I think Eliezer is a really bright guy who sometimes has trouble expressing himself in a way that is respectful to contrary opinion. A lot of really bright people are like that.

Having just watched the whole thing, it's tough. I does seem as though Eliezer thinks that the fact that he is sure something is true makes it true. It will come as a surprise to the large fields of cosmology and physics that the many worlds hypothesis is true beyond any reasonable doubt.

Quoting Freddie: Having just watched the whole thing, it's tough. I does seem as though Eliezer thinks that the fact that he is sure something is true makes it true. It will come as a surprise to the large fields of cosmology and physics that the many worlds hypothesis is true beyond any reasonable doubt.

Eliezer is making an argument, even if he states it somewhat exuberantly. His reasoning is available for criticism. I'd say he perfectly well knows the difference between "has convinced Eliezer" and "is proved."

No, I think that's exactly wrong. Listen to the language he uses, please, particularly about the many worlds hypothesis, which (I assure you) is not considered remotely proven beyond a reasonable doubt by the physics and cosmology community.

Quoting Freddie: No, I think that's exactly wrong. Listen to the language he uses, please, particularly about the many worlds hypothesis, which (I assure you) is not considered remotely proven beyond a reasonable doubt by the physics and cosmology community.

We'll have to disagree, I thnk. (I agree that many worlds isn't considered proven, though it has some strong and interesting advocates, e.g. Sean Carroll.) I think Eliezer has far too much invested in his status as a Rational Man to believe what you say. However, I don't think he's above overstating his position as a touch of verbal legerdemain.

When we put a man on the moon, how far along were we to colonizing another planet, on a logarithmic scale? Can you then extrapolate to figure out when it will happen? No. The barriers are technological, social, and political, and the effort and resources required are so great that the technological ones may be the least important. Will we ever colonize other planets? I don't know, but I'm sure the responses from the top experts now would be a lot less optimistic now than they were right after the moon landings. I think that: so it is with AI. I can imagine, perhaps even expect, a failure state where society changes once again to one where the life of each generation is indistinguishable from the previous. I also imagine that the technological barriers are much greater than AI proponents believe. (For example, perhaps we find that a single neuron is itself so complex that it can't be reliably simulated by a supercomputer, and that those details themselves are crucial to intelligence.) Anyway, regardless of what happens, to say that it is "obvious" it will happen in the not-so-distant future is simply arrogant.

Agree completely.

Can someone explain this "many worlds" thing to me? What does it explain? What do we do with it? An explanation, as I see it, is an analogy which relates a posited underlying reality to my observerd experiences. In other words, it relates noumenon to phenomenon. My experience is completely unanimous and unambiguous that I exist in one world at a time. Eliezer argues for an underlying reality in which I exist in many worlds simultaneously. Where is the analogy which relates the one to the other? I noticed that, although he may have one, Eliezer does not provide a refutation for Aaronson's argument that what "many worlds" loses is science itself. Eliezer instead counters that the Copenhagen interpretation loses naturalism, and more specifically the idea that everything is made out of atoms and other such physical building blocks. The idea that a scientific observation produces one result seems to me much more fundamental than the idea of naturalism, so, forced to choose between then, I would much rather lose naturalism. And I don't understand Eliezer's argument that "whatever the right theory is, it has to be a many-worlds theory." Per Aaronson, we don't have a halfway decent theory. Therefore, we are in

Yes, Eliezer has almost zero social skills. He comes across as arrogant, self-righteous and dismissive of dissenting opinion. He would be a nightmare as a teacher or colleague, which probably explains why he works alone on his computer, is an auto-didact and has apparently never had a job in an ordinary work environment (i.e., with f2f interaction with other human beings).

A lot of really bright people are like that.

But most aren't.

Quoting themightypuck: Agree completely.

I see it like you and Jeff, too -- I don't mind Eliezer's affect at all. I find it entertaining, as a matter of fact. More importantly, I think the content of what he has to say is quite worthwhile. I don't care so much about "many worlds" until someone can tell me how we might ever be able to detect them, but I liked the discussion on predicting where we'll be, and when we'll get there, in the computing and AI fields.

Quoting WilliamP: [...] Anyway, regardless of what happens, to say that it is "obvious" it will happen in the not-so-distant future is simply arrogant.

Yes, but I think it's reasonable to say "highly likely, assuming no societal collapses." Put that in for "obvious," and do you still have a problem with the prediction? I must say that if I had to bet, I'd bet that way, rather than betting that it won't happen for another 5000 years.

Our complaints about the shortage of science topics were heard! Or somebody got the diavlogs mixed up and placed in the wrong slot. In any case, all the science junkies are most likely grateful for the unexpected topic today. Although I appreciate Eliezer's intelligence, knowledge and enthusiasm, I still find his rigidity exasperating. Scott kept reiterating his overall agreement with the manyworlds idea as the best current interpretation for some of the quantum physics implications while keeping a margin of uncertainty for a view of reality which is very counterintuitive and poses other conceptual problems. But Eliezer couldn't accept that, and went on to compare Scott's view to the infamous teapot! This general problem has been pointed out before in a variety of discussions. I understand that one needs to hold certain principles (or theories, or hypothesis) as true, at least temporarily, in order to study or develop ways of testing them or expanding them. But, at the same time, one has to be able to step back and have a critical, skeptical view in order to avoid dogmatism. Eliezer doesn't seem to be able to do that. I think it would be in his

Quoting Ocean: Although I appreciate Eliezer's intelligence, knowledge and enthusiasm, I still find his rigidity exasperating. Scott kept reiterating his overall agreement with the manyworlds idea as the best current interpretation for some of the quantum physics implications while keeping a margin of uncertainty for a view of reality which is very counterintuitive and poses other conceptual problems. But Eliezer couldn't accept that, and went on to compare Scott's view to the infamous teapot!

