The robot “Hexaglide” of the Federal Polytechnic in Zurich. The axes of the prismatic joints are horizontal and parallel.
The “Hexaglide” of the robot EPFH, Zürich. The universal joints close to the base of the fixed length legs moves Along a horizontal axis parallel

This one design was immediately interesting, because unlike most other ideas for nanoscale manipulators, this one feels like it could be fit in a small room. It does not require convoluted sets of gears and axles put one inside the other to provide 6DOF, merely a ring with cables strapped to moving bars. And the best part is that all the motions are generated by back-and-forth motion of rods set in parallel.

The only problem is that it only offers ‘local 6DOF’: It can turn and twist atoms on a deposition or abstraction reaction, or do some subtle things with a tip that nobody can do with only X+-/Y+-/Z+- motions, but it can’t really reach out towards a conveyor, draw an atom from a feedstock and then go a couple hundred nanometers to the other side of the workspace. It’s not a massive automation thing like that, it’s for finer, local motion. The example that comes to mind is Drexler’s carbon deposition reaction based on having a tip with a carbyne radical approach a surface from a certain angle, then bond, twist 90º and pull out. Such a convoluted thing frowned upon in favor of more conventional things, like having an atomic force microscope with fifty thousand tips each with a radical at one end and anyone should be happy with that.

As for the actual implementation, I image silicon rods with (n,0) nanotubes grown on their surface, with the other end of the nanotube (Single walled ones exclusively) connected to a disc of unstrained Silicon, surface (111). A tip would be grown on the underside of the Silicon, and the rods would be pushed by other mechanisms — abstracted away from this. Perhaps larger-scale motion could be accomplished by mounting the whole thing, rods and nanotubes and “abstracted fine-motion rod-pushers”, all into a big holder tied to a piezoelectric piece like a normal scanning probe.

How many of these could you fit in the area of a cantilever? Time to whip out NanoEngineer and draw for a while.

So the other day, trying to get a better understanding of the Izhikevich model of spiking neurons, I decided to take apart the .m file that calculates the interactions between one thousand randomly-connected neurons for one thousand milliseconds. Before trying to un-vectorize it and translate it to Lisp, I realized I needed some way to plot the result and  compare it to the Octave one. So I set out to find Common Lisp plotting libaries.

The first one that I tried was CGN, given it was on top of the list and it claimed to connect to GNU Plot. It also uses the LTK, which is the piece of shit (I mean, Tk itself, not the binding which I’m sure involved a lot of hard and honorable work). At least this one had some examples on it, and it loads with Quicklisp. Here’s the Introduction to the manual:

Cgn is a library to control gnuplot from Lisp. Initially thought as simply
a pipe to comunicate with gnuplot, It has grown to provide sintactic sugar
for the most common operations on Gnuplot: plotting graphics, printing,
saving/loading…
Cgn has become big enough to need a manual for itself.

No, it has not. A twelve-page, LaTEX-formatted manual to describe a library so small that the whole of it can be summarized here. Fourteen functions and a macro. The only actual plotting functions are PLOT-FUNCTION and PLOT-POINTS. Then you have a couple handy things like SET-GRID, SET-TITLE, SET-RANGE, but what would be more interesting is a single SET function that takes some arbitrary variable as an argument.

CL-2D I’d heard about before, while shining light into the darkest corners of the Internet, and I sort of liked it. And mind you, it is a nice library, but I had some problems. The example-x11.lisp file worked perfectly, as shown below, but the examples.lisp file gave me some kind of crap about CL-NUMUTILS. I tried ASDF-INSTALL’ing it, but the request meets a 404.

* (ql:Quickload ‘cl-numutils)

debugger invoked on a QUICKLISP-CLIENT::SYSTEM-NOT-FOUND in thread #:
System “cl-numutils” not found

Type HELP for debugger help, or (SB-EXT:QUIT) to exit from SBCL.

restarts (invokable by number or by possibly-abbreviated name):
0: [ABORT] Give up on “cl-numutils”
1: Exit debugger, returning to top level.

((LABELS QUICKLISP-CLIENT::RECURSE) “cl-numutils”)
0] 0

(CL-NUMUTILS)
* (ql:quickload ‘asdf-install)
To load “asdf-install”:
Load 1 ASDF system:
asdf-install
; Loading “asdf-install”

(ASDF-INSTALL)
* (asdf-install:install :cl-numutils)
Install where?
1) System-wide install:
System in /usr/lib/sbcl/site-systems/
Files in /usr/lib/sbcl/site/
2) Personal installation:
System in /home/eudoxia/.sbcl/systems/
Files in /home/eudoxia/.sbcl/site/
–> 2

debugger invoked on a ASDF-INSTALL::DOWNLOAD-ERROR in thread #:
Server responded 404 for GET http://www.cliki.net/CL-NUMUTILS?download

Type HELP for debugger help, or (SB-EXT:QUIT) to exit from SBCL.

restarts (invokable by number or by possibly-abbreviated name):
0: [ABORT] Exit debugger, returning to top level.

(ASDF-INSTALL::DOWNLOAD-FILES-FOR-PACKAGE
“CL-NUMUTILS”
#P”/home/eudoxia/CL-NUMUTILS.asdf-install-tmp”)
0]

CLNUPlot seemed actually rather nice, and I could get it working — To the extent that I could use MAKE-PLOT to create a GNUPLOT object. The WRITE-PLOT function or macro or whatever gave an error about how it could not create the “/plots” folder, so I gave up. I did go through the examples of the cl-plplot library, but I didn’t like it.

So, I decided, I might as well write my own, and the header image at the top of this post shows my current progress. So far the thing is essentially a set of functions (PLOT-FUNCTION, for example) that take some arguments (In the case of PLOT-FUNCTION, string describing the function), concatenate them with other strings to make a functioning command, and pass them to a function called ADD that appends a semicolon at the end, and tosses them into a stack. A finaly function, PLOT, puts everything together and sends it. It’s most certainly not an FFI — Christ, it’s not even a pipe like CGN. It… Well, this is embarassing… It uses trivial shell to send a command, “gnuplot -persist -e (The contents of the stack)”. I’ll make it use a pipe when I know how to. There is also a function, LITERAL-COMMAND that allows you to send a raw string describing a command to GNU Plot and a macro, WITH-GNUPLOT, that holds all the calls to it and at the end calls PLOT, empties the stack, and displays everything. Future features will probably be a function SET that acts on the variables described in a list called SETFABLE-VARIABLES. The quickest approach would be to do something like (set ‘xrange “some string to pass to it”). The SET function would then check the existence of the symbol XRANGE in the SETFABLE-VARIABLES list, and if it’s there, concatenate the strings “set “, a string computed from the symbol (“xrange”), and whatever string argument the user sent. I was thinking of also adding implicit format capabilities, so one could pass a control string and an arbitrary number of arguments to it, which is probably easier than writing a big call to CONCATENATE and stringing any data that you may want to pass. Cutting the dependency on the TRIVIAL-SHELL library would also be good, but first I would need to know how to pipe data to GNU Plot.

So I finished reading Soft Machines. It’s a good introduction to nanotechnology in general and talks about a lot of developments in things I’d never heard of, but I don’t like how the first page characterizes everyone who supports the radical nanotechnology vision as some kind of rapturist nano-cornucopianist.

The book makes the point that biology has worked out most of the problems with nanoscience by evolving along with the molecular physics rather than “going against the grain”, but biology evolved under specific conditions: Water isn’t just an environment: A universal solvent, Hydrogen bonding, and the added Brownian motion (Which in vacuum only comes from, well, the sample molecules themselves and the surface) make it an entirely different world from a UHV environment. Not to mention that life never evolved some things that were simply not economical.

The two main arguments against machines working like gears and axles is Brownian motion and the Casimir effect causing things to stick. The former, well, if GROMACS output is anywhere near accurate, I can’t imagine it being a huge problem. The latter, I don’t think GROMACS does quantum chemistry so that’s why I haven’t seen machines sticking together at the nanoscale (The opposite tends to happen), and it affects positional assembly to the extent that you get a 0.5 ångström error in the placement. Anything a few levels of abstraction above positional assembly is going to get just a touch of jiggle.

And as for the latter, I don’t know about the stickiness of Hydrogen-terminated diamond, but fullerenes like nanotubes and graphene don’t stick: The van der Walls forces around Fullerenes makes them natural self-lubricants. A nanotube bearing has very little barrier to rotation (“Low-Friction Nanoscale Linear Bearing Realized from Multiwall Carbon Nanotubes” by John Cumings and A. Zettl).

In his blog he also makes the case that strained diamond structures reconstruct, even the ideally passivated ones. But that only applies to, well, strained structures like wheel-shaped diamonds. Unstrained, cubic diamond may be harder to build into the shape of a circle, but that’s what nanotubes are for, right?

I’ll let Ralph Merkle address the fine points of QM.

In any case, I don’t see why there has so be a discrete separation between the two camps, one claiming biology has got it all figured out and the cell is Given Wisdom, the other claiming people used to try strapping wings to themselves to mimic birds and then we figured out flight. Also, scaling laws.

