Fun fact: AGI-related concepts can fit quite well even into regular D&D games. Last year I ran a Planescape campaign where the characters had to stop a literal paperclip maximizer (which started turning paperclips into paperclip golems making even more paperclips). Maybe not exactly the best way to prepare ourselves for real-world scenarios, but at the very least it was useful to introduce the concept to my players.
Bezzi
Karma: 597
Trying to extend the metaphor: in order to see monsters you have to look through the Overton Window. But first you have to shift the window and reach the monster; otherwise it will remain invisible.
However, I would remark that the Overton Window does not necessarily always shift towards a better world. Sometimes there aren’t any monsters around, and you end up summoning one.
For example, push too much towards Killing Animals Is Really Bad, and you might get Killing Animals Is Worse Than Killing Humans. Maybe I’m not enlightened enough, but I doubt that an enlightened human would value the life of a dog more than the life of another human… yet this position seems to be actually hold by a disturbingly large number of people (10-seconds Google search, not double-checked: 14% of 537 people would save one random dog over a foreign tourist… raise to 40% when asked to choose between their dog and a stranger).
The point of Being the (Pareto) Best in the World is that you can quite easily become a true expert in X if X is a sufficiently narrow intersection of domains. For example, the best person for writing music notation software is neither a programmer nor a musician, but someone having reasonable skills in both music and programming.
This post seems to make a broader claim, namely that multi-domain expertise can improve your confidence in every single domain, even if you never face actual cross-domains problems. I find this point much less obvious. If you just want to raise a general level of overall confidence, fine. But I fail to see how your expertise in a field could possibly raise your confidence in another (unrelated) field. When your piano teacher is yelling at your terrible execution of the Well-Tempered Clavier, you can’t just repeat to yourself “Screw it, I’ve a PhD in Computer Science!”… in order to improve, you need to actually exercise at the instrument, and your PhD isn’t going to help you with that.
After ten years spent between computer science and formal music training, my overall evaluation is that formal music training will definitely not make you a better programmer in general (and vice versa). The only meaningful exception for me were the music notation software courses aimed at musicians, which I found ridiculously easy for obvious reasons.
There are an infinite number of integers, so there must be an infinite number of primes.
This is the main thing I am going to retain from this post (sorry, Lakshmi).
Until now, if asked to prove why there are infinite prime numbers, I would have answered “because Euclid’s Theorem” (Euclid’s proof is very easy to remember and explain verbally without formal notation). It never occurred to me that I could have just answered “because there are infinite integers, duh”.
What really shocks me is that I’ve just spent 20 minutes digging “infinite primes” results, and no one points out this. Every “infinite prime numbers” page I found reports either Euclid’s Theorem or something even more complicated (it seems that people still publish 1-page papers proving the infinitude of primes). I’m very confused. I mean, “infinite integers imply infinite primes” is not a fully-fledged formal proof, but it shouldn’t be difficult to write an actual formal proof from that statement… am I failing to see some obvious things? (like, maybe that proof boils down to Euclid’s Theorem?)
For some reason nobody points out that what the humans are doing is absolutely 100% the correct thing to do! Our theoretical Homo Economicus who accepts 1 cent is clearly beaten by real life Becky who refuses to settle for less than 50⁄50, and so gets $50 every time!
I am still not convinced that staying intentionally out of a perfect-subgame equilibrium should be a good choice. What do you mean with “every time”? The standard version of the Ultimatum Game assumes to be played once, and that you will never encounter the same opponent again. There is no concept of “this player is well-known for rejecting anything less than 50/50″.
Suppose to play the Ultimatum Game not in person, but via Twitter accounts with silly nicknames. Every account randomly change name after each game, so you cannot distinguish one player from another. And all players play from their room with absolutely no one else wandering around. You are not going to be judged greedy or dumb or whatever by anyone. Do you still reject 1 cent in this scenario?
Greetings, LWers!
I’ve finally
found the timemade up my mind to write this, so here I am.I’ve noticed that many new members have stumbled upon the rationalist community because of HPMOR. As I never read fanfiction sites (or sites talking about fanfiction sites), my case was quite different. For some reason I distinctly remember the ridiculously long chain of links that brought me here, so I’ll post the whole list just to give an idea of how long it can take to realize the existence of a site like LessWrong:
Search for insights about the P=NP conjecture during my PhD.
Find the P-versus-NP page, a very good summary that also links to this excellent post by Scott Aaronson.
Start reading Scott Aaronson’s blog.
Scott Aaronson mentions Unsong (in this post).
Start reading Unsong.
Return reading Scott Aaronson’s blog.
Scott Aaronson dedicate this post to the infamous NYT article about Scott Alexander.
Fail to realize that Scott Alexander is the author of Unsong.