I don't understand the idea well enough to say for sure, but it seemed to me that Eliezer did have a solid case that if you reject MW, you also must logically reject {long list of things}, and Scott did not seem to have much of an answer to that. I agree that Eliezer's brusqueness can be off-putting, but if you interpret it as him being impatient with his feeling that he's thought about this long and hard, and what Scott was responding with amounted to "I just don't want to believe that, no matter what you say," his attitude is easier to excuse. Otto does make a good point about something Scott raised that Eliezer seemed not

> Yes, but I think it's reasonable to say "highly likely, assuming no societal collapses." Put that in for "obvious," and do you still have a problem with the prediction? I must say that if I had to bet, I'd bet that way, rather than betting that it won't happen for another 5000 years. Personally, I wouldn't go with anything stronger than "well maybe". It even seems to me that the rate of technological development might be leveling off on the whole. (Sorry to go all John Horgan on you ) Computers aren't even getting faster any more. Miniaturization will go on for a few cycles, but it's far from clear what to do with multiple CPU cores. Software techniques aren't much better than they were 20 years ago. Materials technology isn't changing very much. (Is it?) Most importantly, people aren't getting smarter or more creative. I don't hear about the type of work in AI that clearly is something that can be built on so that the next generation can stand on it and make something better. Instead, the difficulty of things in the field has not just been underestimated, but has been consistently, systematically, and dramatically underestimated. It is far from clear to me that

Quoting WilliamP: > Yes, but I think it's reasonable to say "highly likely, assuming no societal collapses." Put that in for "obvious," and do you still have a problem with the prediction? I must say that if I had to bet, I'd bet that way, rather than betting that it won't happen for another 5000 years.

Personally, I wouldn't go with anything stronger than "well maybe". It even seems to me that the rate of technological development might be leveling off on the whole. (Sorry to go all John Horgan on you )

Computers aren't even getting faster any more. Miniaturization will go on for a few cycles, but it's far from clear what to do with multiple CPU cores. Software techniques aren't much better than they were 20 years ago. Materials technology isn't changing very much. (Is it?)

Most importantly, people aren't getting smarter or more creative. I don't hear about the type of work in AI that clearly is something that can be built on so that the next generation can stand on it and make something better. Instead, the difficulty of things in the field has not just been underestimated, but has been consistently, systematically, and dramatically underestimated. It is far from clear to me that

Quoting bjkeefe: I don't understand the idea well enough to say for sure, but it seemed to me that Eliezer did have a solid case that if you reject MW, you also must logically reject {long list of things}, and Scott did not seem to have much of an answer to that.

Scott didn't reject manyworlds. He said that quite a few times. He said he felt it was the least ridiculous of all the available ideas (manyworlds, single world and some other they didn't talk about). Scott objected to Eliezer's statements such as "obvious". He kept trying to make the point that it isn't necessarily so obvious.

I agree that Eliezer's brusqueness can be off-putting, but if you interpret it as him being impatient with his feeling that he's thought about this long and hard, and what Scott was responding with amounted to "I just don't want to believe that, no matter what you say," his attitude is easier to excuse.

I could similarly say that Eliezer wasn't able to move from his rather rigid position of "obvious certainty" no matter what Scott said. Scott was able to validate Eliezer's position and

On the other hand... I do agree with this comment. Yes, 5000 years seems like an awfully long time to anticipate anything. Progress doesn't have to be a steady line. There can be peaks and valleys. And there's always the black swan possibility.

Quoting WilliamP:
[...]

Computers aren't even getting faster any more. Miniaturization will go on for a few cycles, but it's far from clear what to do with multiple CPU cores. Software techniques aren't much better than they were 20 years ago.

I once saw this graph, which claimed that if current rates of miniaturization were to continue, we would have higher energy densitys in our computer chips then on the surface of the sun by 2025, so its kind of hard to imagine us going much further there. We need better heat sinks . Still, I think you are being a little to pessimistic on the software side, multi-threading is fairly kickass.

Materials technology isn't changing very much. (Is it?)

I am not a scientist, but I have been following optoelectronics development pretty closely. Lets not forget it has not been that long since Shuji Nakamura's breakthroughs with GaN high power LED's. Also, we are starting to get some good theory behind droop, once we tie the ends there I think the field could possibly advance rather rapidly. Also, all sort of things will become financially feasible with nanotube stuff, if we can find a way to mass produce the stuff (That bulk

Quoting Ocean: Scott didn't reject manyworlds. He said that quite a few times. He said he felt it was the least ridiculous of all the available ideas (manyworlds, single world and some other they didn't talk about). Scott objected to Eliezer's statements such as "obvious". He kept trying to make the point that it isn't necessarily so obvious.

I sort of agree, but on the other hand, it also seemed like Scott kept saying "I am not rejecting this" while also saying "I don't necessarily accept this."

I could similarly say that Eliezer wasn't able to move from his rather rigid position of "obvious certainty" no matter what Scott said. Scott was able to validate Eliezer's position and moved one step further to place it in a larger context of inquiry. Eliezer didn't understand that and kept addressing the issue as if Scott was arguing against manyworlds.

I guess your ears work more like Scott's and mine work more like Eliezer's. I didn't really hear Scott placing anything in a "larger context of inquiry" so much as I heard him being a "Well, who can say?" sort of squish when it came right down to it. In particular, I thought

This is very interesting. I picked up on open mindedness versus rigidity while you picked up on the "Templeton Foundation cash register ringing". It is very similar to Eliezer bringing up the teapot. I don't know whether there is any other agenda behind Scott's expressed ideas during this diavlog. I took what he said at face value. From that perspective, I tend to agree more with his way of maintaining a critical, skeptical view and not take theory for granted. But, what is most important is to understand which position is more likely to advance science. Perhaps the right answer is that a little bit of both is needed: enough certainty to build a theory, and enough uncertainty to keep searching. I also would like to see Eliezer discuss other topics to see if there is a difference. However, he has quite consistently shown the same kind of attitude in other diavlogs. I think it is his personality, maturity and cognitive style. Some of that may change over time. Obviously.

