374 karma · joined August 9, 2011
[ my public key: https://keybase.io/iamchristopher; my proof: https://keybase.io/iamchristopher/sigs/VowNpMA9kVyTUU3frlUry9YrGe07sdkKGDcfea58EGw ]
On one hand, the self-annihilation idea implies that super-predator civilizations, being extremely rate to exist, would not have strong motivations for exterminating competitors.
What would possibly be the need, given the vastness of time and space, and the thermodynamic arguments given? Indeed, following this logic, the extraordinarily rare galactic survivor civilizations are almost guaranteed a long, peaceful existence.
On the other hand, it all develops on the curve of the galactic gaussian justifying possible pre-emptive defense investments. There's a difference of 3-4 or n sigmas statistics likelihood. In the former case, the likelihood of a rare rival emerging is high enough that pre-emptive defense strategies will be considered, and rudimentary systems of deterrence put in place. In the (high) n theory, it probably would not. How a civilization in the early stages of hegemony determined this fact, would be an interesting and possible civilization-saving (or destroying) feat....
Side note, haven't heard of the Revelation Space trilogy. Does it offer a workable theory for the origins of the predator dynamic?
Erm, this premise is factually untrue though. A lot of next generation resistive RAM devices, especially OxRAMs, have been demonstrated to be rather rad hard, making them good candidates for future space electronics platforms or.. all the other attendant apocalyptic scenarios.
Wait, does that really follow? What if you have a better than linear bootstrapping compiler. To unpack that a bit, imagine we not only have $n$ such units, but we have them wired to together in a creative way-- i dont know whether it is hierarchy, or some clever topology, but lets say that the bootstrapper now gets sub-linear scaling properties as it grows $n$.
If we look at the brain, there is a lot to be understood from the dynamics of recurrent neural fields. They are wired in a very complex way which seems to allow for some kind of very special booting (re-booting) operations. And thats just at one level of abstraction, then we re-wire them into meta-fields (like the columnar abstractions that Hawkin's builds his HTM theories around). If we have a sort of fractal information encoding, we ultimately approach shannon efficient coding. Is that what evolution has selected brains to do? And do you think it is possible the first seed AI may realize this and exploit the same strategy, just 1000x (10kx?) faster?
-The fact that you have a bunch of capsules in each layer, and each one of them is intrinsically performing a non-linear filter function (a particular squashing function shown as an image in that blog), seems like both a great asset (it looks like a 'meta-network') and also a potential problem. If you need to tune weights within these caps by the derivative of such a composite function, it doesn't seem straightforward.
-The 'routing by agreement' feature is interesting, but I dont quite get why it is superior to max pooling. If the feature is simply that it punishes weak links rather than selecting only the strongest, one interesting analogy is that it seems a bit Hebbian , and related to a concept in un-supervised learning called STDP (spike timing dependent plasticity).
If you have open-sourced this in any sense, would enjoy to check it out and/or contribute.
On the other hand, at very large scales the universe is expressly homogenous (smooth) https://www.space.com/17234-universe-fractal-large-scale-the...
W/r to larger and more wide-ranging questions, e.g. if the mutli-verse has some sort of deep fractal meta-structure, and/or the recursion of fundamental quantum computing operations, the jury is still entirely out.
>>I could go on all day; this was a chapter in my dissertation
I'd love a link (assuming it is publically available!)
To be honest I also at times thought the author's alarmism went a bit far and could have scared off serious readers, but as a general principle I still don't think its valid to critique someone on a basis that is non-existent: e.g., the author at no point explicitly endorsed any of the concepts or ideas you mentioned in the GP comment.
Where is the second half of the article where, after discussing at length many bad possible outcomes, the guy goes on to propose detailed social engineering schemes to solve the looming planetary crisis??
Either you did not read this article closely, or you are simply trolling. In my opinion, the real ending of the piece was marvelously open-ended. The author concludes by highlighting the irrational , but quintessential human, tendency to see some bright-side to the whole thing-- here are our climate scientists like Hansen and others, that know the most and see of everyone how screwed we are, still maintaining some faith we'll figure it out. And they didn't need 'communism', 'mega city' or some other nebulous concept to help them to that conclusion-- it was/is either habitual, or fantastical.
>>The other thing is that people don't really think this is a true emergency. An evidence for this is the quote "What if global warming is wrong and we made the planet better?" If it is a true emergency, we should be doing stuff that make the planet worse if it is wrong. We should be pushing nuclear power - even to the the point of reducing existing safety regulations. A Chernobyl every decade is preferable to global warming. Politically, we should be willing to trade existing environmental regulation for those which reduce CO2.
-Human perception of danger/emergency has, from the evolutionary perspective, been optimized for concrete, near-term events/entities, eg terrorism, explosions, enemies. On the other hand ,it has not prepared us for preparing against abstract, medium to long term adversaries, eg planetary or physics scale changes that threaten civilization, malevolent ETs, malevolent super-intelligences [AIs].. etc etc.
-Even if this weren't true and we didn't have these unfortunate cognitive bias, your argument about broad public wisdom of emergency relies upon a well-informed populace that is familiar with statistics, and the scientific methods. Unfortunately, that is not the case in almost every advanced Representative state on the planet. This broad ignorance renders the ambient public awarness point you have made quite moot.
-Finally, I see the logic behind the 'what if global warming is wrong and the planet gets better' case as analogical to Pascal's wager, or the false postive; what you forgot to mention is the false negative, which is metaphorically relatively similar to the outcome of Pascals (Hell/Earth becomes like Hell).
[1] https://climate.nasa.gov/solutions/adaptation-mitigation/
>>At issue is the global warming potential (GWP), a number that allows experts to compare methane with its better-known cousin, carbon dioxide. While CO2 persists in the atmosphere for centuries, or even millennia, methane warms the planet on steroids for a decade or two before decaying to CO2. In those short decades, methane warms the planet by 86 times as much as CO2, according to the Intergovernmental Panel on Climate Change. But policymakers typically ignore methane's warming potential over 20 years (GWP20) when assembling a nation's emissions inventory. Instead, they stretch out methane's warming impacts over a century, which makes the gas appear more benign than it is, experts said. The 100-year warming potential (GWP100) of methane is 34, according to the IPCC.[1]
Doesn't 'reducing' (really, normalizing) the GWP for the ~80 yrs of the 100yr period (when it is not existing) seem sort of like an accounting trick? In other words, if 20GT of CH4 were released next year from the Bering Sea into the atmosphere (this would be spectacularly bad), the overall shock to the climate/atmosphere systems is precisely the same.
[1]https://www.scientificamerican.com/article/how-bad-of-a-gree...
Re-optimizing said supply chains in an oil constrained, and eventually oil free world seems extremely non-trivial. I wonder if we will soon have to scale up the methods (both computationally, & in human/practical industrial re-organization terms) to plot new paths and then connect them, before the graph changes so much that the potential for profit disappears.
In the near term,I could see human engineers using deep learning or other computational methods for multi-scale optimization to reduce costs and carbon footprint based on locality of original sources (food/biofuels/..) In the medium term, I would bet that AI agents will exist precisely to optimize these tasks. There's hundreds of billions to be (re)-made in (re)-wiring the economy properly.
And why shouldn't we (he) be doing that? Actually the author's goal is transparent, to shake up perception, to provide fodder in the imagination of folks to ultimately change how urgently we treat this in public forums. I thought this was a quite trenchant point (from the article):
>>Over the past decades, our culture has gone apocalyptic with zombie movies and Mad Max dystopias, perhaps the collective result of displaced climate anxiety, and yet when it comes to contemplating real-world warming dangers, we suffer from an incredible failure of imagination. The reasons for that are many: the timid language of scientific probabilities, which the climatologist James Hansen once called “scientific reticence” in a paper chastising scientists for editing their own observations so conscientiously that they failed to communicate how dire the threat really was; the fact that the country is dominated by a group of technocrats who believe any problem can be solved and an opposing culture that doesn’t even see warming as a problem worth addressing;
Back to you: >>I really didn't understand the part where they said methane has 34 times the greenhouse effect as carbon over a period of time and then changes the timeframe to bring the multiplier put to 84 times.
I also thought this statement was scientifically unclear, but I think the author was trying to say that the rate of methane release increases, thus, the impact on climate systems multiplies relative to an equivalent release of CO2 in that period (100 yrs v 24 yrs). If someone else can mention exactly how he got from 34x to 84x though, id love to hear it; i didn't get it.. I do want to mention, however the 30x GHG effect number is not set in stone. In fact, depending on the rate of release, there may be 'force multipliers' depending on how (quickly) ecosystems can absorb and use these gases. Ominously, this multiplier seems to go up anyways as temps rise. [1]
[1] http://www.nature.com/nature/journal/v507/n7493/full/nature1...
In other words, all of these happening is not a probable outcome (the article is not based on a statistical inference or climate model). However, as we climb up the \delta +C ladder, all of these effects/implications become increasingly probable. If you crank up \delta +C to +5C for instance, at least half if not all of these will occur (due to positive feedbacks eg albedo);
I would not say we are beyond the hope of any action at this point. There are two sub-sets of conventional actions being discussed in the context of official orgs, eg UNFCCC at the moment:
-Mitigation- reduce emissions directly,- eg turn off coal plants or plan to do so to dozens within the next few years, massively scale up solar adoption
-Adaptation- take actions now to reduce devastation from past [inertial] and unavoidable future emissions, e.g, invest in heat-proof seeds and crops for certain tropical biomes
Broadly, both are still possible. The former is notably more susceptible to political landmines and fossil fuel monopoly blocking than the latter, so a lot of climate scientists are urging folks to focus on the latter. Is it enough to save us? In terms of physics and climate systems--no, but metaphorically it could reduce pain like palliative care for the swathes of humanity, largely from tropical regions, who probably wont make it. The real heartbreaker here: mitigation is crucial to avoiding all the evil long-tails, but our poltical and economic systems seem allergic to it. How about just a little mitigation- eg lukewarm Paris implementation-- do we still miss some of the worst evils? Very unclear-- now we need to consult the statistical physicists.
There's a third bin too which I would call pure engineering solutions; These could involve for instance, building plants to suck CO2 out of the atmosphere [1], seeding the oceans with certain chemicals to increase Co2 oceanic uptake [2], or straight up engineering the atmosphere [3]. I listed those in increasing order of possible risk/backfiring (complexity of the climate engineering operation).
I invite anyone to clarify or add to which they think are feasible and infeasible amongst these; I listed this all for context, and not to advance a particular legal, moral, or political argument, although clearly there are a litany to be made.
[1] http://www.climeworks.com/ [2] https://www.sciencedaily.com/releases/2007/06/070608142214.h... [3] https://www.wikiwand.com/en/Stratospheric_aerosol_injection_...
Its true that LQG is not known a natural schema to implement hyper-computation. However, since we don't well understand the time-dynamics of twistors ( how these operators may interact non-linearly through time), i don't see any a-priori reason why it is not a possible scheme for super-computation (you could make a good counter- response based on occam's razor, which I'd grant).
Look, lots of thereotical CS folks get disturbed by the idea of hyper-computation/super-turing machines, but in truth Turing machines are a toy model in comparison to true physics; as such, it doesn't take a whole lot more to get something more powerful; Siegelmann and colleagues have shown that real weighted, analog recurrent nets have super-turing abilities [1] [2]. While Aaronson and other raise good questions about physical realizability of such systems, a good thing to keep in mind is that these discussions often take place at computational 'limit' cases, eg solving intractable PSPACE problems, which may not be as relevant to more pedestrian problems solvable by biological systems. Central point: dynamically evolving systems iteratively exploring through (from) in-consistent systems towards more and more consistent ones have many of the same compelling qualities we would call 'super-turing'. Also see the lit on evolving turning machines.
Finally, while I don't agree with the magnitude your Pauling analogy, I certainly agree with you statement that Penrose is out of his depth here. While I do not , presently, buy his argument that quantum effects are necessary to realize consciousness, I remain open to the idea until we know more about BOTH physics and computation.
[1] http://www.mitpressjournals.org/doi/abs/10.1162/NECO_a_00263 [2] http://www.kurzweilai.net/super-turing-machine-learns-and-ev...
