To support George follow and subscribe on twitch.tv/georgehotz and help George get Twitch Partner! | Programming archive🡿 ru-vid.com/group/PLzFUMGbVxlQs5s-LNAyKgcq5SL28ZLLKC
Today I learned I'm not the only one that clicks out of every website that asks me any questions or has any pop up ads whatsoever. And I'm not the only one who's irritated by the gluttony of terrible information. Today this weird channel makes me feel like a normal human.
Have you considered if the 100 trillion combined teraFLOPS are actually fragmented into smaller groups of Humans that are willing to work with one another? With that in mind you could make the argument that the computing power capacity should be measured after the largest population of Humans that collaborates with other humans. Now, I have no idea how you'd measure that but my point is that the line (3:05:22) between Humans and Silicon (I'm capitalizing it just in case...) will probably be crossing much sooner than 2038.
There is 25 billions PC connected to the internet. If one PC will emulate 100 neurons we get neural net of 2500 billions neurons which is 100 times more than in our neocortex. So it seems we have a substrate right now which is ready for superAI, we only need software. It would be potentially most intelligent entity on the earth, far beyond any human.
2:31:42 , Embryo selection for intelligence. Don’t believe breeding intelligence would work. Since standardised tests are somewhat co-related we could assume the people who graduated from these colleges were smart. But we look at history a bulk of intellectuals have been behind a lot of horrors. Communists were intellectuals, all of the german intellectuals supported the nazis. Same goes for the French Revolution. This goes to the argument by Michael malice ‘ it is easier to train a smart dog than a dumb dog’ and that a lot of smart people often fall prey to propaganda. Also Some form of these ideas were tried by eugenists and didn’t lead to any place good instead were used to segregate minorities.
Awesome stream as usual, but you seem to have missed Taleb's main criticism of IQ: the correlation is non-linear for any metric you care about (wealth, say); for low IQ scores, IQ is very predictive of wealth. For high IQ, the correlation decreases significantly. Hence IQ is a predictor of non-intelligence, but the right tail of intelligence is not captured by the test. Also, I agree I find Taleb's ad hominem and moralizing annoying, but it doesn't discredit the rest of his arguments, which should be analyzed in their own right.
I don't know man, his article sounded like BS with extra steps. These outliers in IQ seems to be a technical problem, and Geohot is right, we need a revamped IQ test for the 21st century, with better assessment to handle Taleb's outliers.
Taleb is a cocky shill. People think they´re high IQ when they are reading "Taleb" . Thats why people still following him and other shills like Harari. No real value at all.
2:50:10 "The lizard man lives within all of us." Simple. Powerful. Useful. Truth. One of the first lessons we should learn as humans, and one we need to be reminded of throughout life. If we combine it with the lesson that we have a responsibility to ourselves and to humanity to learn how to control that lizard man, the question at 2:52:05 "Who's going to save the world?" becomes easier to answer. I love your passion for comma ai, and I know it's good to stay focused, but I hope to see comma bio (the idea from a previous video ) and comma edu come true. They are worth pursuing. (And if Elon can successfully work on several life changing projects at the same time, so can others, including you. 😁) ❤️
Man, what a stream! The topics. Profound speak, you made my day. That's a tough discussion about the singularity, surely we can't be doomed, we adapt and overcome, but a species 2.0, megring humans with machines could be the only solution in this dystopia. We just need more genius minds asking genuine questions and involving in endless discussions, and then doing shit together, I don't think, that beating fake Pinata's is gonna get us through this, but that will continue nonetheless. More open minds, more forward thinking minds, that's what we need.
He is too high for the conversation. He can show his boiling mind like this, but a conversation would be too boring for him, only one topic for a long time and here he just switches topics every 20 minutes.
