Hacker News new | past | comments | ask | show | jobs | submit login

It's Yudkowsky from LessWrong, he is actually a quite insightful thinker, but his disciples claim stuff like this all the time. LessWrong is an extreme echo chamber.

For me the breaking point was Roko's Basilisk, a self-proclaimed "most-dangerous idea of our time".




Are you honestly claiming that the author believes that Yudkowsky was literally the first person in history to suggest making lights brighter?

Also, you should read anything Yudkowsky himself has written about Roko’s basilisk. He’s never claimed that it could work, or that it’s dangerous at all.

(“Why did he ban it then?” Because he wanted to prevent it from inspiring an infohazard brainstorming session that could produce a real danger, and he apparently hadn’t heard about the Streisand effect.)


From the article:

>Elevator pitch: Bring enough light to simulate daylight into your home and office.

>This idea has been shared in Less Wrong circles for a couple years. Yudkowsky wrote Inadequate Equilibria in 2017 where he and his wife invented the idea


Yes, I read the same article. Do you believe, based on a single word, that the author believes that literally zero people in the entire course of human history have ever suggested “brighter lights” until Yudkowsky came along?


No one in the parent posts has said anything about "literally zero people".

The OP made a claim ("invented in 2017"), and this very claim is being criticized.

I found perplexing that you interpret the very same words in almost opposite ways.


I cant find it again but I read an article about how brightness goes up with technology and human spend roughly the same proportion of income on light. Kind of disproving the led efficiency is good for the environment argument. Not revolutionary, better to break that pattern.


You contradict yourself.

If humans spend same proportion of income on lumens, and you get more lumens per watt, then you will use less wattage per the same proportion of income. Ergo LEDs are more efficient.


Are you thinking about the cost of the physical lights? Because what is being discussed is the cost of energy.


Watt-hours are energy. If cost of energy is decreasing and you get more lumens per watt with LEDs then the benefits stack, they do not cancel.


If energy price is going down, then even more watts are being bought (since spending is fixed).


Not all those watts are going into lighting. If spending is fixed on lighting but total expenditure is up that means energy is being bought for things other than lighting. Kirchoff's law applied to the grid in toto.

Adds up considering the proliferation of computing devices, particularly cryptomining.


I don't know who any of these people are, but that is exactly what the article says, so why should we not believe that that's what the author believes?


Because it’s far more likely that the author chose slightly wrong words that ended up conflating a specific invention with being the first human “to invent the idea” that “hey, brighter lights would be closer to obviously brighter daylight” only 5 years ago. I’d wager that the author realizes that more than half of adults today had that insight sometime prior to 2017.


Use of the word "invention" is usually a red flag. I've certainly done a number of things I never had knowledge of any prior art on, but I never said I "invented" those things.

If I wanted to, I'd probably go and research prior art first before making use of the word.

Also, I'd say that the article author is not referring just to brighter light as the invention, but rather as the use of extremely bright light indoors for different psychological effects.


Some people read websites the way they would read a math proof. One tiny error and the whole thing comes down. A more charitable reading is often perfectly appropriate.


The LessWrong community is known for self-citations all over the place and heavy use of jargon created by themselves, and words redefined as terms-of-art. In light of this, if a member claims that one of their prominent figures "invented" something, you better take the claim at face value.


On the flip side, interpreting everything you read through your own lenses of experience and bias instead of assuming that the author meant what they wrote seems fraught with pitfalls.


If you are reading it as a math proof, I am only disputing the claim when the idea was invented.

As I don't bring up the rest of the article's points, I am neither supporting nor refuting them with that particular comment.

Math (or actually, logic) is a bit more resilient than you make it out to be.


Yudkowsky claimed that the exact basilisk as described by Roko wouldn't work, but he also believes that something like it would.


Sort of. He believes that if people try to come up with infohazards and then immediately spread them, there’s a nonzero chance that they succeed. It’s unlikely, but still not something to encourage.

(This is not the same thing as “being certain that a minor modification of Roko’s basilisk would produce an evil time-travelling AI,” to be clear.)


This kind of utter gibberish is exactly why people don't take LessWrong seriously.


“Ideas can be harmful” is gibberish?


The amount of self-importance on display is what makes it sound like gibberish.

  A: "I have the most ground-breaking idea ever!"
  B: "Okay, let's hear it?"
  A: "No, it's too dangerous, it may well destroy the world, but trust
      me it's brilliant and completely novel."
It's a religion at that point.


How about bitcoin? It's not so much an idea as a piece of software, but being predicted on wasting energy it's going to do a lot of harm. That's an example of this class of game-theory based ideas.

I believe that there are other ideas like that, things that set up spirals that are a lot worse than just "destroy energy to make money". Obviously they're not going to be universally compelling, an idea that's dangerous like that in one context might be perfectly safe in another. It depends on the target audience.