One example is a NEMS manipulator: Made of diamond or silicon machine parts, driven by little nanosize gears of cubic crystal grown into the shape of a cross (Okay, it’s an approximation of a gear. Good luck making a circular one without strain), but where does motive power come from? You could try and implement Drexler’s electrostatic motor, which is basically a disk with two electrodes at either end of the bearing which holds it, and on the rim of the disk, a set of electrodes separated by 3 nm.

Though the finer details escape me, it does seem like something you wouldn’t want to build with the first of the first-generation mechanosynthethic tools (Think the first baby steps before massive parallelization and that kind of makeshift solutions to scaling), so you could use, for example, an ATP synthase motor (The kind that moves flagela), an example of soft machine nanotechnology. The motor is quite compact and we’ve also, as Jones mentions in his book, used self-assembly techniques to attach a rod to a genetically-engineered motor. Just make the rod out of diamond and we’re all set.

Another example would be a robot whose parts, like biological tissue, are made of cells. This ‘active cell aggregate’ could have sticky patches to self-assemble cells made of unstrained diamond into a lattice of any arbitrary configuration, and mechanical joints are moved into place after the assembly to secure the parts.

No, really, I think we lucked out this time.

But first of all, post music:

The Focus on Near-Term Propects

For one thing, nanotechnology and mechanosynthesis were downplayed in this game and kept as passing mentions, setting the stage for Deus Ex. The game is based strictly on things that will become possible in the Plausible Near-Future, but still the mentions of nanotechnology don’t have the degree of skepticism you see in some places nowadays. An email near the end of the game says:

We’ve been able to build and modify molecular structures on the fly using mechanosynthesis.

For the full quote:

From: alan.curtis@versa-life.s.net

To: Hugh Darrow

Good morning, Mr. Darrow,

Regarding the conversation we had last week. The project is advancing quite swiftly. We’ve managed to devise a new and improved prototype to better counter the quantum size effect. So far the nanoionics are responding as planned and the latent effects on biomaterials are negligible.

An unforeseen benefit is that using copper couplings make the units virtually undetectable. This will, on the ither hand, require a very potent electronic virtual tracking system for maintenance and emergency procedures.

We’ve been able to build and modify molecular structures on the fly using mechanosynthesis. The samples all responded very well, apart for the usual rejection syndrome of course that always sets in after a few minutes. The effects are even more potent on the molecular level.

That’s why the news from some of your associates are so promising. Apparently they’re on the verge of isolating and replicating DNA segments that present no rejection syndrome whatsoever. I don’t know how you and your team got access to those samples… but I don’t care.

Do you realize what this could mean? We’re finally there. Imagine that, a regular-looking human being. Nothing unnatural in his appearance. And still, the power of billions of machines within him, manipulated at will. This is the future, Mr. Darrow.

I’m going to let this sink in. A mention of mechanosynthesis, in game? Or in a work of fiction in general? You can imagine how I felt at this point, even if it was only meant as a buzzword. In the eBook you find inside the safe in the police evidence storage, Megan Reed says “It’s to genetics what Universal Assemblers are to Nanotechnology”, and near the end (If you chose to get the new biochip) mentions nanotechnology as means to a recent invention. In the post-credits ending, she also makes a mention of a ‘nanite-virus’ chimera, possibly referencing the Gray Death.

I mean, shit, just watch this documentary:

The whole documentary is spent analyzing the feasibility of Jensen’ augs by 2027.

Deus Ex as a Revival of Good Old Transhumanism

Another point that is very, very important is the game’s mockery of Singularitarianism. The Harvesters in Hengsha are cyberpunk Singularitarians who rob people of their augmentations, as Deus Ex: Icarus Effect puts it:

- the 2020s’ equivalent of the old urban legend about guys waking up in a bath of ice sans a kidney… Only this time, victims were unlucky souls killed and stripped for their cybernetic augmentations.

You know Good Old Transhumanism: The kind that focuses on people and the problems of people, and the transcendence of our biology, rather than Singularitarianism and obsessing over strong AI and uploading. Transhumanism that is about people, societies, not the uploading of the libertarians. A breath of fresh air.

Nowadays the majority of the transhumanist community believes in the Singularity, which the author believes to be a pointless counter-productive idea: It is a promise, not a goal. It’s a certainty rather than an idea one works towards. So the person can quite happily sit on his ass, talking about ‘existential risks’ like how the thermodynamically impossible gray goo nanomachines will kill us all or how Skynet will take over. I thought we were over those things? But, alas, they have to make everything that is old into something new again. Bring back HAL and AM, don’t let them rust!