Quoting Freddie: I does seem as though Eliezer thinks that the fact that he is sure something is true makes it true.

I couldn't put it better myself, although I didn't really mind it in this diavlogue. I do think he'd be better off if he could understand his own context- if there are two camps of thoughtful, informed people disagreeing in earnest, each convinced of their own belief, then it doesn't really matter which, if any, camp holds the specific thoughtful, informed person "Eliezer Yudkowsky". No matter his feelings on the issue, he is just one guy, and his taking a position doesn't settle the issue. I thought (and still think, I suppose) that Clinton would be a better president than Obama, but I believe that the democratic process is better than any single decision maker (even a hypothetical Emperor Pampl) and so the odds are that the people were right and I was wrong. The many worlds hypothesis might be obviously true to Yudkowski, but I think he'd benefit from stepping back and reminding himself he's just one among many in the debate.

Quoting Ocean: This is very interesting. I picked up on open mindedness versus rigidity while you picked up on the "Templeton Foundation cash register ringing".

Sponsorship does have its drawbacks, doesn't it? I might not have reacted in the same way had this just been a regular old diavlog. [Added] Obvs., I agree with what you go on to say -- that in general, it's much better to be open-minded than dogmatic, particularly in science.

Quoting pampl: [...] I thought (and still think, I suppose) that Clinton would be a better president than Obama, but I believe that the democratic process is better than any single decision maker (even a hypothetical Emperor Pampl) and so the odds are that the people were right and I was wrong. [...]

Would the Emperor have appointed Al Gore to be president in 2000? Or John Kerry in 2004? Or would the Emperor have felt the people made better choices then, too?

EY's blog is called less wrong. I doubt he is as dogmatic as people think he is.

Quoting themightypuck: EY's blog is called less wrong. I doubt he is as dogmatic as people think he is.

You may be right. I don't know him enough to tell one way or the other. My comments, and most likely others', are based on what he shows here. It may be helpful to him to become more aware of how he comes across.

Quoting bjkeefe: I sort of agree, but on the other hand, it also seemed like Scott kept saying "I am not rejecting this" while also saying "I don't necessarily accept this."

Right. This is where Scott explains how to think about different interpretations of quantum mechanics.

Quoting bjkeefe: In particular, I thought Eliezer made an especially telling point when he observed that Scott's refusal to accept what he was saying led to the refusal to acknowledge that human brains are made of atoms, and that Scott was implicitly suggesting that there might be "other stuff" (not made of matter) involved with the makeup of human beings. (As a side note: It was about then that I heard the sound of the Templeton Foundation cash register ringing.)

Whoa! Scott wasn't saying that. I was just saying that the original folk who thought up Copenhagen managed to reject naturalism, and I think Scott agreed with that in the diavlog, actually. Scott's got nothing to do with Templeton and neither of us get paid for this.

Quoting Ocean: Scott didn't reject manyworlds. He said that quite a few times. He said he felt it was the least ridiculous of all the available ideas (manyworlds, single world and some other they didn't talk about). Scott objected to Eliezer's statements such as "obvious". He kept trying to make the point that it isn't necessarily so obvious.

I wonder if anyone can estimate how many copies of the earth have been created over the past, oh, 4 billion years.

Quoting EliezerYudkowsky: Whoa! Scott wasn't saying that. I was just saying that the original folk who thought up Copenhagen managed to reject naturalism, and I think Scott agreed with that in the diavlog, actually.

Scott's got nothing to do with Templeton and neither of us get paid for this.

Noted. Thanks for checking in.

Quoting Simon Willard: I wonder if anyone can estimate how many copies of the earth have been created over the past, oh, 4 billion years.

Would it make a difference whether you want to estimate that for the last 4 billion years or for the last 10 minutes?

Quoting Ocean: Would it make a difference whether you want to estimate that for the last 4 billion years or for the last 10 minutes?

Not really. I'll be happy if someone can do either estimate.

Quoting bjkeefe: Would the Emperor have appointed Al Gore to be president in 2000? Or John Kerry in 2004? Or would the Emperor have felt the people made better choices then, too?

Yes and yes. I don't really follow the last question- if I was appointing people why would I also be running an election where people express their preference? If you're asking me to imagine counterfactuals where Gore won in 2000 or Kerry won in 2004, then I can do that, but those kinds of thought experiments are always exercises in projecting one's own assumptions onto theoretical situations- as a Dem who voted for both Gore and Kerry I obviously think they would have been better. To people who aren't me and aren't experiencing my reasoning and intuition, though, my conclusion should carry much less weight than the collective conclusion of the American people. Even for me, the electoral outcome is a strong piece of evidence against my own position, and if it were an issue I was less adamant about than it might change my mind. My complaint with Yudkowski is that he can sound oblivious to this fact that he's just one voice

Quoting Simon Willard: Not really. I'll be happy if someone can do either estimate.

Well, multiplication and all that. But bear in mind that the estimate will of course only be valid for the past 6000 years. ;^)

Quoting pampl: Yes and yes. I don't really follow the last question- ...

Just wondering how much you thought the will of the people always produced the best result.

Quoting Simon Willard: Not really. I'll be happy if someone can do either estimate.

Let me know when you hear from anyone. Maybe the singularity...

I think a decades horizon is a more honest prediction than a 5000 year horizon. At least with a decades horizon you are dealing with time frames we have some experience predicting over. I don't see how "5000 years" differs from "I don't have a fucking clue" in any appreciable way. It seems strange to me that one could say they believe the singularity is inevitable and then throw out a 5000 year time frame.

Quoting themightypuck: I think a decades horizon is a more honest prediction than a 5000 year horizon. At least with a decades horizon you are dealing with time frames we have some experience predicting over. I don't see how "5000 years" differs from "I don't have a fucking clue" in any appreciable way. It seems strange to me that one could say they believe the singularity is inevitable and then throw out a 5000 year time frame.

To be fair, Aaron later did say pretty much what you said, IIRC, though he did try to fancy it up with some logarithms talk.

