Posts
Comments
although I'm not sure if he claims that every ruliad must have observers
Of course yes, since there's only one ruliad by definition, and we’re observers living inside it.
In Wolfram terms I think the question would more be like : "does every slice in rulial space (or every rulial reference frame) has an observer ?"
Possibly of interest : https://writings.stephenwolfram.com/2023/12/observer-theory/
One part that I don’t see as sufficiently emphasized is the "as a time-persistent pattern" part. It seems to me that that part is bringing with it a lot of constraints on what partition languages yield time-persistent patterns.
Do we have the value of the sum as a function of x, before going to the limit as x goes to 0 ? If yes, it would help (bonus points if it can be proven in a few lines).
To me, the correct way to do this is to compute the (implied) rate of returns of solar investment over the lifetime of the panels :
(x-x^25)/(1-x)=20/3.2 => x ~ 0.85
x = 1/(1+r) => x ~ 0.17
So yes, a 17% rate of returns is insanely good (if the two assumptions, "25 years lifetime" and "3.2k/year", stands) and will beat pretty much every other investment.
(which should makes you suspicious about the assumptions)
A person stating which entities they admit into their hypotheses, that others may not (“I believe in atoms”; “I believe in God”).
This one does not looks like the others to me.
I am not a community organizer, but if I was, I would just edict the MAD rule : if any dispute can’t be reasonably settled (in a way that would escalate to a panel), both the plaintiff and the defendant are kicked out of the community. The ratio of assholes to victims being low (hopefully), frame it as a noble sacrifice from the victim to keep the rest of the community safe (whoever is the actual victim here). "Some of you may die, but it’s a sacrifice I’m willing to make".
True, you’re slightly disincentivizing the reporting of true abuses (but not so much, since you’re saving a lot on the trauma of investigation). But you’re also heavily disincentivizing both false accusations and "I’m a good manipulator so I’m confident I will be able to get away with it". Add to it the man-hours saved, and I’m pretty confident it is a net gain — as long as you can enforce it.
As a bonus, if a dispute end up with community bans, you can be reasonably sure that the plaintiff was the victim. But of course, that only works if you don’t go from that observation to concluding "therefore the victim is un-banned". A noble sacrifice, really, in service of both community and truth. A sacrifice nonetheless.
category boundaries should be drawn for epistemic and not instrumental reasons
Sounds very wrong to me. In my view, computationally unbounded agents don’t need categories at all ; categories are a way for computationally bounded agents to approximate perfect Bayesian reasoning, and how to judge the quality of the approximation will depend on the agent goals — different agents with different goals will care differently about a similar error.
(It's actually somewhat interesting; the logarithmic score doesn't work as a measure of category-system goodness because it can only reward you for the probability you assign to the exact answer, but we want "partial credit" for almost-right answers, so the expected squared error is actually better here, contrary to what you said in the "Technical Explanation" about what Bayesian statisticians do)
Yes, exactly. When you’re at the point when you’re deciding between log-loss and MSE, you’re no longer doing pure epistemics, you’re entering the realm of decision theory ; you’re crafting a measure of how good your approximation is, a measure that can and should be tailored to your specific goals as a rational agent. log-loss and MSE are only two possibilities in a vast universe of possible such measures, ones that are quite generic and therefore not optimal for a given agent goals.
Naive question : about immunogenicity, what are the problems with the obvious strategy to counter it ? (target the thymus first to "whitelist" the delivery method).
A prime example of what (I believe) Yudkowsky is talking about in this bullet point is Social Desirability Bias.
"What is the highest cost we are willing to spend in order to save a single child dying from leukemia ?". Obviously the correct answer is not infinite. Obviously teaching an AI that the answer to this class of questions is "infinite" is lethal. Also, incidentally, most humans will reply "infinite" to this question.
E=mc2 is only valid in the rest frame of the system. The formula in a non-rest frame is E^2 - p^2*c ^2=m^2*c^4 See https://en.wikipedia.org/wiki/Four-momentum for more details.
In general Perpetuals trade above the price of the underlying coins
I’m confused by this. Doesn’t this means that long positions almost always pays short positions, even if the index is increasing ? If so, why would anyone go long on the future ?
What’s the point of buying bitcoins in your scheme ?
I don’t understand where that 1/2 comes from. Unless I have made a gross mistake P(A|A => B) < P(A) even if P(A&B) > P(A¬(B)). In your first example, if I swap P(AB) and P(A¬(B)) so that P(AB) = .5 and P(A¬(B))=.3 then P(A|A=>B) = .5/.7 ~ 0.71 < 0.8 = P(A).
You’re right, the 2^(-3/4) (and the 2^1/4) is probably quantitatively wrong (unless each side is perfectly heat-conducing but both are isolated from each other. Or if the planet is a coin facing the sun. You know, spherical cows in a vacuum…). But I don’t think that changes the qualitative conclusion, which hold as long as the bright side is hotter but not twice as hotter than the perfectly-heat-conducing planet.