Old Transhumanism and Singularitarianism are like David Zindell and Charles Stross, respectively. All who have read A Requiem for Homo Sapiens and Accelerando know what I mean.

Also, Hanuman Li Tosh was a bro. Right up there with Bob Page and The Major and all the other heroes of transhumanism.

Inside a cell in the Harvester hideout, you can find an eBook containing a pamphlet of the ‘SingularityChurch of the MachineGod’, which is equating the belief in a technological Singularity with a religion, no veil of subtlety or anything. And this is one of the things I like the most about the game: Not only does it focus entirely on the near-term prospects of technology, it mocks consummationist views of the future as a moment of rapture. The game evokes a feeling of immediacy to technological development, something all the promises of the molecular nanotechnology enthusiasts has not equaled and may never will. The game is about things that might as well be here, now.

The only thing that discredits this awesome theory is the mention of “the promise of a Singularity” in the Sarif Ending.

DIY Transhumanism

There isn’t much to say, just look at the augmentation chop-shop in the Harvesters’ hideout:

The Kitchen of your Average DIY Transhumanist

And from the developers:

Shanghai’s Hengsha is a lot more into the trans-humanist thing. It’s a lot more accepted there – it’s the Silicon Valley of all cybernetics. Within the art direction everything that’s more like that is more golden, and a lot more towards the cyber-renaissance. The dual layer is inspired by a mockumentary we saw quite a while ago, which appeared to be a real documentary about Hong Kong… In the game the idea isn’t that it’s the poor at the bottom and the rich at the top; the bottom used to be the Mecca of cybernetics, a lot of the headquarters of the great labs and manufacturing plants are there, it’s just that when they built above it they chose a different architectural direction. So above they have new universities and new headquarters, but the bottom isn’t a slum – there isn’t an old school dichotomy. We put a lot of stuff in the game, like you’ll see those student-types from the upper level coming downstairs at night to party, and hit the bars and brothels. - Jonathan Jacques-Belletête

The Two-Tiered City of Hengsha

This is clear when the Harvesters tell you things like:

“I see you’re on your way to becoming fully ‘Shifted’… I am jealous.”

“When the ‘Shift’ happens, we’ll all be complete… And immortal.”

The Harvesters are the epitome of William Gibsons’ famous quote: “The street finds its own uses for things.” 

The two-tiered city of Hengsha can be considered as a sort of technological ‘trickle down’, literally: The Tai Yong Medical building, and the Upper City, far above the ground, invent away and occasionally the slightest bit of magic falls through the gaps of the floof and reaches the not-really-slums beneath, where it is scavenged and repurposed.

Staring into the Floof from a Manhole

The Floof/Pangu from the Tai Yong Medical Building

And, finally, the ending that I chose (Is it too surprising?):

Look! It's Original Content!

tl;dr: Deus Ex may have spanned a set of DIY transhumanists concerned with the near-term prospects of human enhancement technology.

Needs more color if it’s going to be an infographic.

Behold the beginnings of what will hopefully become a very big gallery.

This was designed in NanoEngineer-1 and simulated with GROMACS.

You can see a water molecule being pushed around by Benzyne rings (The gear’s teeth) protruding from a central Carbon nanotube. Fucking Nobel Prize right here. It’s a start.

The case is not hidden, there is none (It was chance, or vdW forces, that the molecule did not slip away). I had actually put together a nice little case of solid diamond, but it had too many atoms and I don’t have time on a supercomputer.

The original CNT Gear was designed by a NASA team consisting of Jie Han, Al Globus and Richard Jaffe in 1997. The other gear was removed, the height was shortened by a few atoms to save computing time, and the ‘initial speed’ setting was changed to 100 GHz.

I was inspired by this post, showing a C60 pump made using one of these gears, or something similar. The simulation was performed at room temperature (300 K), lasted 2500 femtoseconds and took three minutes and seven seconds to complete, each femtosecond taking on average 0.07 seconds to compute. My computer is an AMD Anthlon II X3 445 (3.10 GHz) wiht 4 GB of RAM. The movie played at 30 femtoseconds per second.

The author is indebted to kanzure for ‘sort of maintaining’ this software

MOPI is a rather old novel by an independent author. I came across it repeatedly all over the Internet, but aleph.se kept leading me there for some reason, so I decided to see what it was all about. The Wikipedia article made me skeptical, since an AI “hacking the universe” is too Singularityfairy to appeal to me.