>>It's not so much an idea as a piece of software, but being predicted on wasting energy it's going to do a lot of harm. That's an example of this class of game-theory based ideas.

It's not a waste if the product the PoW generates is valued as equal to or greater than the cost of generating the PoW, and the market ensures that is the case, as miners cannot operating at a loss.

Where it can be socially harmful is when energy consumers don't pay for the cost of the negative externalities that the energy they purchase created in its generation, but that applies to all energy usage, not just that used in generating PoW.

You should know that costly signalling strategies are widely employed in nature because costly signals are reliable:

https://en.m.wikipedia.org/wiki/Handicap_principle


Money laundering is useful but that doesn't mean it's good. A Ponzi scheme can generate wealth but it can't generate value. Wealth without value is maybe not wasteful, but it is harmful.

> Where it can be socially harmful is when energy consumers don't pay for the cost of the negative externalities that the energy they purchase created in its generation, but that applies to all energy usage, not just that used in generating PoW.

Bitcoin alone uses as much power as a medium-sized nation. The increased power usage has caused old fossil-fuel power plants to come back online.


1. There is no evidence that cryptocurrency is widely used to launder money

2. Money laundering can improve economic efficiency when the crimes that generate the illicit revenue are socially beneficial, as is the case when people in China escape the Communist government's capital controls, or when people in Venezuela escape their Communist government's steep inflation tax via mandates to use the state's fiat currency.

Even in ostensibly free societies, socially harmful laws that repress the right to voluntarily interact, like price controls on N95 masks that prolonged a shortage in them for months, or heavy taxation to pay the pensions of a bloated and over-paid public sector kleptocracy, can be socially beneficial to undermine.

So your deduction that money laundering or other forms of escape from institutional rules, is socially harmful, is overly simplistic. In some cases it is harmful, and in some cases it's beneficial.

3. Cryptocurrencies are by definition not ponzi schemes. You could argue they're speculative bubbles, but that is itself speculation.

>>Bitcoin alone uses as much power as a medium-sized nation.

Because it generates billions of dollars of currency a year, and its production is costly in energy and affordable in labor. The energy consumption of labor-intensive industries obfuscates the real energy requirements of those industries, by effectively outsourcing the energy consumption to the workers.


It generates zero dollars. Every dollar in all cryptocurrency markets comes directly from someone buying in. Claiming otherwise is basically a blatant lie.


It's not generating dollars. It's generating a digital asset that is worth something. Of course the value of any commodity is based on what people are willing to buy it for, so your observation in no way discredits the point, that Bitcoin mining is generating something that objectively, based on what the market is paying, has value.


> That's an example of this class of game-theory based ideas.

That's a good point on its own, but it's not a response. At no time did I or anyone here in this thread assert that "info hazards" do not exist. What is being discussed is whether it's worthwhile to be deferential to a person/community who is making claims of being in possession of an exceptionally catastrophic piece of information that by its nature would preclude public scrutiny on ethical grounds.


I don't think Yudkowsky is making that claim anywhere?

>At no time did I or anyone here in this thread assert that "info hazards" do not exist.

>What is being discussed is whether it's worthwhile to be deferential to a person who is making claims

It's honestly really not that clear what you're trying to discuss, certainly you haven't been making a clear statement like that. I don't think anyone is making claims of being in possession of such an infohazard for one thing, and if there were I don't think anyone would be saying you should be deferential towards someone who claimed to have one, sight unseen.

I think you're arguing with a bit of a straw man? Like no one is being all "let's be deferential towards Yudkowsky because he has the idea equivalent of a nuke". He doesn't, to the best of my knowledge, claim to have the idea equivalent of a nuke.

I assumed you were claiming that info-hazards don't exist because otherwise what exactly are you claiming?

Anyway, if that was your point, granted. Let's not be differential towards people who claim to have info-hazards, unless they can go into a room and make someone laugh themselves to death or something.

https://www.youtube.com/watch?v=FBWr1KtnRcI


He doesn't, to the best of my knowledge, claim to have the idea equivalent of a nuke.

Well, he isn't willing to reveal how he convinced his opponents to let the AI out of the box.


Yeah, but his win rate is probably not that much above control.


Can you link me to when anyone on LessWrong has behaved like your "A"?


Now you've just generalised the statement until it becomes a truism. So you are now essentially claiming the statement is garbage because it is devoid of useful meaning.


[removed by author as off-topic]


In hindsight, I should not have included the comment about Roko's Basilisk. It was really nothing more than a personal observation that had nothing to do with the topic at hand. As a result, this thread attracted a lot of people who have very strong opinions on an off-topic matter, and for that I'm sorry. It muddled an already tense discussion on whether any sane person should take the claim of inventing brightly lit homes in 2017 literally or not. For what it's worth, I would have deleted the Basilisk portion if people hadn't already responded to it.


Cool, I didn't even know comments could be deleted! I appreciate your reaction.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: