This is our monthly place to discuss relevant topics that have not appeared in recent posts.
I enjoyed your Adam Ford Great Filter discussion. What is your estimate of the probability of mankind (or some intelligence created by us) colonizing the galaxy?
Seems to me it should sit in the 10-90% range.
This estimate will be right 90% of the time for anything the probability of which is completely random or unknown.
Compared to most people who post on this topic it’s a controversial view. Most tend to be outside that range at one end or the other.
Is anyone willing to explain why the Doomsday Argument doesn’t make the probability very small–as it also would seem to make Robin’s EM scenario?
The Doomsday Argument is conditioned on nothing. Conditional probabilities change a lot: we know we have existed for only a short portion of the total time the universe will be able to sustain minds (if there is a limit at all), we know lineages of animals can survive for much longer than we have up to now and EMs introduce a new condition and highlight a fundamental problem with the Doomsday Argument as well. The new consition is that EMs will see a large population explosion early on, we know we haven’t had that explosion yet so that’s evidence we are among the early minds. The fundamental problem EMs highlight is that the Doomsday Argument is very sensitive to the definition of a species: EMs are not human, biologically, so the advent of EMs could be seen as the extinction of biological humans, but things change if you do consider EMs to be human and at the other end you can include homo erectus or all primates or all animals or all life on Earth and those choices will significantly alter the prediction of the Doomsday Argument. Also, there’s nothing actually physically enforcing the Doomsday Argument, just like after getting heads up 11 times in a row the probability of getting heads again is still 50%, not infinitesimally small and just like a good sports team will keep winning trophies if nothing changes, even though a “Trophy Argument” would predict something different.
The fundamental problem EMs highlight is that the Doomsday Argument is very sensitive to the definition of a species:
Right, you can find any number of reference classes. But if the rule is: use the most salient reference classes, then one of them would probably include EMs. And this is fine, because as you say, the inference is statistical rather than physical.
The probability serving the Doomsday Argument is a priori, but that will dominate (even correcting for the degree of freedom of choosing from the three reference classses) if it’s vanishingly small.
It seems to me that the main question is whether the probability is indeed vanishingly small–as opposed to merely very small. But if anything seems like a candidate for making that probability vanishingly small, it’s the likes of EMs and super-AI.
Interesting thought of “open” house…discussion
I would like to discuss fractals in immunoglobulin behaviour in vivo..
regarding the probablity of “human” colonisation of galaxy…….what was the probability of “human”ity evolution at the time of big bang? who estimates that? in the evnt of a non-zero probability, (we are actually here), the estimated probability of simultaneous evolution in the universe is also non-zero and non negative!
the estimate of human colonisation of Mars (which is also a part of the galaxy) is already happen(ing or ed)! so to calculate the probability of “intelligence” (non human) in the galaxy is closer to one (our existence is proof of our existence) …..and the probability of intelligence created by us (is already existent) out there is also closer to one (read Hubble/ voyager 1 and 2)….what is still in the realm of fantasy is that there is a co-evolved humanoid intelligence out there somewhere , struggling like us , waiting and wanting to be “discovered”
I think you should debate Nick Bostrom on the “AI foom” issue.
Does anyone know why those who write on topics such as rationality, A.I. the future etc are typically unclear, verbose and meandering? Is it just the subject matter?
Keep avoiding plain English guys, all the better to make yourselves seem deep.
Witness the latest “exchange” between Robin and Yudkowsky. The goal seems to be the say as little as possible while changing the subject matter every other sentence. Oh, and then to carp about how misunderstood you are, and to deliberately talk at cross-purposes.
Unluckily there is no apparent empirical refutation of this far-flung speculation, so it just sort of goes on for ever. Status and self-confidence decides who scores a point in the “debate”. Oh and naturally no-one changes their minds on anything despite the commitment to overcoming bias etc
Does anyone know why those who write on topics such as rationality, A.I.
the future etc are typically unclear, verbose and meandering?