Computable number are merely subcountable and cannot computably be put into a 1:1 correspondence with the natural numbers. This is just like how you can't know upfront whether a program will halt. As such, the Computable numbers aren't "given" just like the natural numbers are, instead they are elusive as a set. The bigger set of Definable numbers is still subcountable, but Definability itself isn't really easy to formally capture without stepping into a metalogic. So it's hard to get at ease with any useful class of numbers that includes common Computable examples. In fact it's not so clear from what you say what your ontological point of view of Definable Uncomputable numbers is. For example, the real number starting with 0 and with n'th binary digits defined as either 1 or 0, depending on whether the n'th Turing machine/compiling program (w.r.t. a chosen language) does halt. This number is definable (I just did), but uncomputable. What's your stance on its "existence"?
Lovin the skepticism, critical thinking and philosophy that guides your opinions. I would like to be more skeptical about things as you are, its just more difficult and takes more effort, add to that, that there are so many potential things to be skeptical about, I strive for a healthy amount of skepticism.
Guys I'm looking for the stream/moment where George says "Don't do politics! Do cocaine before you do politics! It's better for you." Can anyone point me to it? I might make it my ringtone or maybe I'll make a geohot soundboard. inb4 personal army I'm just hoping I get lucky here.
Promoting websites based on uniqueness would be terrible, we’d end up websites that look like a random patchwork as people continue to manipulate SEO results, and you’d never know where to look or click on shit. Worse than it already is. And longer websites are already terrible that’s why ever time I go looking for a recipe I have to read through some ducking story like “it’s was a warm summer night as I took my sabbatical in a remote island of Greece when I first discovered homemade tzatziki.” I’m screaming just giving me the ducking recipe!
RU-vid isnt the best place for this, but it will have to do. Another user had linked this meta-analysis ( www.ijph.in/article.asp?issn=0019-557X;year=2020;volume=64;issue=6;spage=192;epage=200;aulast=Aggarwal#ref19) that analyzed studies on masks; there were a couple more meta-analyses linked too, but I only had time to go through one. It could be an interesting stream dissecting this meta-analysis and the others in further detail. I numbered all studies in the above linked meta-analysis in order top to bottom summarizing the results and limitations (see table 1 for the study's own summary). I also added my conclusion to each which resulted in considering 1/9 studies and ignoring the others due to reasons mentioned below (ranging from low adherence to ineffective control groups to low power). While the conclusion of the meta-analysis states no effect of masks were seen; it needs to be put into context. 8/9 of these studies are flawed by there own admission. Low adherence (mask-use compliance) was the main flaw among these. Simply put, a mask that is not used it obviously not effective, so I fail to see the significance of putting any weight to these studies. The one study that had the least concerning limitations was #2. It concludes that no cumulative effect from any intervention arm (mask, mask/hand), but there was a significant reduction in rate of ILI 3-6 weeks in the mask/hand arm. However this was not seen in the mask only group, so it's possible this effect was from hand washing alone. From this, we could conclude that hand washing is good and masks are at a minimum neutral. All I think this really says is that we need more research and anyone claiming either side of the argument should not be so set in their side. META-META-ANALYSIS "Hand" refers to hand hygiene like washing with soap or using sanitizer "Mask" refers to triple layered medical/surgical masks (only #7 used a mask similar to N95 in addition to a surgical mask) Feel free to point any flaws/biases #1 - Result: masks/hand showed no cumulative effect, however showed significant reduction in 4-6 weeks - Limitations: - no data on adherence - 25% of cases not laboratory tested - My Conclusion: Ignore due to no adherence data. If we are uncertain how often and how well masks were used, we cant conclude effectiveness. Curious about the short term reduction vs no cumulative reduction and it's relation to compliance #2 - Result: masks/hand washing showed no cumulative effect, however showed significant reduction in 3-6 weeks. - Limitations: - masks only worn in residence hall and transmission may have occurred outside - protective effect may have been from hand hygiene as it was not significant for mask use alone - My Conclusion: Consider because of good adherence (this study is by the same group as #1, it seems they improved on the limitation of the first). At a minimum, hand hygiene is effective; I would label masks as beneficial in the short term and neutral in the long term according to this study. #3 - Result: no significant