Scott Aaronson directly quote I Can Tolerate Anything Except The Outgroup (in this post).
Follow the link and read my first SSC post.
Start reading SSC from some top posts.
Still fail to realize that Scott Alexander is the author of Unsong.
Finally notice the “Scott also writes Unsong” note in the about page.
Continue reading SSC.
SSC mentions LessWrong.
Finally land on LW frontpage.
Start reading the Sequences.
Start reading the Codex.
Start reading HPMOR (directly from LW).
Finally sign up (after several months of lurking).
I’m not sure about which conclusion we can draw from this. Maybe that wondering about P=NP has a small chance of making you a better rationalist. Maybe that you can spend more than one year following a computer scientist professor who declares himself on the fringes of the “rationalist movement” without realizing that a rationalist movement even exists (in my defense, I started reading Shtetl-Optimized in mid-2019, and I didn’t exactly dig through the older posts… still, it took me more than one year to finally land on LW). In hindsight, many posts from Scott Aaronson are quite obviously related to rationalist concepts. For example, I first learned about the classical paperclip maximizer from Shtetl-Optimized (here), but even googling “paperclip maximizer” I didn’t land on the rationalist blogosphere. I just learned the paperclip maximizer classical description. It may be worth mentions that after reading the relevant Wikipedia entry, my first thought was “an amoral paperclip maximizer can fit perfectly into my Planescape campaign”, which indicates that maybe I’m a bit too much addicted to D&D.
Of course there are people who believe this! Actually, PETA-like morality seems to be increasingly common (I highly doubt that 14% of people would have saved random dogs over humans in 1901). My point was that an increment in PETA-like morality is not necessarily a good thing that will bring us all closer to the next stage of humanity.
Suppose that Future Buddha declares that there is absolutely no moral difference between me and one random dog, and he’ll happily flip a coin to decide which one deserves to live. I’ve thought about it for several hours, but I still lean towards “Future Buddha is an idiot”.
Well, I’ve never used online dating apps, but to me the comparison with in-person dating seems unfair in an even broader sense… with online dating apps you’ll find only people who are actually interested in dating in the first place (eg not me). I think this should be emphasized a bit more, because not absolutely everyone reach the level at which you are ready to metaphorically wear a glowing sign reading “please date me”.
Well, I agree that in general you should ask consent before pulling people into any game, but I suppose that part of the purpose was precisely to see how people react to random responsibilities (which can definitely happen in real life). I mean, the Soviet probably didn’t bother to get Petrov’s consent before putting him in the control room. And all that’s required in our case is basically “please do nothing”… I didn’t receive the email, but I don’t think I would be upset by one message just asking to be ignored (an email asking me to actively do something to prevent destruction would have been a different kettle of fish).
Full disclosure: I clicked the button. Actually, I misclicked the button while hovering on it. I suppose that’s the reason why GitHub and similar services are very careful to hide the “delete repository” button behind long page scrolls and also add an additional “are you absolutely sure?” popup.
For the next Petrov Day, I think we should at least add the blocking popup instead of just having an “Are you sure?” title over the button. Being tricked into pushing the button is one thing, but it should not be possible to push the button purely by accident.
I agree, but the people who actually received the codes are supposed to be carefully selected LW users, not totally random people. I would be quite impressed to learn that someone between those 100 users didn’t actually understand the context (on the other hand, I do expect random LW users who didn’t get codes to press the Red Button for the lulz without necessarily knowing the context, and I agree they shouldn’t be blamed for this).
That said, adding more things clarifying the context is probably good. Petrov himself surely didn’t have the context problem.
Ok, then I publicly declare to be quite impressed.
(I’ll treat this as further evidence that inferential distances tend to be longer than expected)
Speaking of useless adaptations, Weapons + Armor is only useful up to 10. If your Weapons + Armor exceeds 10 the extra armor is useless. 40 submissions (7%) had Weapons + Armor more than 10. I think the idea here was to create big exciting powerful monsters.
Actually, is even worse than this. Weapons 5 + Armor 5 seems fine, but Weapons 8 + Armor 1 isstrictly better(assuming you wanted a successful predator).Basically, any Weapons + Armor >= 10 combination is dominated by another Weapons + Armor >= 10 combination having +3 Weapons and −4 Armor: same cost and same protection, but with more effective weapons.That said, I’ve submitted a Wurm with Weapons 6 + Armor 6 anyway, because my inner Dungeon Master kept saying “It’s a sodding Wurm, it should have higher Armor Class!”. Weapons 9 + Armor 2 on a giant desert predator seemed just wrong (can someone please provide real-world examples of such “low-armored” deadly predators?).
Right, maybe it’s not always better. But if we restrict to “invincible” predators (Weapons + Armor >= 10), it is, since no one can prey upon you anyway. I’ve edited the previous comment.