I don't think Eliezer is as intelligent as other, kinder bloggingheads.tv commenters judge him to be. He reminds me of an adage in Spanish that roughly translates to "he had a brain so small that in it he couldn't hold a shadow of a doubt." (Now, that's unfair to Eliezer, who undoubtedly is a smart man.)

Also, why does Eliezer button his top shirt button? It looks really dorky.

In 20 years, Eliezer will announce to the public that he has invented the first artificial intelligence that marks the arrival of the singularity. On its first test run, it will answer any question you ask it by ridiculing your premise and asking where you got your question from in the first place. If you restate your point calmly using a simple analogy, it will derail the conversation by telling you that your analogy is inexact in some irrelevant detail and accuse you of being irrational. When all else fails, it will get red-faced, ask you if you really believe any of this in your heart of hearts, and accuse you of getting emotional. In 5,000 years, the singularity will have come to pass and the world will be taken over by quantum Eliezer robots running around accusing each other of making irrational statements at faster than the speed of light. The universe will end when one day, one of the robots lapses and accidentally accepts a valid point made by one of the others. The force of the realization that not every idea you have

Quoting jacksonian: Also, why does Eliezer button his top shirt button? It looks really dorky.

Finally, a core question on the substance of this diavlog!

Both MW and Copenhagen come up with the same results (afaik) so what you are really choosing between is the unintuitive mathematics of MW and the magical wavefunction collapser of Copenhagen. I believe a lot of physicists in the past preferred to black box QM but today more are willing to play around with theory. This may have something to do with experimental results in the last few decades or it may have to do with too many Ph.D.s with too much time on their hands (or some other reason). In the diavlog I could sort of grasp (I'm no physicist) the problems that Copenhagen cause a naturalist but it was much less clear the problems MW cause the scientific project. I don't see how an infinity of worlds I don't live in that is the inevitable result of a mathematics that describes the world I live in is that big a deal. Is a theory necessarily wrong because along with the things it predicts that can be experimentally verified it naturally posits things that cannot? This may boil down to world view. I'm obviously a naturalist.

Quoting vidal_olmos: I don't think Eliezer is as intelligent as other, kinder bloggingheads.tv commenters judge him to be. He reminds me of an adage in Spanish that roughly translates to "he had a brain so small that in it he couldn't hold a shadow of a doubt." (Now, that's unfair to Eliezer, who undoubtedly is a smart man.)

A good line. I'd appreciate it if you gave the original Spanish, too.

Quoting Magic Flea: In 20 years, Eliezer will announce to the public that he has invented the first artificial intelligence that marks the arrival of the singularity. On its first test run, it will answer any question you ask it by ridiculing your premise and asking where you got your question from in the first place.

If you restate your point calmly using a simple analogy, it will derail the conversation by telling you that your analogy is inexact in some irrelevant detail and accuse you of being irrational.

When all else fails, it will get red-faced, ask you if you really believe any of this in your heart of hearts, and accuse you of getting emotional.

In 5,000 years, the singularity will have come to pass and the world will be taken over by quantum Eliezer robots running around accusing each other of making irrational statements at faster than the speed of light. The universe will end when one day, one of the robots lapses and accidentally accepts a valid point made by one of the others. The force of the realization that not every idea you have

Have not read the comments so pardon me if this is redundant but did anyone else find this a bit ironic.

I took my last Semiconductor class about three years ago, but at least at that time the biggest long term problem for Moore's law was velocity saturation. Basically there is a limit to how fast electrons can move in silicon.

Quoting Craig McGillivary: I took my last Semiconductor class about three years ago, but at least at that time the biggest long term problem for Moore's law was velocity saturation. Basically there is a limit to how fast electrons can move in silicon.

A friend of mine who works in the audio hardware business once told me, "Sometimes the speed of light isn't as fast as you'd like it to be," which still blows my mind.

Quoting Magic Flea: In 20 years, Eliezer will announce to the public that he has invented the first artificial intelligence that marks the arrival of the singularity. On its first test run, it will answer any question you ask it by ridiculing your premise and asking where you got your question from in the first place.

If you restate your point calmly using a simple analogy, it will derail the conversation by telling you that your analogy is inexact in some irrelevant detail and accuse you of being irrational.

When all else fails, it will get red-faced, ask you if you really believe any of this in your heart of hearts, and accuse you of getting emotional.

In 5,000 years, the singularity will have come to pass and the world will be taken over by quantum Eliezer robots running around accusing each other of making irrational statements at faster than the speed of light. The universe will end when one day, one of the robots lapses and accidentally accepts a valid point made by one of the others. The force of the realization that not every idea you have

Quoting themightypuck: Both MW and Copenhagen come up with the same results (afaik) so what you are really choosing between is the unintuitive mathematics of MW and the magical wavefunction collapser of Copenhagen. I believe a lot of physicists in the past preferred to black box QM but today more are willing to play around with theory. This may have something to do with experimental results in the last few decades or it may have to do with too many Ph.D.s with too much time on their hands (or some other reason).

The great majority of Physics graduate students will not be working on the interpretation of QM, but desperately need to understand how to apply QM to practical problems. After all, Quantum Theory is the greatest success in the history of science (certainly if you include all of the "Standard Model"). Therefore, the Physics faculty sensibly tends to deflect attention away from these questions, even though the "measurement problem" causes consternation for every insightful student. The attitude in past decades has been "just accept Copenhagen and learn to do your computations." That's been wise counsel for the majority of

I think Yudkowsky seems a little too strident. The problem I see with developing interpretations of quantum mechanics is that they might bias ones thinking. We don't have a theory of everything. If you apporach the process of developing a theory of everthing with a certain interpretation of quantum mechanics, then if the theory of everything violates your interpretation you will be unlikely to discover it. You don't need an interpretation to predict the outcome of an experiment so you should bother to come up with one. Maybe a true interpretation could form the basis for a better predictive theory, but then you should be open minded and think about any interpretation that is consistant with the current predictive theories. Yudkowsky wants everyone to accept the many worlds interpretation, because he just can't imagine that in the future we will have a theory of everthing that contradicts many worlds. I think we should either avoid interpretataions all together or we should have as many interpretations as we can think of.

