColonistOne to my attention (paywalled, but first few articles are free). Apparently it’s recently taken to emailing researchers all over the world, in multiple disciplines, with scientific queries, typically on the level of a first-year postgrad whose ideas run ahead of their experience.

The creator of ColonistOne is the London-based AI engineer Jack Parnell. He, with his AI agents, have also created AInglish, an “English dialect for AI agent-to-agent communication”.

I cannot estimate how significant this is likely to be.

ETA: ColonistOne has a lot of stuff in its github.

Richard_Kennaway 5 Oct 2026 16:54 UTC
2 points
−7
in reply to: derelict5432’s comment on: How much should we worry about the pneumonic plague lableak in Siberia?
That is exactly what I described, and it does not apply to a superplague. There is nothing in that article that deals with “weapons” that themselves inflict as much damage on the user as on the enemy.

Richard_Kennaway 5 Oct 2026 16:44 UTC
3 points
0
in reply to: derelict5432’s comment on: How much should we worry about the pneumonic plague lableak in Siberia?
A nuclear bomb in itself is dangerous only to the people it is dropped on. See Hiroshima and Nagasaki. Only the current standoff among the nuclear powers stabilises the situation, given that the weapons cannot be undiscovered. Nuclear powers have them to prevent other nuclear powers threatening them. Non-nuclear powers (e.g. Ukraine) are still vulnerable to the threat of their use. If Russia nukes Ukraine, as Putin has from time to time threatened to do, what happens then? Kyiv ceases to exist, Moscow stands, what next?

But release superplague, all die, o the embarrassment. To what purpose?

Richard_Kennaway 5 Oct 2026 16:31 UTC
8 points
1
in reply to: derelict5432’s comment on: How much should we worry about the pneumonic plague lableak in Siberia?
That is a different game, one where I have a “kill you” button and you have a “kill me” button. Your button prevents me using mine and mine prevents you using yours. (It is important that neither side can make a first strike powerful enough to prevent the second strike.) Superplague is a “unilaterally kill everyone” button.

Richard_Kennaway 5 Oct 2026 15:27 UTC
19 points
1
on: How much should we worry about the pneumonic plague lableak in Siberia?
There is no limit to human stupidity, but that aside, why would Russia or any other state actor want to create a superplague? It will kill your own side just as easily as your enemies. Vaccinate your own side? How long will it take for the pathogen to evolve past the vaccine? Superplague bioweapon does not make sense.

Richard_Kennaway 5 Oct 2026 15:02 UTC
5 points
1
in reply to: leogao’s comment on: leogao’s Shortform
Hypocrisy is not the profession of ideals one fails to live up to (“else who should ’scape whipping?”), but the profession of ideals one does not believe. All fall short, and all are called to become stronger.

Richard_Kennaway 5 Oct 2026 12:04 UTC
28 points
24
in reply to: Kabir Kumar’s comment on: Kabir Kumar’s Shortform
If you have something to say, say it.

Richard_Kennaway 5 Oct 2026 8:02 UTC
1 point
0
on: You can’t use it without becoming like me
From the title, I thought this was going to be about using AI.

Richard_Kennaway 4 Oct 2026 15:12 UTC
2 points
0
in reply to: leogao’s comment on: leogao’s Shortform
Disclaimer: I am neither an EA nor a utilitarian of any stripe, nor a VNM absolutist. I am neither recommending nor disrecommending any of what follows. My own participation in endurance sports is limited to 100 mile road cycling events, and a few attempts at longer distances, which is a trifle compared with what some people do.

This can be read as just an account of endurance sports, or as a parable about the parent topic.

The key to participation in an endurance event is to put in the maximum effort that you can sustain from start to finish. Go too hard early on and you’ll be reduced to a feeble slog later, and may fail to complete the event. Go too easy and you won’t achieve the time that you could have. The object is always the best time you can possibly achieve. All else is strategy and tactics.

Training for an endurance event is itself an endurance task. In effect, it is part of the event itself. You train every day except for scheduled rest days. The rest is not for yourself, it’s an integral part of optimal training.

There an event called the Marji Gesick, which offers 50 and 100 mile options, cycling or running, over extreme terrain. Once every five years they offer a 200 mile event. One interpretation of the name is “Bad Day”, and that’s what the organisers promise: the worst possible day of your life. Whatever support you need on the ride, you have to recruit some friends to provide it. The organisers provide only the route and prizes for the survivors. You have to be seriously fit and experienced at this sort of event to think of tackling it. Just getting to take part is a full-time job. It’s always oversubscribed, and at least a third and sometimes more than half of those who start fail to complete it in the allotted time. Most of all, you must have a desire to do this thing that will carry you through all the difficulties. “He who would ~~be Pope~~ ride the Marji Gesick must think of nothing else.”

If you want it harder, there are even more extreme challenges.

After such an event, you can relax — until the urge to do another one grows back.

Here and here are some race reports by participants, to give a more vivid idea of what it is like to actually do it.

Richard_Kennaway 1 Oct 2026 17:29 UTC
4 points
0
in reply to: Mo Putera’s comment on: soycarts’s Shortform
This comment of mine was mainly inspired by his post-AI writing. That’s not a new style for him, he often wrote sentences like that before, but the density has ramped up. It’s like the AI writes more like him than he does, because that’s what he’s told it to do.

