Moore’s law ends. Technological singularity postponed indefinitely

The fabs will soon be delivering “16nm” chips.  But they are not in fact 16nm chips.  That is just marketer spin.  The wire to wire spacing, the pitch, is still 64nm, as it has been for some considerable time.  There have been substantial improvements in power consumption, and this and that, but chips have just stopped getting denser.  There are no more transistors per unit area than in previous technology generations.  They are 64nm chips, and we have been stuck at 64 nm for some time.

For a long time, social decay and dysgenesis was masked by the march of science and technology.  After World War II, the march of science pretty much stopped, but technology continued.  After 1972, march of technology stopped in many areas, and severely slowed in most areas.  Since then, one technology after another has been stopping.

Living standards in the US have been stagnant or falling since 1972.

DNA technology continues with exponential growth.  Possibly hard disk storage and fiber optic bandwidth does also.  For the moment.

If we are going to make it to the singularity, either have to have a political and cultural renewal, or else DNA technology has to make us smarter people.  Unfortunately, while DNA reading and DNA editing continues to progress, DNA writing has maxed out.  It is far from clear that we can make smarter people without a thousandfold improvement in DNA writing.

50 Responses to “Moore’s law ends. Technological singularity postponed indefinitely”

  1. Henk Poley says:

    Just to add some data. In terms of single-core computer power, “Performance Moore’s law” is now at a doubling about every 38 months, since 2006.

    http://imgur.com/a/jbu7I

    • jim says:

      Those curves are not a straight line. Look sigmoid to me. Moore’s law (doubling ever X months) is a straight line on a semi log graph.

  2. James James says:

    The article http://spectrum.ieee.org/semiconductors/devices/the-status-of-moores-law-its-complicated doesn’t show what you say it shows.

    See pictures here http://anandtech.com/show/8367/intels-14nm-technology-in-detail of actual 14nm transistors.

    • jim says:

      Which tells us:

      The tradeoff for that is that wafer costs continue to rise from generation to generation, as double patterning requires additional time and ever-finer tools that drive up the cost of production.

      So, they are getting smaller features, from masks that are the same size as ever.

      • James James says:

        What’s wrong with that? That’s like complaining about a farm which has increased production, because it’s using the same tools.

  3. […] Jim’s Blog – Moore’s Law Ends. Technological Singularity Postponed Indefinitely […]

  4. […] declining rate of technological […]

  5. […] Moore’s law ends. Technological singularity postponed indefinitely « Jim’s Blog […]

  6. […] Moore’s law ends. Technological singularity postponed indefinitely « Jim’s Blog […]

  7. Zach says:

    I will make it out there. We shall drink some wine….

  8. Zach says:

    +1 for Jim. Thanks brah.

  9. Zach says:

    (off-topic as usual)

    Ever thought about taking a Gwern approach?

    http://www.gwern.net/

    For example: Long Essays. On any given topic, say what needs being said in totality, and then move on. Edit where necessary.

    Take the liberty site as an example. Therein lies a great Natural Rights essay by you I assume. I personally love it – fantastic shit. That being said, you continually repeat your insights here. Not once or twice… but perhaps hundreds of times.

    … just sayin’ brah! I figure a blog is fine for recent events, but a good thorough ASS RAPE would be nice as well, like your epic essay (Natural Rights).

    Stamp the insight, summon the demons, make your mark, and move on.

    Jeah!

    • jim says:

      Libertarianism, anarcho capitalism, natural law, moderate realism, and the rest were figured out a very long time ago.

      We are still figuring out the dark enlightenment.

      • Zach says:

        Fair enough. If that is your response to my inquiry, then I accept you too, are not all that confident. Correct?

        I have no idea what “moderate realism” means. Nor can I “tag” any one blog piece to the subject matter given by you above. This further heightens my suggestion. sigh

        I (think) I know what the dark enlightenment is, but have failed to find a simple definition of it.

        In your own words, what is it? I’m curious as to how you define it.

        • jim says:

          Moderate realism:

          Moderate realism is the mainstream ontology of which Ayn Rand’s “existence exists” is a variant – or rather what used to be mainstream over the past couple of millennia, though it is out of fashion in today’s academia.

          Universals exist in particulars, red exists because red apples exist, and have redness in common with all other red objects. The color red is a noun, but “red” in the phrase “the red apple” is an adjective. “Three” is both a noun, referring to the universal, and an adjective, referring to what a particular group of three objects has in common with all other groups of three objects.