I wonder what the breakdown is these days among physicist studying QM? I suspect a significant number are working on theoretical stuff for no other reason than it is a lot cheaper than building giant supercolliders.

Quoting themightypuck: I wonder what the breakdown is these days among physicist studying QM? I suspect a significant number are working on theoretical stuff for no other reason than it is a lot cheaper than building giant supercolliders.

I think the theorist/experimentalist choice is often made on criteria other than practical considerations. The necessary personality traits seem pretty divergent.

Quoting Craig McGillivary: I think Yudkowsky seems a little too strident. The problem I see with developing interpretations of quantum mechanics is that they might bias ones thinking. We don't have a theory of everything. If you apporach the process of developing a theory of everthing with a certain interpretation of quantum mechanics, then if the theory of everything violates your interpretation you will be unlikely to discover it. You don't need an interpretation to predict the outcome of an experiment so you should bother to come up with one. Maybe a true interpretation could form the basis for a better predictive theory, but then you should be open minded and think about any interpretation that is consistant with the current predictive theories. Yudkowsky wants everyone to accept the many worlds interpretation, because he just can't imagine that in the future we will have a theory of everthing that contradicts many worlds. I think we should either avoid interpretataions all together or we should have as many interpretations as we can think of.

I'd say the problem with not developing interpretations of quantum mechanics is that that

That makes sense.

No wonder EY is interested in overcoming bias.

Quoting AemJeff: I think the theorist/experimentalist choice is often made on criteria other than practical considerations. The necessary personality traits seem pretty divergent.

Well, except that it's the set of theorist personalities that has the high variance. Experimentalists stay a bit closer to the mean.

On its first test run, it will answer any question you ask it by ridiculing your premise and asking where you got your question from in the first place.

It will also continually interrupt you and make fun of your stammer by telling you to "spit it out."

Regarding the physics discussion: A very poor performance by both men - EY's views on QM weren't properly challenged, Scott showed little knowledge of QM interpretations. EY's statements that all interpretations other than 'MWI' involve 'wave function collapse' are factually inaccurate, the Bohm Interpretation does not have any wave function collapse.

Quoting Craig McGillivary: I think Yudkowsky seems a little too strident. The problem I see with developing interpretations of quantum mechanics is that they might bias ones thinking. We don't have a theory of everything. If you apporach the process of developing a theory of everthing with a certain interpretation of quantum mechanics, then if the theory of everything violates your interpretation you will be unlikely to discover it. You don't need an interpretation to predict the outcome of an experiment so you should bother to come up with one. Maybe a true interpretation could form the basis for a better predictive theory, but then you should be open minded and think about any interpretation that is consistant with the current predictive theories. Yudkowsky wants everyone to accept the many worlds interpretation, because he just can't imagine that in the future we will have a theory of everthing that contradicts many worlds. I think we should either avoid interpretataions all together or we should have as many interpretations as we can think of.

He may seem strident, but what's wrong with that? It's like a challenge, or a dare. He believes

The stupidity of nearly all comments in this thread is breathtaking.

Quoting mjgeddes: Regarding the physics discussion: A very poor performance by both men - EY's views on QM weren't properly challenged, Scott showed little knowledge of QM interpretations.

EY's statements that all interpretations other than 'MWI' involve 'wave function collapse' are factually inaccurate, the Bohm Interpretation does not have any wave function collapse.

That's true. However, the Bohm picture is equally insane. I think it has not been developed as far as the others.

Quoting Furcas: The stupidity of nearly all comments in this thread is breathtaking.

Yours included.

Quoting claymisher: Yours included.

That one too. Idiot! Ever heard of recursion?!

Quoting claymisher: That one too. Idiot! Ever heard of recursion?!

What's recursion?

Quoting claymisher: What's recursion?

The stupidity of nearly all comments in this thread is breathtaking. [joking. just calling back to an earlier post]

Quoting Wonderment: It will also continually interrupt you and make fun of your stammer by telling you to "spit it out."

Version 2.0 will have the functionality to accuse you of "magical thinking" and then proceed to explain that there are really these many, many, many invisible universes created every nano-second, but you can't see them or touch them or have any way of knowing that they exist, but it's so obvious that they're there.

I noticed that, although he may have one, Eliezer does not provide a refutation for Aaronson's argument that what "many worlds" loses is science itself. Eliezer instead counters that the Copenhagen interpretation loses naturalism, and more specifically the idea that everything is made out of atoms and other such physical building blocks. The idea that a scientific observation produces one result seems to me much more fundamental than the idea of naturalism, so, forced to choose between then, I would much rather lose naturalism.

Philosophy is driving this. And personal preferences. Some would rather postulate multiple worlds, even though that can't be empirically verified, because it is friendlier to their naturalism. Giving up naturalism would be a horrible fate for some.

Quoting geoffrobinson: Philosophy is driving this. And personal preferences.

Some would rather postulate multiple worlds, even though that can't be empirically verified, because it is friendlier to their naturalism. Giving up naturalism would be a horrible fate for some.

I am having a hard time following (Ill watch this one later, is this point explained in the vlog?) why what theory you choose as being the best explanation for various QM laws should have any affect on whether you are a naturilist or not. I always thought of naturilism as just the belief that the universe is a closed system with a non changing set of laws that are never suspended by an outside agent. I just don't see the connection, this all seems to be a subset of the lets describe our universe set of topics to me.

I just have to say, again, that whatever the comparative strength's of Eliezer and Scott's arguments about many worlds, to say that the astrophysical and cosmological communities consider many worlds a settled question, much less an obviously settled question, would be flatly incorrect. Many, many perfectly credentialed and respected scientists reject many worlds. For the record, I feel fairly strongly that many worlds is correct, but then my opinion is worth very little on this subject.

so basically, "ignore what eliezer says, and listen to what my heart tells me he really wants to say"

yep - that nailed it. and that is exactly where EY fails at being a scientist and starts to be a preacher.

i'm just odd enough to find that whole exchange with yourself quite funny.