difference between mask and control - Limitations: - low power (38%) due to trial ending early - My Conclusion: Ignore due to low power. Study itself states they cant conclude th effectiveness of masks due to this ("As a consequence of the lack of power, the lower bound of the 95% confidence interval of the multivariate adjusted odds-ratio was 0.44 meaning that we cannot formally exclude that our trial could have missed a substantial face masks effectiveness; i.e. a relative reduction of the ILI attack rate of up to 56%."). I thought the study design was particularly relevant since it focuses on source control (only required index cases to wear masks); it's unfortunate it has low power. #4 - Result: no significant difference between mask and control - Limitations: - low adherence (45% in index cases and 21% in contacts) - My Conclusion: Ignore due to low adherence. A mask that is only used 45% of the time by subjects with an infection isn't useful #5 - Result: no significant difference in three arms (hand, mask/hand, control) and significant difference if masks and hand washing used within 36h of illness onset in index case - Limitations: - time lag between symptom onset and variable adherence (use of hand washing or mask/washing) may have mitigated intervention (hand washing or mask/washing) effectiveness - some subjects in control and hand only also used masks (unstated compliance it meta study) - My Conclusion: Ignore due use of masks in the the non-mask arms #6 - Result: no significant difference in rates among three arms (hand, mask/hand, control) and significant difference in in rate of transmission among mask/hand - Limitations: - under-reporting of cases - low adherence for mask/hand arm - My Conclusion: Ignore due to under-reporting of cases and low adherence #7 - Result: no significant difference among arms (surgical mask, N95 LIKE mask, control) - Limitations: - 21% adherence in self-reported mask groups - My Conclusion: Ignore due to low adherence #8 - Result: no significant difference among arms (hand, mask/hand, control) and significant increase in odds of infections for hands/mask (ie appositive of expected result of this being protective) - Limitations: - possible exposure out of house - poor adherence to interventions (hand, mask/hand, control) - 2009 H1N1 pandemic caused increase in hand/mask use among control arm (4%-52% among index cases and 17-67% among families) - sensitization bias: intervention arms may have been more likely to report perceived symptoms (fever w/ cough or sore throat) than control arm - My Conclusion: Ignore due to low adherence and use of masks in non-mask arms #9 - Result: no significant difference in ILI rates in arms (mask/hand, control) - Limitations: - 50% adherence to mask use - My Conclusion: Ignore to to low adherence
@@WhyGoThere No where else, just reviewed w/ notes. As for the other two meta-analyses(didnt review in as much depth, might be more to each study): Second Meta-Analysis (onlinelibrary.wiley.com/doi/full/10.1111/jebm.12381) - Result: no difference between surgical mask and N95 mask for ILIs - Limitations: - Number of community studies were small possibly making this subgroup analysis unreliable (community in terms of non-hospital setting) - My Conclusion: Ignore because his study summarizes surgical mask vs N95 (not mask vs no mask). The community and hospital results showed similar outcomes, so the limitation does raise any major concern. The analysis notes compliance may have explained the little difference ("The reason for the similar effects on preventing influenza for the use of N95 respirators versus surgical masks may be related to low compliance to N95 respirators wear, which may lead to more frequent doffing compared with surgical masks") Third Meta-Analysis (www.cmaj.ca/content/188/8/567) - Result: no significant difference between N95 respirators and surgical masks for acute respiratory infections - My Conclusion: Ignore due to studies own admission ("However, wide 95% CIs from our meta-analysis must be interpreted as insufficient evidence to determine whether there is a clinically significant difference. Findings from the surrogate exposure studies suggest that N95 respirators are superior to surgical masks for filter penetration, face-seal leakage and total inward leakage under laboratory conditions.) Also notes possible effect of compliance ("N95 respirators are often considered uncomfortable for regular use, and improper wearing or adjustment of the respirator because of discomfort could lead to inadvertent face contamination, thus negating the potential protective benefit."). Finally, the conclusion ("Although N95 respirators appeared to have a protective advantage over surgical masks in laboratory settings, our meta-analysis showed that there were insufficient data to determine definitively whether N95 respirators are superior to surgical masks in protecting health care workers against transmissible acute respiratory infections in clinical settings."). Also should add that this is again surgical vs N95 (not mask vs no mask) and I didn't not the limitations since the conclusion is clear.
