The Sinularity Summit at Stanford

I attended the Singularity Summit at Stanford University on Saturday, the 13th of May.



I've included below the (mostly) raw, unedited, notes that were taken collaborativelyby myself and several others through the magic of SubEthaEdit and Bonjour (ne' Rendevous) networking. All of the contributors are listed below in the "Contributors" section of the document.

I would like to note that this is the first time I've worked on collaborative notes like this. Watching the others, particularly Cory (from the stage as a speaker and participant) and Meng, take notes and the the speed at which they took notes is a thing of beauty.

The Summit was an all day event; toward the end of the day the notes start to become more comentary than notes, but still well worth the read.

The structure of the document grew out of the needs of the particpants. It was quite organic in its growth.

So, without further ado, I present...

Singularity Summit

Stanford University

Memorial Auditorium

May 13, 2006

http://sss.stanford.edu/

tags: SingularitySummit, SingularitySummit2006

http://csua.berkeley.edu/~marked/SingularitySummit

irc://irc.freenode.net/#singularitysummit


--

CONTRIBUTORS

Cory Doctorow [[CD]] (on stage!)

Mark Rosetta [[MR]] (front balcony)

BJ Snider [[BJS]] (back orchestra)

Chris Anderson [[CA]] (remote, sip audio feed)

Meng Wong [[MW]] (upstairs, front row)

Paul Schreiber [[PS]] (back orchestra)

Tantek Çelik [[TÇ]] (3rd row, center section)
Am posting photos tonight w above tags at http://flickr.com/photos/tantek
I have photos of most slides, so notes of slides can be verified afterwards.
Camera battery ran out of power just at end of Christine's talk.

[[TÇ: Anybody have an extra Canon SD400/SD300/SD450 battery?]]
[[MR: I have a camera, just short SD memory, which MW has offered to fix]]

--

NOTES:

Tyler Emerson -- introduced schedule, etc

[[CD: Tyler has been a machine about getting this event organized -- he's been working his guts out. This thing is as planned as a military op]]

Peter Thiel -- Former CEO of PayPal

First involved in this in 7th grade (g), Max (chess champion)[[?]] said computers would never be as good as humans at Chess. Humans had the creativity to make great players, computers couldn't match it.

Eventually computers got better at algorithmic approaches.

1996: Deep Blue v Kasparov: Kasparov said he was "defending the dignity of the human race." Kasparov barely won. Post-mortem: computers kept getting stronger and stronger. Last year, PIPER12 v John Adams -- John Adams was smashed: 5.5 - 0.5. Computers are the strongest chess players in the world.

Is chess a good proxy for human life? Will computers get better than humans in other domains? In the 70s, 80s, early 90s, chess was seen as a proxy for human intelligence: Wall Street hired me on the basis of my chess skills. These days, that's a mere "Cool party trick"

Will computers get better than humans in every domain over the next few years? Some speakers say it'll happen in the next 20, or next several hundred, or never.


--

Ray Kurzweil: The Singularity: A Hard or Soft Takeoff?

AI is the class of problems you haven't solved yet.

Initial DARPA Car failure - AI can't happen, DARPA car completion - "oh well, that's not intelligence"

[Demoing New Product]

Ray points a camera at his book, takes a picture, does realtime OCR, and a machine voice reads the page aloud, letting a computer talk.

Sounds like "Victoria" on a Mac.

Audience: w00t!

Developed this for the National Foundation for the Blind. When we started developing this, we had to project the power of technology in 5 years.

People ask where all the AI apps are? They're deeply integrated in the technological infrastructure. Cell phone routing, Airplane Landings, financial applications.

Hardware and software [?] are growing at exponential pace. "Computers can't tell the difference between a cat and dog, something a kid can do."

Deep Blue, ? positions / sec
Kasparov, 1 position / sec (Humans pattern matching)

Now computers can detect cats, dogs and ex-wives. What's changed in seven years? Improved pattern-recognition, but much better data-mining for training sets. There are three million pictures of dogs on Google, and 3.6 million of cats [[laughs]].

The National Federation of the Blind has another project: a car that blind people can drive. Four years ago this was ridiculous, now it's plausible.