Do they write this way all the time or only on these topics?
I think we’re talking about a cohort of folks who don’t understand the importance of clear expression in natural language. Part of the explanation is that engineers tend to have a low opinion of natural language. But the main explanation is that their individualistic worldview minimizes the importance of expression, the objective to be atomistically “less wrong” rather than to take part in a discussion that yields insight.
well, correct me if i am wrong, but “open” house means that specifically….it is a sort of brainstorming, where most talking heads talk for talkings sake and see if meaning emerges from the excercise. It could be “directed” or organic …depends on the precondition of the moderator
That’s part of the problem: they think they’re talking instead of writing. “Talking” in a written forum carries none of the advantages of direct conversation to offset those of precise writing.
Yudkowsky is “largely self-educated” as Wikipedia puts it and seems to like fields like philosophy, where this verbose style is very common. Maybe my memory’s working set is too small, but half-way through one of Yudkowsky’s sequence articles, I have usually forgotten the title and the hypothesis.
I find Hanson clearer and I believe his scientific background is a contributing factor. Peter Norvig is great, but concentrates on the technical side.
Challenge: Show me a philosopher who writes as imprecisely and meanderingly as EY.
[Philosophers are among the clearest academic writers. Both EY and his critics disparage philosophers, EY to pretend novelty, his critics to play the “ridicule the abstract” game.]
Zizek as an obvious example, but in fact I struggle to think of a philosopher that has a ‘plain english’ writing style.
It’s hard to adjust for what you’re used to, unfortunately. I don’t remember having the impression of ‘unclear, verbose and meandering’ writing the first time I read the sequences, though it was a couple years ago and so the memory is foggy.
There are many, many philosophers who are capable of making themselves understood without sacrificing substance.
It is your sort of half-posturing that gets no where. I was initially presented to less wrong, via a “Cult of Bayes” article, and held such prejudice for a long time. Yudkowsky’s writing style does leave something to be desired, but I got over it.
You see, it’s quite a bit to ask to have every one do *everything* at once. The people from less wrong and here got very very far on their own strength and will power.
Less wrong doesn’t make a lot of sense if you don’t consider that it is a bunch of people who tried to go further than the standard tropes of Popper/Naive scientism from say, Popper, Kuhn, and Occam’s razor and the typical traditional science tropes is already so much further than the rest of the world, but this is a narrow group of people who tried to do more, and largely succeeded.
So the cross-section of people involved is already thin. You have to take the allies you can get. There are not that many people “This side of the sanity waterline”. I am very appreciative of Hanson/EY and the measures they have done to give safe haven to allow a new generation of people to develop.
Some times, I feel a little sorry for Elizier in that he get’s all this flak. I’d take some time to sympathize for him and read his personal website about his younger brother. I keep that vividly in mind, and that reminds me of his nobility.
Your posturing and venom is largely unwarranted. It is doubtful he is trying to seem deep. There are many clear speakers in that community, you sow discontent without perspective. Such venom is only facilitated by long chains of symbolic thought over and over. If we were in meatspace, keeping a view of the human would soften your words.
Why don’t we put some losses for your aggressive signalling? Every half-assertion laced with aggressiveness you lose $dollars. You get a lot for free for no reason.
When people say “You are not overcoming bias,” they hardly measure what they actually learned from OB/LW that they would not have figured out themselves, and then chide OB/LW for the subtractive difference in what they know but OB/LW might not have covered and only measure *that*, and do not count the former.
Less wrong doesn’t make a lot of sense if you don’t consider that it is a
bunch of people who tried to go further than the standard tropes of
Popper/Naive scientism from say, Popper, Kuhn, and Occam’s razor and the
typical traditional science tropes is already so much further than the
rest of the world,
This is indeed LW’s delusional self-image.
One has to wonder whether this self-delusion isn’t advanced by imprecise writing, but I shouldn’t overemphasize this aspect.