          Universals, such as the color red or the number three, exist, but only in that particulars exist. The mathematics of number is thus the physics of arranging piles of sea shells on the beach. Thus maths is out there in the world, accessed through our senses, not manufactured in our heads, nor existing in some other realm accessible to pure reason but not to the senses. A straight line is a collection of points which can correspond to a line of sight. It is all real, and all out there in this world, and if it is not out there, not in this world, it is not real.

          The Dark Enlightenment:

          The Dark Enlightenment is defined in opposition to the enlightenment.

          The enlightenment is based on subjecting the world to reason, that people should think for themselves rather than accept authority, which is all well and good, but reason has limited value unless you start from what is real.

          The enlightenment proposes good news beliefs, that all men are created equal and endowed with certain inalienable rights.

          The Dark enlightenment, however, starts with the bad news, that all men are not created equal, that rights are entirely alienable, and observes, among other things, that most people cannot think for themselves and do not wish to do so. The Dark enlightenment focuses on the bad news, as this is a lot more important than the good news. The Dark enlightenment proposes to subject the world to reason, that superior people should think for themselves rather than accept authority, starting with the bad news about the nature of man, which leads to bad problems, for which all solutions are unsatisfactory. However, most people should accept authority, because most people are stupid.

          But if the superior people quietly do not accept authority, then the inferior people will not either, and trouble will ensue, because inferior people will be licensed to believe stupid things. Therefore authority, official truth, has to be compatible with the senses, may not be obviously falsifiable. It does not necessarily have to be true, but has to be something that a superior person can give lip service to without being stupid, and can sincerely believe without being provably wrong, so that the superior people can accept authority, or go through the motions of accepting authority, while at the same time subjecting the world to reason and thinking for themselves.

          Thus, for example, Global Warming can never be official truth, even if true, unless we achieve an enormously better understanding of the world’s climate, because if it were official truth, official truth would then be vulnerable to a decade or a century of cold, and we would then have a conflict between the duty to believe in official truth, and the duty of the superior man to subject the world to reason and to think for himself. Natural selection, that man is adapted to the ancestral environment should be official truth, because natural selection happens and we are. That man is related to the apes should be official truth, but the obvious implication that man is related to the apes by blood, by some particular creature ancestor to them and us, a great great … great grandparent, should not be official truth, because God or aliens might have created us all ten thousand years ago, though I think that most well informed people, believing in official truth, would be inclined to believe in the likely implication of official truth.

          Kids should be taught that men and chimps are related, and that Darwin argued, therefore related by blood, and different by reason of differential natural selection, but that other people disagree. They should be taught the evidence that led Darwin to that conclusion, and that that evidence is true, but not taught the conclusion.

          This is the reverse of the current practice, in which children are taught evolution as a creation myth, but are not taught natural selection nor adaption to the environment of evolutionary adaption, taught that man evolved from the apes, but not taught that man struggled and suffered and was shaped by struggle and suffering – not taught the bad news.

  10. Zach says:

    64 is not the problem. Heat is.

    Oh, and after the simple worm is emulated, then everything will be, shortly thereafter.

    Burden the right questions to receive the right answers 🙂

    • jim says:

      The simple worm has, I think, 22 neurons, yet every effort to emulate it finds the solution is further away, not nearer.

  11. I think political and cultural renewal to be of inestimable value in its own right. All the more the case if a supposed technological singularity is a possibility.

  12. Handle says:

    Meanwhile, the left singularity doesn’t even decelerate. A race to the finish!

  13. Thales says:

    Whoa, singularity is not salvation — it is a consumption. It’s the fulfillment of a dream, but you’re not the dreamer.

    • jim says:

      Humans, being products of evolution, have their own purposes. I don’t expect that we will succeed in creating AI, and if we do, it will not have its own purposes, but the purposes of those that created it.

      What I expect from the singularity is much smarter humans, with godlike technology. Quite possibly those humans will forcibly upload all the other creatures, including less advanced humans, that they find no longer useful or convenient to have around, put them in storage, and forget to take them out of storage, though they will probably keep some of them around in zoos or nature preserves, or in ornamental gardens. The rest will be put in libraries, and sometimes taken out of the library for a spin, or perhaps a re-run of history with some of the details changed to render it more entertaining, or provide a better background for a good story.