Penrose doesn't appeal to quantum woo, he has a quite well articulated set of assumptions and arguments , related to loop quantum gravity [1], about how objective collapse of the wave function may occur at bio-physically feasible decoherence times.
Now, whether you think his biophysical theories of consciousness are valid, or even required (whether you buy the microtubules argument/hypothesis), is one thing. But stating that Penrose- arguably the 20th century's foremost mathematical physicist- is a practicioner and/or a spreader of 'quantum woo'-- that belies a level of mis-informantion so catastrophically high that it renders your quoted statement indistinguishable from a pure ad-hominem.
I have a mild case of synesthesia and this is one of the closest things I have seen/heard (:) ) that comes close-ish to capturing how I see/feel when I experience great music.
Of course, I was not surprised to see the creator of the video list a similar motivation in his caption for the upload.
The possibility space between relatively simple and insufficiently general unsupervise/clustering approaches and rigid SGD schemes is large, and probably contains the brain's true inference engine. Personally, I am excited by some of the ideas brought forward in this Bengio paper: https://arxiv.org/pdf/1602.05179.pdf
Nevertheless, a STDP type learning rule can inspire interesting applications. One of my co-advisors authored an article [1] which shows a completely unsupervised classification on the MNIST challenge in a crossbar environment, achieving 93 \% . Nothing like state of the art CNNs etc, but considering this was done without labels, that's pretty impressive.
[1]http://www.ief.u-psud.fr/~querlioz/PDF/Querlioz_PIEEE2015.pd...
Individual mega-investors are no doubt powerful, but what about investor collectives. if a single company and/or entity can take on even 1/10th the power of Icahn collectively, it is something worth considering.
Apps like Robinhood have opened the ability for folks earning even less than 50k , to comfortably invest.
Many tens of thousands may be interested; there could be a bigger market opportunity than you think.
Random thought that just came to mind. Have coalition of debt holders (borrowers ) ever considered buying stock in these companies to influence them in the form of another company that is jointly owned or invested in by thousands or tens of thousands of folks?
Eg, mass activism campaign launches request for everyone to invest X% of what they normally would have paid into loans for the month (above the minimum thresholds of course), and/ or y% of disposable income into this fund. Now obviously one or ten people doing this will own just a few hundred stocks, but thousands doing this get tens of thousands of stocks-- thats when it could get interesting ;)
Obviously, would not be available to those really struggling under massive loan burdens and struggling to make every payment, BUT for those in the AND gate between reasonably high loan burden and comfortable income, which is a huge overlap with HN by the way, the impact could be massive.
In the end the results could be beautiful though: cannibalizing capitalism , - an inside job.
What may be MORE interesting to see, not just for HN folks but in general, would be a parameterized set of possible/reachable (as well as 'stretch') goals, for folks interested in learning more about math/physics/cs at various bands of natural/innate ability. While these bands may not be set in stone and raw effort may help level up someone up one tier above their 'home' range, it is for instance frankly impossible for at least 99% of folks to ever become competent at QFT - and likely, 95-99% cannot even grok it in part.
I don't know what her IQ really is, I don't think astronomical, but it is certainly very anomalous. I only have a truly deep perspective on my own abilities, and just based on this I think I can say somewhat confidently she is at least a generation above mine in general ability-, for myself, having jumped from standard engineering undergrad, to software engineering and algorithm design, to applied physics, and now finally to nanoelectronics , each of which I can say I somewhat mastered within the space of a year-1.5yrs, and statisically I konw I am 3.5-4 standard deviations above mean (at age 10 and 14 I tested 164 and 158 respectively), but all of that did not give me the power to so quickly grasp QFT- remained incomprehensible to me in my first year grad school classes, and probably always will, unless I invested massive time into mastering it.
But I want to comment on another thing I find highly strange, which is that someone of such rare intelligence would leave academia (ostensibly she was at one time considering staying in Physics as a grad student) back to software architecture, rather than gravitating to the edge of mathematical physics- or continuing to follow that road to reality, to riff on Penrose. Indeed, I found myself following the inverse path. And ultimately, Moffat & his correspondents in our generation,- possibly another lady, like Pasterski looks like one possible member of this cohort- will be the ones to pave the next part of it.