Brady McDonough 100% agree the poor adherence in studies could (and likely does) represent masks’ effectiveness in a real-scenario. However, I think a high compliance study would be telling - if there was then a significant difference in transmission compared to the low compliance study then we would know it’s a “people” problem, not a mask problem.
George...When you've talked about the petaflops and synapse relation, you've just massively oversimplified everything. As an ex-expert biologist, I can tell that the human brain does not simply function as neuron - the synapse - neuron. It's just hilarious to compare it like that. It's like comparing an apple with a stone.
Hypothesis 4 in the beginning: These websites don't write articles to be informative, but they write articles so they show up high in google search. Because that's where they get read, and that's how they earn money... Yeah this stuff sucks. You don't need to produce a good product, you need to produce one that the algorithm likes.
@44:50 that's fair but wouldn't you agree that with a certain amount of publicity/fame, your public identity is turning into something that may be called a "brand".
I guess my concern with using IQ tests is that they can also be gamed. The fact that, if you're wealthy enough, you can buy SAT tutoring that will teach you how to take the SAT and nothing of actual importance. I suppose you need a base level of intelligence to get a certain score, but the score itself could pretty easily be inflated. But you're also right that "holistic" admissions just lead to nepotism or selection for ideology, or whatever other BS quality admissions people are looking for. I don't really have a good solution to this. It's a pretty central Goodhart problem.
@@geohotarchive when searching for something specific, you search specific but broad enough. most of the time if it requires some explanation of an experience, just add "reddit" at the end. basically, just use websites and it will search the website for info relevant to the search. Most of the things he complains about can be fixed just that you have to know about the little tricks to optimize it for the specific thing you want, but you might not know how to search for it. but SEO probably have biases, which i havent found how to bypass to see better websites with the info i like or want when i do not have a website for that question. so that could be fixed. But most things he talks about already exist in some form that can be used to get what he wants.
it's so fascinating to me how we humans, in general, don't tolerate much. Like, when someone with very good reputation says or does something, or even has something in their past, that's doesn't align with someone else's view, that other person is going to completely change their opinion about that highly regarded individual. I'm not that way though. Especially when it's like an actor or a musician -- if they would turn out to be a molester, absolutely nothing would change for me, though I'd prefer not to know.
About Yandex image search: I paused the video in the end when George makes the image from his cam larger, took a photo with Yandex app, and it actuallly gave it following tags: 'man', 'george hotz', 'george francis hotz'. So it is pretty good. When I gave it my seplhie, it found some pictures of me online. It is also good when you encounter some species of birds or insects you don't know.
I don't really see why singularity point is a problem. Within my lifetime computers might be smarter but their purpose is created by us. Basically without emotion, robots have no reason to create their code away from our directives. The smartest bots are going to be the ones that follow market demands. (Supply and demand) Human and machine will continue to have a "mutualistic relationship". Am I wrong?
Rebuttal to 3:05:00, even though the amount of silicon compute will be greater than biocompute, the network effects of biocompute are arguably higher since the connectivity between each person is greater and more flexible, whereas silicon compute runs on software designed by humans and is thereby limited to the goals the human has. Machines don't have a survival instinct that ants, mice, cats, and humans do and that is the key difference. The real problem is if more powerful software/compute will be concentrated into fewer hands, e.g. governments, tech companies, etc. Then we'll have huge wealth redistributions that can destabilize society. It doesn't seem like its going in that direction though with cheap cloud computing and OSS.
These videos are very valuable because they show a real relatable human, not any smarter than anyone else really, but he keeps going for hours straight meaning he can do anything
Lol website's like that were what I remember in the Geocities, freeweb, anglefire days. Usually it was one person writing everything + no web standards. At some point, frames became obselete and content farms ranked higher.