      • Thales says:

        Sounds more like a Diamond Age, which is more likely than the Singularity, less likely than Idiocracy. Uploading is going to be a bust, but backing-up flora and fauna into a virtual Noah’s Ark could be really helpful.

  14. […] For a long time, social decay and dysgenesis was masked by the march of science and technology. Afte… […]

  15. SMERSH says:

    Yes, the singularity is a relatively silly idea and this creates a problem with certain strands of neoreaction.

    Neoreactionary singularitarians want authoritarian rule to make libertarian economics possible, to speed up technological growth and bring about the singularity.

    But if the singularity doesn’t happen, their globalist ultra-capitalist scenario ends to same way that progressivism’s global crony capitalist scenario ends; a permanent cognitive elite class ruling over a permanent, mostly African underclass on a ruined garbage earth.

    • jim says:

      None of us are proposing a global solution, nor crony capitalism.

    • spandrell says:

      I’d say the defining characteristic of any neoreactionary is that no matter the views on the economy, nobody wants a mostly African underclass. HBD, i.e. NAM exclusion is the litmus test.

    • peppermint says:

      I like the idea of a permanent cognitive elite class. Preferably chosen by a process like the Mandarin exams. Which breeds faster with plenty of downward socal mobility.

      Oh, and the cognitive elite should be responsible along with authoritative.

      No one wants an underclass that breeds faster than the overclass, that’s dysgenic.

      • jim says:

        Exams too easily gamed. Turn into a ritual. Mandarinate worked poorly for China. People seeking to pass search for flaws, people attempting to filter the best in and the worst out don’t really care about flaws.

        • Melonhead says:

          IMO, there are three factors which determine where one falls within the cognitive elite class:

          1) maximized baseline genetic IQ potential (this presupposes that you’ve inherited the potential for high IQ and includes things like good childhood nutrition and diverse learning experiences),

          2) self-control and the ability to delay gratification (lots has been written about these already)

          3) grit (i.e. dedication and perseverance – how many smart people do you know who amount to less than they could be b/c they spend too much time smoking the herb instead of slogging it out at the lab bench or workbench?)

          • jim says:

            Perhaps once upon a time, but these days the ruling elite recruit on the basis of ideological conformity, and corporations recruit on informal grounds, because any attempt to recruit the cognitive elite is racist.

  16. NoBreakSpace says:

    I’ll admit I had to look up “the singularity”, and I assume this to be “Technological Singularity” as described on Wikipedia. Oh boy… this is fail-science based on nothing. How did this gain any traction at all? Write a book, and now it’s real I guess. The whole thing is based on fallacy; Moore’s law, therefore AI.

    It reminds me of Nanotechnology trend in the 90’s. I read an article which described a device that you would simply throw grass into, and have an army of miniature robots recombine the atoms into a succulent steak. Well, here we are 20 years later… nothing. The problem being that the technology does not exist. Funny how some things resonate with the public and other are simply laughed off.

    Also, genetically modified intelligence is going to do what now? Beef up the intelligence of evolution’s greatest mistake – oh my sides…

    I do love this blog though. It is literally the only one I’ve ever bothered reading. Mostly I find myself in complete agreement with Jim.

    • Steve Johnson says:

      It reminds me of Nanotechnology trend in the 90?s. I read an article which described a device that you would simply throw grass into, and have an army of miniature robots recombine the atoms into a succulent steak.

      That machine exists now.

      It’s called a cow.

    • peppermint says:

      turning various food stock into something resembling steak actually isn’t such an outlandish idea, and it is actually happening right now. A lab in Holland came up with a synthetic hamburger, that a rich person actually bought, a while ago. It will, of course, take some time for it to compete with cattle.

  17. cimon alexander says:

    It looks like the aspies over at Less Wrong won’t have to worry about terrorist actions to take out chip fabs now:

    http://lesswrong.com/lw/apm/slowing_moores_law_why_you_might_want_to_and_how/5zqs?context=1#5zqs

  18. Ron says:

    Why would you say something so … Off like “unless DNA tech makes us smarter”? You of ALL PEOPLE should know that the problem we face has nothing to do with lack of intelligence. It’s the morals and spiritual corruption that is wrecking us.

    The enlightenment began with an idea of freedom. A concept of liberty based on the idea that men could be trusted with power. That is a very honorable view that shows compassion to each other. But this became corrupted, and now we indulge and flatter the wicked in the name of compassion. We cowardly refrain from speaking the truth and we call that cowardice “tolerance”

    Tolerance! The people who preach it the most have no concept of it at all! Those are the most intolerant beasts imaginable!