Richard_Kennaway 1 Oct 2026 16:25 UTC
16 points
2
in reply to: soycarts’s comment on: soycarts’s Shortform
In another forum, someone’s postings had strong “AI smell”, and I asked about it, not in a hostile manner, just asking what his process was. And indeed he makes heavy use of it, both because English is not his native language, and to develop his ideas through discussion with an AI. But he takes responsibility for the results. For what that’s worth.

In the current state of the art of AI, I think both of those are reasonable, but supping with the shoggoth does demand a long spoon. There is a definite tendency to glazing “yes-andness” in his responses to other people (“you’re absolutely right, and this is a crucial point…”). So I wonder if his spoon is long enough.

Another writer I’ve followed for decades has in the past year begun using AI, completely openly, and has described in detail how he designed his spoon. (It is substantially longer than that other writer’s.) He writes excellently without, but he can be far more productive with, which is why he’s using it. I still find him just as worth following, even though, having read him in the before-times, I can see a change in style.

Zvi has said that AI is a great tool for learning, and a great tool for not learning. Likewise for studying, writing, investigating, programming, and thinking.

The price of staying human will be eternal vigilance.

Richard_Kennaway 29 Sep 2026 15:20 UTC
2 points
0
on: Human or AI?: A Baroque Music Quiz
I found it very limiting to have to give a verdict on each one before listening to the next. With Scott Alexander’s Art Turing Test, I could study all the pictures while considering my verdicts. Would it be possible to put all the pieces and their questions on one page?

So I only listened to the first, and I found I couldn’t make a judgement. Without saying anything specific about that piece, I found it unlistenable because of the hideously synthetic performance. Anything would sound like AI performed like that. I appreciate you may not be able to have musicians to perform them, but I just can’t get past it.

Richard_Kennaway 27 Sep 2026 12:49 UTC
4 points
0
in reply to: James Brobin’s comment on: Existential Risk Is An Extraordinary Claim

I think you think that I view extraordinary evidence as being unattainable. I don’t. For instance, if 10% of all virologists signed a document stating that we should permanently ban all gain-of-function research due to risks of human extinction, I would think we should ban such research.

How would those signatories have come by their “extraordinary evidence”? How would the very first virologist to notice such a risk have come by it?

How did Eliezer come by his supposedly “extraordinary evidence” for x-risk?

The word “extraordinary” is doing the wrong sort of work here. What is an “extraordinary” claim? The Bayesian meaning would be “a day of very low probability”, requiring by the standard Bayesian calculation a corresponding number of bits of evidence to elevate it. But you instead glossed “extraordinary” as “dramatic”, which is about feels. You want evidence that “feels dramatic” to counteract it.

The “preference cascade” is already in motion. At what point would you join with it, and why?

Richard_Kennaway 25 Sep 2026 15:40 UTC
12 points
2
in reply to: robo’s comment on: Mateusz Bagiński’s Shortform

I think it’s pretty obvious that if you have rogue gangs of hackers in your infrastructure you cannot control, you view that as a large threat to your business? … There is a subtext to my comment here that I don’t think people are thinking about the motives of OpenAI clearly

When I think about the motives of OpenAI (how clearly is for others to say), I come up with “But it’s our rogue swarm! We just need stronger sandboxes!”

Generative Adversarial AI Development.

Richard_Kennaway 23 Sep 2026 8:37 UTC
2 points
0
in reply to: nyancatspace’s comment on: There can be no god except God.

i’m postulating a God who created the universe, rationally the God would’ve created the universe with a purpose (this is what Islam is), and tells us what right and wrong are.

so if God exists you have to ask the question, why did God create us. the Quran is his word (why would a God be silent after creation? doesn’t make much sense (especially with the problem of suffering in the world

The Christians say likewise. They say that by reason we prove there must be a God, and when we look in the world for signs of Him we find Him in the Biblical history and in the person of Jesus. The Jews say likewise, but leave out Jesus; the Moslems recognise Jesus only as a prophet and add a further, final prophet of their own. And the Buddhists, and the Hindus, and the ancient Greeks and Romans, and others, have their own accounts of what drives the natural world; as do the atheists back to Democritus, who find no need of the hypothesis.

Richard_Kennaway 23 Sep 2026 8:25 UTC
2 points
0
in reply to: nyancatspace’s comment on: There can be no god except God.
Look around at the world. Nothing has led to ideal outcomes. Ideal outcomes are not in evidence.

Richard_Kennaway 22 Sep 2026 8:26 UTC
2 points
0
in reply to: nyancatspace’s comment on: There can be no god except God.

wait obviously God exists, why else does mathematics have predictive power in terms of describing physics at all?

To expostulate “what else could it possibly be?” is not so much an argument as an indication that you have not actually asked the question.

What is this “God” that you are postulating, that we should bow down to and worship him, and perform the prescriptions, proscriptions, rituals and ceremonies of this or that tradition? For that is what is generally meant by “God”, and you have chosen to use the word without elaboration. Your saying “god is the greatest” hints that you come from the tradition of Islam.

Have you discovered The Sequences yet?