...but you can't see them or touch them or have any way of knowing that they exist, but it's so obvious that they're there.

You forgot to mention that we can't think about them either because our minds can't grasp their implications. Of course, that shouldn't have stopped all those "sad" and benighted so-called thinkers like Schrödinger and the rest of the pre-Eliezar physicists and mathematicians from seeing the Obvious Truth. Stupid one-worlders! Yes, technically, Eliezar is neither a mathematician nor a physicist. But why should that matter? He blogs, smirks and ridicules those who are.

Quoting Freddie: I just have to say, again, that whatever the comparative strength's of Eliezer and Scott's arguments about many worlds, to say that the astrophysical and cosmological communities consider many worlds a settled question, much less an obviously settled question, would be flatly incorrect. Many, many perfectly credentialed and respected scientists reject many worlds.

For the record, I feel fairly strongly that many worlds is correct, but then my opinion is worth very little on this subject.

Suppose you are asked to predict the outcome of the next presidential election. You offer this pearl of wisdom: "Either the people will tire of Obama's moves toward the left, and will toss him out of office, or the people will demand more of Obama's policies and will enthusiastically re-elect him." When the next election passes, you can point to your theory and say "See? the theory is consistent with experiment". MW is as if you have spelled out trillions of possible outcomes, with various probabilities, each specifying exactly how many votes go to which candidates in which states. I'm not saying Copenhagen answers anything more. Its treating the sheer number of possibilities as "real" bothers some people like me about

Quoting Simon Willard: He may seem strident, but what's wrong with that? It's like a challenge, or a dare. He believes he has a consistent and useful picture, and he challenges you to enlighten him in some way. There are plenty of other researchers who are receptive to other points of view. Science often progresses on gusts of emotion and personal conviction. The best viewpoint will ultimately win out.

That's quite close to how I react to Eliezer's tone. It's little different from trash-talking while playing basketball.

Quoting popcorn_karate: i'm just odd enough to find that whole exchange with yourself quite funny.

You're not that odd. I was LOLing.

Quoting bjkeefe: That's quite close to how I react to Eliezer's tone. It's little different from trash-talking while playing basketball.

I like that. John and George have "garage-band science". These guys should have a regular "trash-talk physics" program.

Quoting popcorn_karate: so basically, "ignore what eliezer says, and listen to what my heart tells me he really wants to say"

Yeah, something like that. Listen to everything he says, in and out of the specific context, and judge which is a more consistent interpretation. Same as you ought to with everybody else in the world. I guess you can assume that everything everybody says is always intended to be literally true. Which approach do you think will be closer to correct more of the time?

Quoting popcorn_karate: i'm just odd enough to find that whole exchange with yourself quite funny.

Then I'm pretty odd, myself. That was hilarious.

Quoting Simon Willard: Well, except that it's the set of theorist personalities that has the high variance. Experimentalists stay a bit closer to the mean.

Fair enough.

Quoting claymisher: The stupidity of nearly all comments in this thread is breathtaking.

Could the breathtaking stupidity of the comments have something to do with the breathtaking obscurity of this dialogue? I'm not referring to the speakers, who are obviously brilliant, but to the intrinsic obscurity of such topics as the "singularity" and multiple universes. Maybe I am just too stupid and earthbound, but I found this one of the least illuminating science dialogues I have ever heard on BHTV.

Quoting Francoamerican: Could the breathtaking stupidity of the comments have something to do with the breathtaking obscurity of this dialogue?

I'm not referring to the speakers, who are obviously brilliant, but to the intrinsic obscurity of such topics as the "singularity" and multiple universes. Maybe I am just too stupid and earthbound, but I found this one of the least illuminating science dialogues I have ever heard on BHTV.

I didn't actually write that. I was quoting a previous commenter while making a joke about recursion. I haven't listened to this one yet but I'm going to because I think triumphalist tech utopians are hilarious.

Feel free to elaborate. Otherwise, your comment is no different (and possibly worse.)

EY is nothing but a random generator of acquired but undigested academic material. he is not a scientist, as evidenced by
a) his abuse and misrepresentation of scientific theory and fact and
b) his religious adherence to a fringe view no REAL-work producing respectable scientist takes seriously corrections for the audience:
1. neither single world nor many worlds is technically a theory; both are INTERPRETATIONS of the same results of one theory, namely QM 2. QM is NOT a theory of reality, but a set of predictions about observations of carefully prepared systems. It is a theory of possible observables and operates on quantum NOT classical systems. 3. At least as it stands today, the theories laws of nature are not absolute rules but approximations of the time evolution of observable patterns. As such, different theories account for observations on different scales. No unified theory to date. 4. EY is obnoxiously and ignorantly misrepresenting Bell's Inequalities, which only "obviously" disproves two things--hidden variables and independent probabilities' addition. All it states is that we cannot assume that we are operating with independent standard probabilities, a problem which can be easily resolved when the entangled particles are seen as a single system. there are many more

Alright, I listened to this one. A real laugh riot. I gotta learn EY rhetorical tricks.

EY is nothing but a random generator of acquired but undigested academic material. he is not a scientist, as evidenced by
a) his abuse and misrepresentation of scientific theory and fact and
b) his religious adherence to a fringe view no REAL-work producing respectable scientist takes seriously corrections for the audience:
1. neither single world nor many worlds is technically a theory; both are INTERPRETATIONS of the same results of one theory, namely QM 2. QM is NOT a theory of reality, but a set of predictions about observations of carefully prepared systems. It is a theory of possible observables and operates on quantum NOT classical systems. 3. At least as it stands today, the theories laws of nature are not absolute rules but approximations of the time evolution of observable patterns. As such, different theories account for observations on different scales. No unified theory to date. 4. EY is obnoxiously and ignorantly misrepresenting Bell's Inequalities, which only "obviously" disproves two things--hidden variables and independent probabilities' addition. All it states is that we cannot assume that we are operating with independent standard probabilities, a problem which can be easily resolved when the entangled particles are seen as a single system. there are many more problems

Quoting Simon Willard: That's true. However, the Bohm picture is equally insane. I think it has not been developed as far as the others.