    In the name of “free speech” we allow every abomination to be discussed, and I myself am terrified of so much as emailing my political views at this point.

    We have banks “too big to fail” when what is really going on is that the banks are simply defaulting on their loans to the people as a whole.

    This is not lack of smarts, it’s a lack of morals, of heart, of decency.

    • jim says:

      Lack of social cohesion.

      Observe that the biggest criminal in the banking crisis was Angelo Mozillo, who is an affirmative action hispanic. Until recently he received the most sever slap on the wrist, but just recently another criminal has been punished more severely than he. The criminal who got the most severe sentence (two years) was Kareem Serageldin, an Egyptian, who merely pissed away half a billion dollars for the sake of a seven million dollar bonus.

      Affirmative action has created a diverse ruling elite, which, predictably, lacks social cohesion. Angelo Mozillo stole from Fannie and Freddie, causing roughly a trillion in losses for Fannie and Freddie for the sake of less than a billion in money for himself, Kareem Serageldin stole from his employer. (Which explains why Kareem Serageldin got a more severe penalty for a less serious crime)

      If our diamond traders are all orthodox Jews, they will not steal the diamonds, and if our bankers were all male white Anglo Saxon protestants, they would not engage in financial misconduct.

      Finance needs trust, trust needs that they all go to the same church, or same synagogue, and marry each others daughters. (This, of course, does not work if the bankers are progressive Jews, since they do not go to synagogue and do not marry each others daughters, hence the progressive Jew Madoff made off with gigantic amounts of money from other progressive Jews.)

  19. peppermint says:

    Chips keep getting faster, but they will never be fast enough to do the impossible.

    The idiots who believe in salvation through singularity also believe in absurdities like P=NP in an easy manner.

    Anyway, part of the impetus for neoreaction was Moldbug’s rediscovery of Carlyle. I think the real thing to notice is that we already have mobile devices that can stay active all day without charging and can pull just about any information anyone has ever written down ever from the air, and also record videos, and have a text chat or voice chat or video chat with just about anyone, anywhere.

    To be sure, as time goes on these devices will get better at doing what they already do.

    But I predict that ads will always make sure all the cores are pegged to 100% utilization and pages take tens of seconds to load.

  20. VXXC says:

    Gee.

    I guess the Kwisatz Haderach isn’t coming. Perhaps we should make do, or better with what we have. And get off the couch.

    • Steve Johnson says:

      Funny you say that because Herbert took “Kwisatz Haderach” from Hebrew where it meant “shortening of the way”.

      Yep, no shortening of the way coming.

  21. Alrenous says:

    The known limits of lithography were hit when they started marketing dual-core chips.

    There’s a few factors in this. One is diffraction. To limit diffraction, they use ultraviolet lasers, but the next step up is going toward x-rays, which is not straightforward.* Second, at the scales they’re now working at, decreasing wire sizes increases resistance.** Third and subsequently we have heat. Half the point of smaller components is to run them faster, but heat increases with the cube or so when you’re increasing speed in this regime – more heat per flop, less area to dissipate it from. Not to mention silicon’s performance degrades with temperature.

    Strangely the Moore curve was not interrupted. It seemed simple, turns out it isn’t. (The article explains some of their workarounds.) I find it refreshing when someone who’s supposed to know more than me actually does, in this case, chip fabricators. Kahng complains the chips aren’t getting twice as dense, but it seems they’re still getting twice as cheap, which is the point.

    This is an area where physics humbles theory. Not that Moore’s law won’t end at some point, only that predicting it in advance is going to be luck, not skill.


    Would have sworn I got this from Szabo, but can’t find it: https://en.wikipedia.org/wiki/Logistic_function From the inside, the first part looks exponential.

    But also, even though growth slows, the back half has half the total growth.

    * This linked article links to the trouble they’re having with the exotic plasma-based generators. http://spectrum.ieee.org/semiconductors/devices/euv-chipmaking-inches-forward Generating and detecting ‘terahertz’ waves is also really hard, which I find fascinating. T-rays being the name for the middle between infrared and microwaves.

    ** If I recall correctly, it’s because the wire’s surface runs out of free electrons and the current starts leaking into the middle.

    • jim says:

      Progress will continue, but unless we find ways to make stuff smaller, we will be on the sigmoid curve, not the exponential.

      Photolithography has hit its limit. Need a contact technology – printing or embossing.

    • jim says:

      Using free electron laser light at ten nanometers, XUV, extreme ultraviolet, you can go down to five nanometers using lithography. Currently lithography is stuck at 64 nanometers.

      10 nanometer light can still be reflected by mirrors, hence is not considered Xray, but XUV.

      The limit of lithography is 5 nanometers. We are not at physics limits for lithography yet yet.

      This is social decay, not physics, halting progress.

    • jim says:

      The known limits of lithography were hit when they started marketing dual-core chips.

      the known limits of lithography are six nanometers using 10 nanometer XUV with a free electron laser. Currently stuck at 64 nanometers. This is social decay, not physics.

      • Alrenous says:

        “its natural successor, shorter-wavelength extreme ultraviolet lithography, has been long delayed.”

        Linked: http://spectrum.ieee.org/semiconductors/devices/euv-chipmaking-inches-forward

        “That’s [2015] still well behind some sunny predictions made two years ago, when the company said it expected to boost the power to 80 W by the end of 2011.”

        So is this social decay or physical limits?

        “Falling droplets of molten tin are vaporized by a carbon dioxide laser. As the tin cools and its excited electrons relax, EUV light is emitted and then used to cast patterns on a wafer.”

        Or: free-electron lasers are not cost effective.

        From La Wik:”The lack of suitable mirrors in the extreme ultraviolet and x-ray regimes prevents the operation of a FEL oscillator; consequently, there must be suitable amplification over a single pass of the electron beam through the undulator to make the FEL worthwhile.”

        Or: this shit’s expensive, man. Hence the tin plasma. Of course, it’s also saying that tin plasma isn’t cost effective, or they would already be in production. There’s a discontinuity here in the cost-per-photon graph. Physics humbles theory in this domain.

        But, while interesting, it’s not important to my argument. Regardless of the reason, lithography stopped advancing over ten years ago, and yet Moore’s law is only starting to look shaky now.

        You can double-check the discontinuity because of this. Innovation always seeks the easiest route. And such a Moore-obeying route was found, but not in lithography.

        • jim says:

          “Falling droplets of molten tin are vaporized by a carbon dioxide laser. As the tin cools and its excited electrons relax, EUV light is emitted and then used to cast patterns on a wafer.”

          Or: free-electron lasers are not cost effective.

          Hot metal plasma as an XUV light source for photolithography is a desperately half assed idea and always was a desperately half assed idea. It is something that if it could be made to work, could only just barely be made to work while creating problems up and down the line. If free electron lasers are not cost effective, then advancing technology is not cost effective.

          They reached the point in photolithography where further advance required free electron lasers as a light source. At which point you can abandon photolithography for e-beam writing, or for a contact technology, or build free electron lasers.

          Since they did not do any of these things, still stuck at 64 nanometers.

          There are several paths open for them to go to smaller scale. They chose not to walk any of these paths. Should have attempted to walk all of them.

          But, while interesting, it’s not important to my argument. Regardless of the reason, lithography stopped advancing over ten years ago, and yet Moore’s law is only starting to look shaky now.

          Regardless of whether Moore’s law looked shaky, it was shaky the moment lithography stopped advancing, but marketing hype disguised a sigmoid curve as an exponential curve.

          The day that argon fluoride eximer lasers were deployed, they should have started working on the next generation of lasers, and they chose to not do so.

          The day they decided to not work on the next generation of lasers, and instead work on the next generation of hype, that was the day Moore’s law ended.

          From La Wik:”The lack of suitable mirrors in the extreme ultraviolet and x-ray regimes prevents the operation of a FEL oscillator; consequently, there must be suitable amplification over a single pass of the electron beam through the undulator to make the FEL worthwhile.”

          Mirrors work, just barely, down to ten nanometers. They work poorly, so one has to have large amplification in a single pass. However, a ten nanometer free electron laser would still employ mirrors, though it would involve almost the same technological challenge as a free electron laser without mirrors.

          A free electron laser without mirrors has to have enormous amplification in a single pass. A free electron laser with crappy mirrors merely has to have very large amplification in a single pass.

        • jim says:

          There are many big problems with XUV lithography, and that they chose to use hot metal plasma instead of lasers is just one of them.

          But, if not going with lasers, then not really going with photolithography.

          Time for something else, such as soft lithography or dip pen nanolithography.

Leave a Reply for jim