Rooms of algebraic theology
[Image: The supercomputer pictured above is the MareNostrum, "meaning 'our sea,'" New Scientist writes; "it is housed in a 1920s chapel at the Technical University of Catalonia in Barcelona, Spain, and built from commercially available parts." Photo by Simon Norfolk]."The supercomputers I'm showing here are powerful almost beyond human understanding," photographer Simon Norfolk explains, describing his extraordinary new images of supercomputers and their architectural settings. "They can map every molecule of the billions on a human DNA string; scrutinise at the atomic level the collision between two pieces of plutonium in an exploding bomb; or sketch the gravitational pull of every star in the galaxy upon every other star in the galaxy. These are not questions that humans could grapple with given plenty of time, a notebook and a sharp pencil."
Norfolk has also photographed computers used for "mapping and predicting global virus outbreaks" and for "simulating automotive crash tests."
[Image: "Modeling physics inside an exploding nuclear warhead." Simon Norfolk].These computers, Norfolk continues, "are omniscient and omnipresent and these are not qualities in which we find a simulacrum of ourselves – these are qualities that describe the Divine. The problem is not that these computers might one day resemble humans; it is that they already resemble gods."
[Images: Simon Norfolk. The top image is titled "Mapping the human genome." The others are the TERA-1 and the TERA-10].In almost supernaturally sterile rooms, these angelic landscapes of silicon quietly hum their way through introspective worlds of calculation: derivatives, logorithms, advanced topologies. One could, in fact, imagine a whole new series of Duino Elegies, written by a posthumous Rainer Maria Rilke, in terrified praise of these cloistered machines – machines Rilke seems to describe preemptively in his "Seventh Elegy," where the "annihilator" meets the "Angel."
Rilke writes that "the external shrinks into less and less":
- Where once an enduring house was,
now a cerebral structure crosses our path, completely
belonging to the realm of concepts, as though it still stood in the brain.
Our age has built itself vast reservoirs of power,
formless as the straining energy that it wrests from the earth.
Temples are no longer known.
In any case, because all harddrives are actually geological objects – careful rearrangements of minerals under the influence of manmade magnetic fields – then these are mathematical terrains in the most exciting sense: the surface of the earth dreaming of spherical velocities inside otherwise undetectable stellar detonations.
[Images: Two close-ups of cerebral machines. Simon Norfolk]. Finally, Giordano Bruno, following Giulio Camillo, wrote extensively about the idea of a Memory Palace, or Memory Theater. As Victoria Nelson tells us, the basic idea was that an "esoterically trained memory was a godlike vessel for encapsulating the entire universe within a single human mind." This was part of what Nelson calls a Neoplatonic "quasi-religion" that "venerated memory as an organ possessing magical and world-ordering powers." Neoplatonists believed that "the whole cosmos could be 'memorized' in a much more overt imitatio dei and by this act magically incorporated into the human organism" – or, of course, into the air-cooled circuits of a supercomputer.
So if I were forced to take issue with the existence of these machines, it would not be because of their use in modeling new nuclear warheads – as Norfolk makes clear they do – but in something far more secondary, even faintly absurd: what I'd call the lack of a supercomputer poetics, or a more imaginative role for these machines to play in our literary and even religious lives. Oracular, Delphic, radically non-secular: they are either all or none of the above.
(With sincere thanks to Simon Norfolk, who supplied all the images that appear in this post. And don't miss BLDGBLOG's later interview with Simon Norfolk, in which he discusses his war photography in much greater detail).
Comments are moderated.
If it's not spam, it will appear here shortly!
Tim - Sure. I would only point out that a poet like William Carlos Williams or George Oppen could write a mean little poem about an everyday object like a hammer, no matter how much of an idiot the hammer's owner is, or however quotidian the hammer's purposes really were. So calculating astral densities in five separate regions of space simultaneously is, in and of itself, a poetic achievement, I'd say - even if achieved via microchips and for a ho-hum, annual scientific survey. I'd stand by the claim that supercomputers can be (or always are) radically non-secular; and I would point out the claims toward Divinity, omniscience, and omnipresence on behalf of supercomputers are not mine but Simon Norfolk's.