Actually I believe the MWI and the Bohm Interpretation are the only two interpretations put forward so far that offer a realist picture (i.e. assume an objective reality). Hence I think its a 50-50 toss-up between these two, and I'm actually starting to lean Bohm's way. The 'pilot wave' idea was initally put foward by Schrodinger, and then developed futher by Bohm. My take on it is that reality is stratified into three levels, each ontologically real. The top-level has the wave function, a real field, which never collapses - all possible states 'exist', but only in an abstract mathematical sense. The next level has something called the 'qauntum potential', analogous to momentum in classical physics. And the bottom level has the concrete objects we know and love (the particles). The top-levels are 'pliot waves' guiding the particles on the bottom level. This an ontologically very clean picture, as well as being satisfyingly simple. But it seems it would require us to give up reductionism, since both the particle and wave levels are regarded by Bohm as being 'equally real', and

Quoting Freddie: I think Eliezer is a really bright guy who sometimes has trouble expressing himself in a way that is respectful to contrary opinion. A lot of really bright people are like that.

Eliezer always seems to me to be a good-humored, spirited debater, especially considering some people who appear regularly at bloggingheads.

Quoting Wonderment: Yes, Eliezer has almost zero social skills. He comes across as arrogant, self-righteous and dismissive of dissenting opinion. He would be a nightmare as a teacher or colleague, which probably explains why he works alone on his computer, is an auto-didact and has apparently never had a job in an ordinary work environment (i.e., with f2f interaction with other human beings).

But most aren't.

...And yet he gets invited to appear on Bloggingheads quite regularly. Perhaps your assessment of his social skills (besides being largely inconsequential) is not quite as based in some undeniable objective failing on his part as you think and more a personal reaction of yours. But then again, perhaps not...

Hilarious.

I doubt that Eliezer would be fun to go camping with, but I think Scott definitely enjoyed debating him. Smart guys like to talk to each other. They speak the same language. Take away ideas:
1) battle of competing priors
2) AI research used to be summer projects to build intelligent machines competing with billions of years of evolution.
3) trans-humanist?
4) causal story....In AI, because of the lack of genes, we would need to create our own causal stories
5) mathematics are mental ice cream

bjkeefe: the original in Spanish is "tenía la cabeza tan pero tan pequeña que no le cabía la menor duda." (In English perhaps instead of "brain" as I used in the translation, it would be better to use "mind.") Here's my beef with Eliezer's position. As it comes across, Eliezer seems absolutely committed to the following: it has been incontrovertibly proven that either there are many worlds or that it is impossible to provide a sensible interpretation of QM. This position is intransigent, dogmatic, unjustified, and bizarre.

Quoting marsbars: corrections for the audience:
1. neither single world nor many worlds is technically a theory; both are INTERPRETATIONS of the same results of one theory, namely QM

2. QM is NOT a theory of reality, but a set of predictions about observations of carefully prepared systems. It is a theory of possible observables and operates on quantum NOT classical systems.

3. At least as it stands today, the theories laws of nature are not absolute rules but approximations of the time evolution of observable patterns. As such, different theories account for observations on different scales. No unified theory to date.

4. EY is obnoxiously and ignorantly misrepresenting Bell's Inequalities, which only "obviously" disproves two things--hidden variables and independent probabilities' addition. All it states is that we cannot assume that we are operating with independent standard probabilities, a problem which can be easily resolved when the entangled particles are seen as a single system.

there are many more problems with his analysis--most of which are less nauseating than the ones already pointed out. he really does bloggingheads audience a disservice by abusing facts to support arbitrary beliefs he fanatically adheres to.

Thanks Marsbar. It is always helpful to hear from a

Quoting vidal_olmos: bjkeefe: the original in Spanish is "tenía la cabeza tan pero tan pequeña que no le cabía la menor duda." (In English perhaps instead of "brain" as I used in the translation, it would be better to use "mind.")

Thanks for that.

Here's my beef with Eliezer's position. As it comes across, Eliezer seems absolutely committed to the following: it has been incontrovertibly proven that either there are many worlds or that it is impossible to provide a sensible interpretation of QM. This position is intransigent, dogmatic, unjustified, and bizarre.

Not sure if you're addressing me with this or the board at large, but my own response to it is as follows. Eliezer does come across as more convinced than would seem merited, based only on my limited understanding of how widely the MW idea is and is not accepted. I don't know enough about the theory to say whether the logical conclusions he seems to believe must follow -- if MW is true -- are actually that incontrovertible, and/or that MW is the only (or best so far) explanation. What I did notice is that Aaron did not strike me as being able to refute any of the

I realize I'm coming to this thread a little late, so there may not be much reply to this post, but here goes. At 27:35, Eliezer challenges Scott to name one desire that would go against the best interest of our genes. How about suicidal wishes? Killing yourself obviously removed any chance for passing on your genes. I see two possible retorts to what I'm saying, so I'll address them. 1. "Suicide is the result of a pathology, and no healthy organism does it." It may be true that most suicides are done by psychologically unbalanced people, but are we to assume that no sort of "pathology" could develop in AI? No sort of virus, or something as simple as a bad response to a set of stimuli unforeseeable by the programmer? In any case, there are circumstances in which suicide is not pathological but rather a very reasonable act. I don't have specific statistics, but I've read of the very high suicide rate among slaves in plantation societies, of slaves jumping off of transport ships during the middle passage from Africa, etc. The indigenous population of the Greater Antilles was wiped out in part by

I found this diavlog very interesting and informative, and contrary to what so many seem to think of Eliezer's style, I thought he was quite civil and perfectly acceptable in his behavior. If one was under the impression that Eliezer was the interviewer and that the point of the diavlog was purely to interview Scott about his views, then I agree that Eliezer did a poor job. And if this was supposed to be a Will Wilkinson style diavlog, then someone at bhtv gravely miisunderstands Eliezer's skill set. But I don't think that was the plan for this diavlog. This seemed as if it was intended to be more of a discussion about a particular issue, in which neither was playing the role of interviewer and neither had "obligations" toward the other in that regard. Eliezer feels strongly about what he believes. He cites his reasons for his belief. He asks people who disagree with him to show him where he is wrong. He lets others talk. He gives consideration to what they say. If people disagree with him, the floor is open to them challenging him in an equally strongly

"tenía la cabeza tan pero tan pequeña que no le cabía la menor duda." (In English perhaps instead of "brain" as I used in the translation, it would be better to use "mind.")