I have to wonder what “strength and will power” you might think was exhibited by EY when he announced that he “had run out of time” and posted his free associations, when the topic was basic to his endeavors and he was given another opportunity to argue his position against a competent opponent.
[He’s like a person just going through the motions. His potential as a writer was ruined when he consented to be paid to produce a daily blog posting for a year. Thus you get the interminable and (reputedly) unintelligible morality series.]
Clarity means different virtues, namely ease of comprehension and precision/succinctness of expression. (Those who say EY is “clear” may have the first meaning in mind.)
Venom?! My, you are a sensitive soul! Another tedious, long-winded post, too. Great.
I find both of them pretty clear. Yudkowsky does sometimes indulge in flourishes which compromise brevity, but generally not clarity (for me). I have in fact been eagerly following that latest exchange, thinking “Now we’re getting somewhere!”
Of course, if their writing is clear to me and unclear to you, that may also warrant criticism. But at least it should be different criticism than if it were clear to no one.
Robin, what are your opinions on basic income/negative income tax (available to every adult citizen and replacing welfare, social security, college grants, etc…). Specifically I’m interested on your opinions on the far future aspects of such a system, for example that it could put a hard limit on income inequality if coupled to GDP/capita, would make any fixed retirement age meaningless, change labor market dynamics (minimum wage, unpaid internships, etc…) and could change a lot of strategies in politics because it would mean public assistance could no longer be used to divide voters? Do you think such an idea could be successful and if so, do you think the chances of success are highly depended on the culture of the country that implements it?
Effective charity: How many hedons per dollar?
Let’s define a (negative) hedon as the amount of pain from one papercut on your finger, or from half an hour of moderate headache, all else equal.
How many expected hedons can you affect by giving one dollar to a charity of your choice (or a random charity that is recommended as “effective” by dedicated evaluation orgs)? What’s the dollar price of one altruistic hedon?
I know this is a hard question with high uncertainties, but I’m interested in your current best estimates about the very rough ballpark (i.e. orders of magnitude).
If those are supposed to be equivalent, then that’s some very moderate headache, or a very nasty papercut…
I feel a lot of the best of your writing comes from statements such as
1) “Um, the whole point of the blood system is to overcome the squared area vs cubed volume problem. So you can cool larger things fast if you use blood vessels to move fluid that carries out heat.”
2)” Evolution says that many details will be well matched to other details, but to
predict them you must know much about the other details to which they match”
3) See your thoughts about that “It is a collection of small things, as opposed to large architectural ones”.
It would be incredible if you could give some reading lists or some thing to understand statements such as those more. I’d love to read it, hell I’d even pay. I haven’t seen if you’ve published course work, but I’ll check.
Evidently, more than a billion usernames/passwords have been compromised by a russian gang. However, apparently no one is telling the public what particular accounts (whether individuals or organizations) have been compromised.
If the news report is true, then how can any person, or any business, or even any government agency, dare to participate in any form of commerce (or even in communications, such as email) that requires the use and trusting of credit cards and/or usernames and/or passwords, etc? Don’t ALL of these potentially-compromised forms of ID and/or accounts need to be re-numbered or replaced with new, non-compromised numbers, cards, usernames, etc, immediately? Why hasn’t this event already resulted in a massive/catastrophic slowing/stoppage of commerce all over the world? And why haven’t we heard news reports that untold hundreds of billions of dollars worth of assets have already been illegally stolen, bank accounts drained, etc., all around the world?
A billion usernames and passwords may sound like a lot but I’d be surprised if this affects a significant percentage of the world population. It’s likely most of these passwords and usernames were for unimportant things and/or abandoned. Having said that I’ve never understood why Americans are so fond of using credit cards, even online (without a PIN number), when in continental Europe much safer alternatives are popular. Then again Americans seem comfortable handing their credit cards to below-minimum-wage restaurant workers so I guess it’s an offline tradition that went online. People I’ve asked about say insurance will take care of it but why risk having to do that paperwork as well as the possibility of your claim being rejected if you could just use an alternative with a PIN number?