I meant “restrict to both designs having Weapons + Armor >= 10” (which I admit may be a moot point since Weapons + Armor > 10 is out of the Pareto frontier anyway).
Weapons 6 + Armor 8 is obviously strictly worse than Weapons 6 + Armor 4 (cost less and do the same), but it’s even worse than Weapons 9 + Armor 4 (same cost and do more). And Weapons 9 + Armor 4 is strictly worse than Weapons 10 alone.
I agree, most winners so far are Armor 10 + Antivenom (the cheapest way to become invincible).
So far this is the only biome with non-invincible winners. I suspect that having just one large primary source of forage (like previous biomes) pushes too much towards “maximal-efficient invincible foragers eating this”. Now I’m really curious to see how it went in Grassland-Rainforest-Temperate Forest.
Diamond discusses the benefits of preserving different languages (most are becoming quickly extinct), such as some evidence that they can protect against Alzheimer, that they are cultural treasures, etc. but I remained unconvinced by these arguments. In fact, I absolutely love that there are many languages in the world and I hope that this continues to be the case, but I cannot find a better argument for it than a simple “I like languages” (which I think is also Diamond’s real reason and the rest is rationalization).
The “cultural treasures” part may have a point. Many forms of poetry (and witty puns) make sense only in their native language, and translations lead to dull results at best. For example, I’m absolutely not familiar with Haiku, but I suppose they must be a nightmare to translate.
And if we talk about the huge, revered rhymed poems from the past centuries, ensuring that someone will be always around to read the originals seems quite a no-brainer to me. If the whole world switched to English only, we’ll lose forever the ability to read the Divine Comedy (and many other works) taking pleasure directly from the carefully selected rhymes. I acknowledge that some of these translations are actually quite good, and especially this one by Charles Tomlinson, who did his best to preserve the original metrics, but I’m still unconvinced that a single language would be fine for this (or maybe I’m just particularly enthusiastic about strictly metrical poetry).
The legal system must above all be predictable, and you’ll have a very hard time being predictable if your legal rules are written like “citizens should do X unless they have good reasons to do otherwise”. At the very least, those good reasons to do otherwise should be elicited explicitly (but they can, and do, change over time).
Take speed limits, for example. Probably we can all agree that driving at 100 mph in city centre is a terrible idea, so the law forbids it in basically every nation (elicited exception: you’re a police officer during an high speed chase). But the actual speed limits probably won’t let you drive at 100 mph anywhere, not even if you are the only driver along a giant straight road in Nebraska with perfect weather. If you still respect the official speed limits in these conditions, probably is because you don’t want to get a ticket from the police (or because you don’t trust too much your own driving abilities, but let’s just ignore this case for now). Legal positivism applied to speed limits is John Nestor, and I would guess that even the large majority of legal officers don’t drive like Nestor, or he wouldn’t have one-third of his Wikipedia entry dedicated to this. Even in this case, I think that letting an occasional Nestor cause traffic jam is still a better idea than just writing down “drive at whatever speed you judge appropriate”.
On a more general level, I think that a simple fear to get punishment from the upper level of the bureaucracy should be enough to justify the simplest form of legal positivism. Stick to the letter of the law, and no one could formally accuse you of any infraction.
My experience is quite similar; never bought a smartphone and use just a Nokia phone with voice/SMS only. It’s not exactly like not having a phone at all, but in a typical day I don’t receive more than one message/call, and more often than not the phone remains completely silent for the whole day. I’ve watched friends managing their WhatsApp chats (and similar time-sucking services) and I’m still very much scared by the perspective of constantly being pinged for random reasons. With my old-fashioned phone I have to pay some cents for every message sent. I could change my tariff plan at any time, but so far I’ve choose to stick with the old tariff, because it’s a very strong incentive to send only messages that actually matter.
That said, I usually spend several hours a day working at my laptop and don’t travel very often. Even with a smartphone at hand, I would still prefer to work from my laptop if available (much bigger screen, no weight in my hand etc). Also, my colleagues are accustomed to emails and my parents are sort of smartphone-skepticals themselves; putting all toghether, my social pressure for getting into smartphones is pretty low. But it’s definitely possible to live without and I encourage to give it a try.
I wouldn’t fully dismiss the solitary genius as a near-impossibility. Some really weird people actually meet your strict definition, and I think they should count at least as a little bit of evidence in favor of Lone Genius.
Consider Grigori Perelman, who solved the Poincaré conjecture some years ago.
Rejected a Fields Medal
Rejected a million dollar prize from the Clay Institute.
Rejected a bunch of job offers from top US universities
Journalists are unable to meet him
So solitary that we don’t even know where he lives (supposedly Saint Petersburg)
This pretty much classify as solitary genius to me.