So, fundamentally, I understand - and to some extent even agree with - your point; but I stand by my claims. Supercomputers are geological objects. They are not secular, despite who funds them and why. They (seem to) continue the Neoplatonic project of an all-encompassing (and non-secular) memory. Etc. Rilke's poem still reads as if describing supercomputers. Maybe it sounds more serious than I'd intended it. The Vatican is, after all, a fully functioning Jesuit supercomputer...
But thanks for the comment! I'll win you back next time.
beautiful images, i like your description of them
I don't know what to say, Geoff, I'm stymied. Perhaps it is all a practical joke.
I loved the photo of the computer in the chapel.
tim's comments made me think. Why is it that when you might be a bit 'profound' people take this as pomposity. MUST one be always ironic and slightly manic?
At the risk of remaining agnostic about the question of "purple prose" : there's a wonderful article called "Mountains of Pi" that Richard Preston wrote back in 1992, about the Chudnovsky brothers, Ukrainian-born mathematicians who share their Manhattan apartment with a supercomputer named "m zero" that they designed and built themselves. Reading the article the first time, I did get a distinct impression that the two brothers were a bit like acolytes to the god they had created. They are continually adding to it, tending it, replacing its parts with FedExed components. They don't dare shut it down -- if they do, it might die. As if to drive the metaphor home, David Chudnovsky is afflicted with myasthenia gravis, a crippling autoimmune disorder of the muscles, and due to breathing problems caused by the disease, he has a great deal of trouble leaving the apartment. The hermetically sealed, air-conditioned space that allows the computer to live is also what allows him to live (any comparison with the relationship between temple functionaries and their gods notwithstanding).
And what do they do with their machine, this superhuman intelligence that lives in their apartment with them? They set it to the most godlike task possible : computing pi.
Anyway, it's a good article. You can read it here.
The name of this Supercomputer MareNostrum 'our sea' conjurs images of Solaris by Stanislav Lem . The great alien unknowable intelligence which is a huge sea on another planet. So both the name and location of this Suercomputer suggest a reading of 'huge mind' whether fabrication or not.
As others have pointed out, even these super high performance machines - orders of magnitude more powerful than our desktop beasties though they may be - are only mind-less instruction set pushers.
Ironically, the absence of thought makes the machines useful for the elaborate tasks they perform.
I think what Simon Norfolk is (perhaps unconsciously) celebrating isn't actual god-like machines but the apparent apotheosis – via the machines - of various Western ideas of what constitutes admirable cognition. We tend to value a mythologized form of rationality. If performing advanced mathematics, for example, is an indication of a 'superior' mind, then a machine that can flow through exquisitely difficult math like a dolphin through deep water must be a 'superior' entity. Hopefully you can see where the fallacy begins.
Of course, I'm not arguing against actual rational thought, merely pointing out how our ideas of what makes one mind sharper than another can lead a person to conclude that a thought-free machine represents the pinnacle of mental achievement.
Although I have some amount of sympathy for Mr. Norfolk's project, I feel his enthusiasm for the idea of current machines as thinkers demonstrates a poverty of imagination and, perhaps, a lack of knowledge about how real computing machines work.
They are magnificent tools; it would be petty and strange to deny that. But it's important to keep one's feet on the ground about present conditions.
About the future however, we can't know. If a merging of machine and human – such as that imagined in “Ghost in the Shell” and other polished pop-culture artifacts – ever comes even partially to pass our descendants may have to reconsider these criticisms.
Thank you Geoff - but thank you very much Tim! You said exactly what I was thinking when reading the text. I love the pictures - excellent. Machines like this makes my heart skip a beat or two. But, Tim got it dead right - these machines are basic, simple and not autonomous. Without humans, they are meaningless. Compared to the human brain, they are powerless. My life revolves around computers, so don't get me wrong. This machine's majesty is a reflection of its maker, not the other way around. Great pictures!!
Definitely fascinating photos. Definitely interesting point about harddrives as geological objects. Yet I still need to be convinced that supercomputing power is the equivalent of superimaginative power.
Could these non-thinking computers, without desire and with vastly spacious mechano-cognitive abilities, be more like Zen masters than Jesuit saints?
Norfolk doesn't seem to realise that DNA mapping, scrutinizing collisions at the atomic level, etc. are all tasks which are essentially very simple equations calculated on very large sets of data.
These, and every program humans have devised so far, are all programs that can be run on any simple off-the-shelf computer, as long as said computer is Turing-complete (a Turing-complete computer can do any well-defined calculation, no matter how complex, as long as it has enough time and a large enough memory bank to hold the instructions). The only advantage of these giant supercomputers is that they accomplish these tasks much more quickly.