The process doesn’t really involve an insurance claim. You just contest the charge with your credit card company, and the burden of proof is immediately shifted to the company making the charge. They have to provide evidence you approved the charge. (As I understand it, that is mandated by consumer protection laws in the US.) In all the cases I and my immediate family members have dealt with, that is the end of it. The false/incorrect charge has disappeared, end of story. It has also happened so few times that punching in a PIN or something on every credit card transaction I’ve made over the years would’ve consumed significantly more of my life than contesting the one false charge I’ve ever had.
If it realy is that simple how have all the credit card companies not gone out of business due to fraudulent claims?
P.S. isn’t this an existential risk problem? Even if the average cost of fraud is low any lone individual could still go bankrupt if they are unlucky enough because it’s not like credit card users are a collective that help each other our in such situations. A PIN number provides a very cheap (it’s just 4 characters long and you don’t have to type the card number for European bank cards so you might actually have to perform less effort than with an American credit card) and effective insurance against such existential risk.
In the event that the charge is not fraudulent, the company that placed the charge provides evidence of the fact to the credit card processor (hand waving over the exact legal entities involved, because I’m not positive), they evaluate it, and if the charge is legitimate, it goes back on (probably having accrued interest as though it were never gone). I don’t know anyone who has illegitimately contested a charge, so I’m fuzzier on that process.
Regardless, the credit card companies employ fraud detection measure going in both directions. I’m sure that if you contest enough charges without good reason the credit card company can manage to stop doing business with you. You probably get some sort of black mark on your credit history, too.
I’m unclear on how you can only have to provide a 4 character PIN to complete a transaction. I assume there are more than 10,000 bank card holders in Europe.
Still seems very fuzzy to me and why take the existential risk that you’re the one in a million who gets ripped off for $100.000 and gets denied a refund (or has to wait a long time for that refund) by the credit card company. Plus those credit card company anti-fraud measures must take up resources too.
In (continental) Europe you can pay offline with just your PIN number (the card reader in the store sends the card’s identity to the bank which matches it with the PIN number), online you will usually need to provide an additional randomly generated checksum (generated by a personal card reader you can get at your bank). Fraud is pretty much unheard off except when criminals physically crack an ATM (and make it look like it’s untampered with) or when you have malware on your computer). Stealing a bank card doesn’t get you anywhere and neither does hacking into the database of webshops.
Credit card companies do lose considerable amounts due to fraud (as well as to unpaid credit card debt). Their profitability depends on usurious interest rates.
I wouldn’t be surprised if such rates are prohibited in Europe, necessitating more efficient methods.
Also the US credit rating system heavily incentivises credit cards over alternatives that don’t involve borrowing.
Someone at a credit card company once told me they break even on people who pay the full bill every month, make 300% of their profit from people who rack up interest, and manage to collect 1/3 of that. Not sure how much fraud is.
How would human interaction be different if we had evolved with access to reliable random number generators?
One possibility is that we could reach mixed equilibria more easily. Would colluding competitors say “this is the probability I raise prices tomorrow”? Would wise advisors instruct us to “lie with probability p”?
Now that we all have smart phones which can produce randomness, should we act differently?
Robin, I’d like to know how you would answer this question.
I agree that the word “rationality” tends to connote fixing biases more, while “intelligence” tends to connote an overall capacity. An analogy in probability might be “calibration” vs. “accuracy”. But I don’t see that much point in discussing what words really mean, as it should be enough for each author to say what they mean if there is an ambiguity.
I gather that you’re concerned that there may be two different phenomena.
If so, I agree with you. Rationality, strictly defined, means avoiding having contradictory beliefs. So, to have rationality, there must be propositional content–which can be attributed only to language-using entities. I’d deny it makes sense, except metaphorically, to say my dog acted rationally or irrationally.