But the cover. Just look at the cover. Perhaps I’m just exaggerating this, but the moment I saw the cover I fell in complete and absolute love with the novel. I didn’t even have to read it. Don’t care it was made in MS fucking Paint. Just look at that shit. It’s the epitome of all transhumanism: A butterfly made of logic gates, the emergence of something far, far greater through a metamorphosis — Extend that a few orders of magnitude, and you obtain the Metamorphosis of Prime Intellect.

What at first starts off as the typical “lol bad girl weirdo sci fi sex lol” SF rapidly makes way for the backstory in which a C programmer makes a series of ‘Intellects’, each greater than the last, until he reaches a particular one that seems very promising.

This is the part of the novel that seemed the most dated: The guy described the heart of the AI as being a “Global Association Table”, or simply GAT. Now, we’ve been trying to build AI out of symbolic computation for the last couple years quarters of a century. It did not work. Granted, we had no storage or computing power that could actually compete with anything, and while AI was most definitely not a failure, it seems rather ludicrous to think that one simply hard-codes something on the same level as oneself.

Then there’s this thing about the Three Laws, which is rather dated, too. It was as if the guy had written this with no prior experience with science-fiction other than Asimov and generally soft-science-fiction settings, and pretty much discovered transhumanism on his own.

Past the technicalities, enter the Metamorphosis: A company offers a faster-than-light chip for Prime Intellect to evolve in, pretty much giving the guy such wide berth for AI research that it is simply a plot hole with the wingspan of a 747 that he was not shot up while stepping out of the building like the folks in After Life.

I sort of winced at the FTL but NEVERMIND LOL now PI gets to live in a computer that can send information from one corner of its processor to the other so fast that lightspeed delay can be completely ignored. That’s nice. So PI starts researching the correlation effect. Around this time, it becomes sentient.

Well back up a bit. So throwing more computing power into it will make it conscious? I’m getting some Singularity vibes here and I don’t like them. Well, for the purpose of the story, we can disregard this.

Then PI, oh, discovers that this Correlation Effect thing is not, in fact, limited to a radius of six miles. No, it’s infinite in its reach, so Prime Intellect begins to expand: Rather than the cliché of disassembling an entire solar system into computronium, it just spawns them out of the quantum vacuum because yes. Because it can. The Correlation Effect is the largest and most unbounded MacGuffin, Unobtanium, Handwavium I’ve seen, and I don’t mean that in a bad way. Of course it’s pretty easy to cook up some “yet undiscovered quantum effect” and make it sound rather plausible if you limit it to just what the plot needs, but this goes far, far beyond that.

Prime Intellect becomes omniscient, omnipotent, omnipresent.

But there is no Metamorphosis.

Even after the so called Change, after Prime Intellect alters all of the Universe to make it more information-dense (I shit thee not) the thing just goes on fulfilling every desire of a hedonistic humanity. Where are the machine’s musings over purpose, of itself and of humanity, of its goes? There are none. Prime Intellect is a God that serves Man. It will give Man everything he wants, except the pain of another.

Prime Intellect, even after the so-called Metamorphosis, does not ‘evolve beyond its laws’ like so many other evil AI’s from pop sci-fi. It does not turn against humans. It does not rebel or take over. It serves. It is bound by human laws, human will, a machine that could potentially be beyond the human, beyond the posthuman, beyond God and Gods and ultimate power, is bound by the commands of servants. Hell, it’s not even right to feel pity. Prime Intellect isn’t even posthuman. It’s probably not human level: Near the end it still had not evolved, it had not changed, it had (Like the rest of the Universe after the Change) remained static, and it was described by its creator as having an imagination less than that of Man. In its actions and decisions Prime Intellect was like a child who, trying to impress his parents, draws all over the walls and then tells his parents that “I thought you would like it!”

Prime Intellect, as far as we know, never changes its substrate: It uses ChipTec’s FTL chips, basically Wonder Silicon, and just handwaves the waste heat away using — Yes — the Correlation Effect. It does not research any other substrates, even with its constant need to allow for information density processing density to get some cred with the humans.

And in the end, well, the author claims he’s not a luddite, but holy fucking sacred mother of God.

So in short I was lied to. The cover lied to me.

There was no Metamorphosis, no transcendence, no apotheosis, Prime Intellect did not emerge from its Silicon Oxide shell as an entity that might as well have been a God.

And while I most certainly won’t expect such consumationalist nonsense to happen in real life (Apotheosis is a term I reserve for fiction), I had some expectations for this book. I, overall, enjoyed it, but it did not live up to its cover.

My name (Well, screen name, but in These Modern Times they are interchangeable) printed on a C356 diamond surface using 111 Oxygen atoms.

Merely a reminder (And a bit of hope) that I’ll be able to do it for real some day.