Given perfect conduction (uniform surface temperature, bright side and dark side have the same temperature at all times), https://en.wikipedia.org/wiki/Black-body_radiation#Temperature_relation_between_a_planet_and_its_star applies : temperature does not depend on rotation speed. Then T = T_sun sqrt(R_sun/(2D)) ; it is the temperature T that balance incoming radiation P_inc = pi (R_planet^2) (R_sun^2) (T_sun^4)/(D^2) and emitted radiation P_em = 4 pi (R_planet^2) * T^4
Let's suppose no conduction at all. The bright side and the dark side does not exchange heat at all. Let's take two limiting cases : tide-locked planet, and an "infinitely fast" fliping planet.
In the first case, the dark side of the planet is at absolute 0. The bright side of the planet receives the same incoming radiation but emit half its radiation (halved surface) -- change 4 pi to 2 pi in P_em. Its temperature is T_bright = 2^(1/4) T. Average temperature of the planet is (0+T_bright)/2=2^(-3/4) T
In the second case, each side gets half the incoming power from the sun and radiates half the energy (surface halved). Average surface temperature of the planet is the same that the average surface temperature of any side, which is the same temperature that the perfectly conducting planet, T (the .5 from halved incoming power and .5 from halved outgoing radiation cancel each other).
So : rotation raises temperature of a non-perfectly-heat-conducing planet, bringing its surface temperature closer to the perfectly-heat-conducting planet surface temperature.
Yes, "intrication" is the standard translation of "entanglement" in QM. But nobody else uses it, and therefore I fear there is an obvious failure mode where someone Googles it and start shouting "WTF is that?"
"évidence" the noun is just a shorthand for "obvious thing" (most typical usage is « C’est l’évidence même » = “It’s obvious”. « Ce n’est pas la peine d’asséner de telles évidences » = “Such obvious things are not worth stating”).
Great suggestion, I’ll look into that.
Yes, that means « obvious »/« self-evident »
I’m trying to translate some material from LessWrong for a friend (interested with various subjects aborded here, but can’t read english…), and I’m struggling to find the best translation for “evidence”. I have many candidates, but every one of them is a little bit off relative to the connotation of "evidence". Since it’s a so central term in all the writings here, I figured out that it could not be bad to spend a little time finding a really good translation, rather than a just-okayish one.
English readers :
- Could you find a few different sentences that would cover all (slighty differenrt) usages of evidence ? The objective is, if my translation fit well in all those propositions, there is good chances that it will fit well in everything i may want to translate. For example, from wiki : “Evidence for a given theory is the observation of an event that is more likely to occur if the theory is true than if it is false” “Generalization from fictional evidence” “Conservation of expected evidence”. I except that finding a translation that will cover equally well those three usages will basically cover any usage, but can you think of a 4th usage that may prove problematic even for a term that fit well for the 3 others ?
- What would be the less bad synonym of “evidence” : clue, proof, observation, sign (that’s basically my best candidates, translated back in english). I dislike all of them, but that’s the best candidates I found, translated back in english. (substitute evidence in all the test sentences abole, and you will understand my problem. “Clue for a given theory…” is somewhat good, but “conservation of expected clue” less so…)
French readers, if any :
J’ai comme candidats : « preuve », « indice », « signe », « observation ». D’autres propositions ? Laquelle vous semble la meilleure ?
Thanks for your cooperation.
(and don’t get me started on “entangled with”, I think I will lose much hair trying to find an acceptable translation for that one. French sucks.)
Can’t tell for the Romantic Manifesto, but in Atlas Shrugged, Ayn Rand uses the word “value” as a synonym of “rule of conduct”. For example, she argue that “rational evaluation” is a correct value for man in the same way that “flying” is a correct value for birds.
She calls her philosophy objectivism because the thinks that correct values, which means rules of conduct that leads to environmental fitness (in her words says : “survival”), are objective.
I still don’t understand HOW cancer kills.
I mean, we just have some additional cells who does not perform their normal functionality. But we still have a big bunch of normal, functioning cells.
In my (very very) distant family, someone died from lung cancer a few months ago. I still don’t understand the link between the few additional cells in his lungs and the acute hepatic failure that killed him.
I read somewhere that a primary cancer seldom kills ; most of the time the metastasic-induced does. Why ? There should be far more "bad" cells in the primary site, doesn’t it ?
(medecine illiterate there, sorry if half of my assumptions are wrong)
A few day ago, I saw an interesting article on a site somewhat related to lesswrong. Unfortunately I didn’t have the time to read it, so I bookmarked it.
Computer crashed, lost my last bookmarks and now I spent 2 hours trying to find this article, without luck. Here is the idea of the article, in a nutshell : we human are somewhat a king of learning machine, trying to build a model of the “reality”. In ML, overfitting means that in insisting too much on fitting the data, we actually get a worse out-of-sample performance (because we start to fit the modeling noise and the stochastic noise). By carrying this ML idea into the human realm, we can argue that insisting too much on consistency can be a liability rather than an asset in our model-building.
Does that decription rings someone bells ? If yes, please link the article :)