And in direct contradiction to Norfolk, these plainly ARE questions that humans could "grapple with given plenty of time, a notebook and a sharp pencil," because if we couldn't do so, there would be no way for us to write the program to tell the supercomputer how to do it either. The key is that we would take so long to do it, and be so bored in the process, that it's much better to have a specialized machine to do it for us.
Fundamentally, the reason computer instructions must be so banal in their attention to detail is that we humans are rank amateurs at developing machine-learning algorithms that can help computers to operate more dynamically, in a way that more closely resembles a sentient being.
As someone else pointed out, the god analogies fail precisely because of how simpleminded our concept of intelligence is. Many in the field of machine intelligence have abandoned the idea of deriving artificial intelligence from pure calculation, in favor of neural networks. If we ever build a machine that becomes self-aware, it seems unlikely to me that its design would resemble the primitive, glorified calculators of today.
Full disclosure: I am a computer engineer.
left it a little late to comment maybe, but i felt i should get my tuppence worth in.
firstly, its a great article. the idea that supercomputers may be closer to god is very intriguing, the islamic idea of God being found through mathmatics is an interesting concept and good to think about in relation to this. but then i'm a sucker for rationalisations of god, the ultimate of irrational ideas.
secondly the comments here, though most likely true, are evidence that none of these men have souls or any imagination.
it is an intersting concept, computers having godlike abilities, the idea that man can make somethign cleverer than himself. i have never had much time for the idea that mankind's sentience is something unique and special. in that there is nothign biological you can point to that shows mankinds sentience and something that animals lack beyond the simple brain size to body size ratio.
it is a slightly primitive view that mankind must be unique in its cgnitive power, that we are the only animals capable of self conscious thought. it is also very egocentric, we know what we are, therefore we are sentient, animals and computers don't, therefore they're not. i would wager that in a few years we make the discovery that many other animals are in fact capable of sentient thought, we just don't recognise it. if you looked at humanity from the outside youd be hard pressed to find evidence of any intelligence beyond that which keeps us alive. sure we do funny cultural things but so do many other animals.
anyway. basically i like the article becasue of its ideas, and dislkie intensely the comments because they are so unimaginiative and weirdly hostile. don't think Mr manaugh is saying that he belives that computers were like god and should be worshipped. yes they are a tool and rely on human programming but the whole comoputing thing is still in its infancy. how long did it take the human brain to evolve? and can we really rule out the idea that if we program these computers with learning algorithims that after a few decades they will have some level of lfe and develop a cabibility for abstract thought?
its not to say we should be terrified and cower at the sight of godlike AI's. i would of course want to argue that virus's constitute a form of artificial life, in that they can reproduce and have a specific life function. give it a few years and it'll be a war between highly intelligent firewall's/spam filters and highly intellgent virus's
full disclosure: i'm a civil servant.
I suspect computers might tap into the long historical use of mathematical systems as oracles: re the I Ching, the Tarot, Kabbalah, astrology... Indeed, to the extent that computers are already used to host these ancient systems and develop new ones like the "Bible Code", I should think that they'll find a role in our religious and literary lives.
Excellent blog BTW!
I think the god analogy succeeds, if it fails it is not because our idea of intelligence are simpleminded but maybe because our ideas of god are too complex.
I am talking about the computational model of the universe. In a sense one can say that the universe is a giant computer calculating the position of every particle at every quantum of time. What is the difference between simulated existence of particles in a computer and "actual" existence in the universe? They both obey the same laws and behave in the same ways.
The computer is in effect creating a tiny universe the size of a nuclear warhead.
For me the discussion here hinges on the difference between God-like problem solving efficiency (robot) and God-like properties inherent in the technology (being), which people seem to be confusing.
Having seen Simon Norfolk lecture he discusses the idea of Supercomputers as Gods only in the symbolic sense and feels the only Godlike powers adopted are within the users of the technology themselves.
It's worth remembering, however, how far this technology has come in the past 25 years alone - mapping the entire human genome, just to site one example shown here - and how the power of these machines multiplies exponentially.
The idea of conscious thought evolving in these machines within the NEXT 25 years is not as ridiculous or "thoughtless" as some seem to think.
Post a Comment