What do you think the most important considerations are (or should be) when having to decide between having a job done by a larger group of narrow “specialists” (committee, development team or a full blown bureaucracy) as opposed by a smaller group of more broad specialists (such as theoretical researchers or detectives)? Do you think there are obvious places where we can gain a lot by making that choice differently than we have done so far?
The more that tasks A & B require coordination, the more it makes sense to have the same person do both tasks. The more that tasks will change the less value there is in learning particular tasks, and the more it makes sense to do many related tasks so one can more flexibly adapt as tasks change.
Robin, are you in favor of Land Value Tax? I’ve been getting the impression that it promotes economic efficiency to an absurd degree. (By forcing as much work to happen in as small of a space of land as possible.) But I am not an economist, so there could be something I am missing.
It is good to tax things with a low elasticity of supply or demand. The unimproved value of land counts, but there isn’t enough value there to cover government spending. So you’ll have to tax something else too.
I wasn’t mostly thinking of it as a revenue source, but as a pressure towards efficiency. As a thought experiment, suppose the revenues collected from LVT were dumped in the ocean, and all other taxes remain the same. I have the impression that this would cause businesses to cluster into more densely concentrated developments, which would make the economy more efficient. Those that remained sprawled over more square feet of high value land would be at a disadvantage compared to those that used vertical construction, miniaturization, automation, and so on to use the smallest area possible.
I think you are mistaken.
What am I missing?
Since no qualified folks have accepted your challenge, I’ll try. You write, “As a thought experiment, suppose the revenues collected from LVT were
dumped in the ocean, and all other taxes remain the same. I have the
impression that this would cause businesses to cluster into more densely
concentrated developments, which would make the economy more efficient.”
Clustering and efficiency, yes. But it’s doesn’t seem realistic that the degree of increased efficiency will suffice to offset the losses from outright value destruction.
LTV may not be a panacea, but it does seem an unusually efficient and fair tax. So why don’t we hear neoliberal, conservative, or libertarian economists advocating it. (Milton Friedman appears to have been an exception.)
[The cynical answer is the coalition politics played by powerful landowners.]
Well to be fair a lot of countries already tax land rents to some degree, directly or indirectly.
What if all land was leased from the state by the highest bidder (from whom others could lease the land)?
I think that would work. Rent on land is essentially a tax.
That is empirically and theoretically inaccurate Robin. Ground rent is between 20-40% of GDP in every country, and according to Stiglitz’s Henry George Theorem, it is almost always enough to finance government spending.
I know the George Theorem. Have a cite on rent stats?
For some pretty conservative estimates, go to page 19 in Professor Fred Foldvary’s paper: http://www.foldvary.net/works/policystudy.pdf
Mason Gaffney estimates that the potential land rent in the USA is at least $5.3 trillion annually. He has several papers estimating national land values. He applies an additional level of analysis called ATCOR, which not all economists accept.
The Federal Reserve meticulously estimated that the total non-edifice ground value of greater NYC was $28.2 trillion. That is greater than other estimates of nation-wide land value, which rely partly on self-reported land value that people underestimate for tax depreciation purposes (cannot depreciate land).
Foldvary estimates land value tax to 20% of income which he says would be enough if we didn’t have transfer payments. But of course we do have transfer payments, so as I said we’d need other taxes too.
His estimates are very conservative and he does not apply the ATCOR principle, the general principle that ground-rent increases as other taxes are removed, can hardly be questioned, since value of land is precisely determined its opportunities of use. However, even 20% isn’t far off from covering all public expenses. I believe Foldvary also points out that the need for transfer payments would decease dramatically in a system that distributed incomes justly and did not allow the dual robbery of rent and tax appropriations.
Total resource rents of Australia.
I was just listening to Tetlock’s book. He says models make better forecasts than experts. So I was wondering…are we making any progress in automating away the experts?
probably time to remove Roberts from the blogroll. all traces of his site are gone
http://blog.sethroberts.net is still up, with the last post from a week ago. (though it will be updated much less frequently, for obvious reasons)
… be a charity angel.