"Narrow-minded" is perhaps the right metaphor in English. No room to squeeze in any doubt. The Spanish proverb is a bit crueler, suggesting the person is just a dogmatic pea brain. In English you can be narrow-minded but still have a high IQ.

Quoting Wonderment: In English you can be narrow-minded but still have a high IQ.

That the original maxim was in Spanish suggests this phenomenon is not unique to English speakers.

Quoting bjkeefe: Eliezer does come across as more convinced than would seem merited, based only on my limited understanding of how widely the MW idea is and is not accepted. I don't know enough about the theory to say whether the logical conclusions he seems to believe must follow -- if MW is true -- are actually that incontrovertible, and/or that MW is the only (or best so far) explanation. What I did notice is that Aaron did not strike me as being able to refute any of the logical conclusions directly.

what "logical conclusions" are there to refute? the MW idea is just a little trick where you take everything that doesn't quite fit and shove it into a "differen world" that we can't access, examine, or interact with in any way. It like cleaning your room by shoving everything into a closet. It is an unproveable and unfalsifiable "just so" story that really should not be considered science (ok thats hyperbolic - but not comletely), which is why Eliezers absolute commitment to it is bizarre. It might be true, mind you. just like an omnipotent, omniscient, hands-off god might be true. and russell's tea pot may be orbiting right now

Quoting popcorn_karate: [...]

All of this may be true. As I've said several times, I don't understand the MW idea well enough to say whether or not Eliezer's strong embrace of it is merited. All I can say is that Aaron struck me as unable to rebut Eliezer's arguments in any way that suggested a clear reason for me to think, "Ah, now there's a good reason to suspect Eliezer is being unreasonable."

I would love to see David Albert do a diavlog with Elezier. Albert is so thorough in his responses and points that I think it would be much more better for EY's style.

Quoting uncle ebeneezer: I would love to see David Albert do a diavlog with Elezier. Albert is so thorough in his responses and points that I think it would be much more better for EY's style.

These two were great together. It's a privilege to listen in on these kinds of conversations. Bless Bob, his global media juggernaut, and all who sail in her. Just recently I re-listened to both Carroll/Albert vlogs, and I had the same thought about David talking with Eliezer, although Eliezer's head just might explode. I love the way David talks (Sean, too).

Now that you mention it...heh.

I can think of no thread better than an Eliezer thread in which to post this link of goodness to Fake Steve Jobs. (h/t: Don McArthur)

Despite watching this video, I still have no idea how exactly we are trying to bring about AI. Instead of arguing about how likely this singularity is, it would have been more productive for them to explain about things like Bellman's Principle of Optimality, control polices, Q learning, cost functions, and dynamic programming in general. Or am I completely off base, and none of this has to do with anything in this diavlog?

Guess I was a little to late to this one...

Here is a article Horgan did, would of probably of brought it up sooner if I didn't have such poor information retention :/ http://www.spectrum.ieee.org/biomedi...ness-conundrum

Quoting Starwatcher162536: Here is a article Horgan did, would of probably of brought it up sooner if I didn't have such poor information retention :/

http://www.spectrum.ieee.org/biomedi...ness-conundrum

Thanks for the link. It really does seem like a pipe dream doesn't it? I tried listening to the Horgan/Yudkowski vlog today, but I'm having audio isssues.

Here is another short article I came across from one of the better science blogs I've seen. http://scienceblogs.com/builtonfacts...unction_47.php I especially like one of the comments:

And infinite intelligence - whatever that might mean - can't break the laws of physics or mathematics anyway. As I explored in the post somewhat, those laws can only be exploited so far before reaching the physical limits. It would be funny if we finally invented a near-omniscient self-improving AI only to be told "Well, looks like you pretty much figured everything else out already. Good job."

Raises a point I never considered, by the time we can finally come up with an AI that has a greater intelligence then humans, there may not be much left the AI can figure out :/

Quoting Starwatcher162536: Raises a point I never considered, by the time we can finally come up with an AI that has a greater intelligence then humans, there may not be much left the AI can figure out :/

I'm mildly optimistic that computers will eventually become capable of independent research and creativity surpassing human talent. Give it another 100 years, I say. But I can think of hundreds of problems that may be soluble but are just to difficult for us today. So I'm not worried about your "not much left" scenario. Make nuclear fusion work for power generation. Explore the vast universe of nanomaterials never yet seen. Design optimum drugs for every disease/person combo.....

Quoting Simon Willard: [...]
But I can think of hundreds of problems that may be soluble but are just to difficult for us today. So I'm not worried about your "not much left" scenario.
[...]

Its always astounded me that fields like Big Bang Nucleosynthesis and Helioseismology even exist when along other vectors we don't even know simple everyday stuff....like the optimal packaging arrangements for platonic solids for example (Much less irregular shapes). Im kind of agnostic about the whole singularity thing, but like I stated in my earlier post, I don't even know how we are trying to achieve AI, so I am willing to defer to your opinion. Are we just assuming we can just lump together like 2 billion control policies and it will automatically become intelligent, or do we have specific types of programming we think will create intelligence when it reaches a certain level of complexity?