2:00:33 To help you make youtube better here is a chrome plugin chrome.google.com/webstore/detail/magic-actions-for-youtube/abjcfabbhafbcdfjoecdgepllmpfceif?hl=en
38:50 THANK YOU SO MUCH FOR THIS RANT! I've had a sense of the exact same thing for a long time, but I've been too soft to call it out and treat it like the bullshit it is! Thank you for being a true inspiration and showing that it is ok to have standards around thought. Thank you so much.
If some opinions are attacked by the second they are outspoken...be aware... If "people" on the internet "arguing" with "proofs" and something like "The science says..." ... be aware. Social media is infested with a large amount of propaganda account . Try your luck with Non-mainstream opinion on : covid19, Climate change, LBGTQ or Ukraine/Russian War. Good luck
These streams and people that they attract give me hope for humanity. Every time I find out about some new cool shit or people like Ran Prieur and Yudkowsky.
Could you make a custom search engine that filters out results not suited to you by "learning" what types of pages you actually want. Every search you do, you can click "Trash" on all the results which you don't like and then use Deep Learning to eventually predict your searches? Just a thought.
I was thinking about similar idea of a paid search engine. The browser plugin idea is the first that comes to mind. I thought the main incentive for people to switch is that they don't get ads any more and have all kind of other "products" on top of what google offers: add specific websites to personal "blacklist", on the results page highlight which websites are paywalled, allow easy search through your full history of browsing, suggesting what new articles on a specific topic are popular with people with similar profile to mine etc. I think you need to offer people to collect all the data you can get your hands on (unpopular idea, I know) but due to the fact that you have subscription model - data is not a product and you can honestly allow people to delete it anytime and you can guarantee that you don't share it with 3rd parties.
The best argument against that I got from my friend: your assumption is that Google is not optimizing for similar thing at the moment, but it does. Google already tries to give you the most relevant results it can produce with all the data they have + offers you a targeted ads you most likely to click. Which I tend to agree with. As a result you can build search engine with different flavor of index. With different weights. And if users like it better than google's - they will switch. But for many people the difference will be too small to bother
The lesson is if you want to be smart like George, simply get good. Your intelligence that is perceived by others is built. IQ helps but there are many with higher IQ than George, but their perceived intelligence is not as notable. Just saying
After two hours you really went deep into Tikkun olam and the rest of the typical Messianic impulse sans its religion. It's all so tiresome. Do you think these things are hard coded in you?
Hey George if you're seeing this and you haven't found a better way yet, here's a Chrome extension that will hide your feeds: www.hidefeed.com/. In response to ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-jem1unAG4_k.html
oh wow, he just wrote off self help at masturbatory trash. Not that I particularly disagree but the action of writing off whole sections of life like that means that nervous breakdown is cooking. I'm here, 15-25 years early with my popcorn and lawnchair
@@sandworm9528 Many capitalist countries have laws against anti competitive practices and monopolies. Every issue in capitalism can be fixed with regulation aka fixed from within. Even totalitarian communism can be initiated from within capitalism. Centrally planned economies are totalitarian human regimes and therefore evil by definition. Capitalism on the other hand is really the name for a lack of a system (as it should be) with regulation bolted on (to prevent moral failings of its citizens magnifying).
Now you can go get her. Just don’t be weird and tell her how nice she is. Don’t tell her she’s pretty, she’ll think you’re being funny. Also tell her how cute her overly matching outfits are. Send me an invite to the wedding, I’ll sit in the back, like a ghost...
curation of opinions and experiences is hard. What I do is curate people and let the information come in, or I simply ask. Like if I really wanted to know this, I'd search for Iceland Expatriates and then ask people questions.
Would you be able to get by with a google wildcard term that forces the results to never list anything that has a pintrest share icon? btw every time you say "this website just looks better/you can just tell" I think you're just referring to the modernist design of the website. Could be wrong though
Is it just me? Or does anybody else think that asking if a test in Iceland hurts doesn’t make sense. You should be asking the type of test don’t to enter ice land and then ask if that specific test hurts. That is a weird query u wrote there George.
That graph of showing the top 50 visited websites is misleading. Presenting the top X of a sorted data set, says nothing about the size of the data set. There could be 1 million mildly popular websites that when added up, equal google. Or am I missing something?