The Law of Accelerating Returns
IT progresses exponentially. Progress is driven by wealth -- when societies get rich, population growth slows, but technological progress steams ahead. Does exponential growth hit a wall like rabbits in Australia? Yes -- but then new paradigms grow up.

Moore's Law is 5th paradigm in line

[A large, somewhat self-organizing] nanotube-based circuits will hit the market next year. Even if you're skeptical about the feasibility of 3D molecular technology (this isn't a reasonable skepticism). 2D chips will have 1 nanometer features in 2018 -- enough computation to emulate all regions of the brain based on the most conservative estimates for $1000.

This is a scientific theory.

Moore's law is just one example of many of exponential growth. A scientist said it took 18 months to model this part of the brain, and at that rate, it will take a century. But that's predicated on technology not advancing -- spatial revolution of brain scans doubles every 18 months.

The genome project -- 7.5 years into a 15 year project finished 1% of the project. More than 50% was completed in the last year.

* Human Genome - 15 years

* HIV sequenced in 7 years

* SARS sequenced in 31 days

Our intuition is linear. If something can be expressed in information, then it can take advantage of exponential growth.

Rate of paradigm shift changes is doubling every decade.

People didn't have search engines, social networks, viable net advertising just a few years ago. Pace of change is accelerating.

Evolution breeds capability.

Evolution used DNA, an information backbone, to create the Cambrian Explosion, and homo sapiens grew faster still.

3 changes: 10s of thousands of bytes of information

* jaw

* neurons

* thumb

When I came to MIT in 65, an $11MM computer filled a room this size and was 7000X less powerful than your cellphone.

Intel says 2022 for 4nm process using photo-lithography - don't even need molecular nanotechnology or other nano tech to bring on Strong AI.

Bought the equivalent of a transistor, for $40, the size of a brick, telephone relay circuit.

Look at how smooth the table of world-wide growth of semiconductors is -- how can it be so smooth? There are other similar phenomena, e.g. thermodynamics, individual particles are unpredictable, but the system produces predictable outcomes.

1990 $10/base pair to make changes or sequence

Now = 0.01/base pair

Nano-MEMs going into animal trials already. Cancer detector/cell destructor.

Doubling the resolution of Brain Scanning every year.

Language: an ideal laboratory for studying human ability for hierarchical, symbolic, recursive thinking

[[BJS: I wonder if he's read Matruana "Understanding Computers and Cognition" )Publisher: Addison-Wesley Professional; 1st edition (January 1, 1987)
Language: English ISBN: 0201112973) and "The Tree of Knowledge" (Publisher: Shambhala; Revised edition (March 31, 1992) Language: English ISBN: 0877736421)?]]

The genome, which contains the entire design of the brain (4 types of neurons, wired this way, repeat 10 billion? times) has fewer bytes than Microsoft Word -- it's a probabilistic fractal that expands itself.

A Mandelbrot set is 6 bytes long and produces complexity that takes more than 6 bytes to express.

Models often get simpler at higher models -- modeling a whole pancreas is simpler than modeling individual pancreatic cells. You don't need quarks or protons to explain chemistry. Neurology has several levels of abstraction that we're beginning to reverse engineer -- and this is an exponential process.

Pattern recognition used to create Arabic to English translation using large data-sets [[BJS: is this similar to current discussion of NSA "large database" pattern recognition with phone data and "terrorist" look-outs?]]


--

Douglas R. Hofstadter: Trying to Muse Rationally about the Singularity Scenario

4/1 Spiritual Robots Symposium
Two books: Age of Spiritual Machines + Robot

"Who will be we in 2093?"
On a republishing, changed number to 400 years in the future (2493). Scared by his own imagery

I don't like the idea that my grandchildren will be made obsolete by technology.

Age of Spiritual Machines and Robot instilled a lot of fear in me. I organized a panel at the University of Indiana and none of the panelists talked about the stuff I was interested in.

I came to Stanford for a sabbatical and organized a followup panel. My goal was to think rationally about the scenarios suggested by these books. Hence "Trying to muse rationally about the Singularity." Ray wanted to know if I meant that I was the only rational commenter on the subject, but I make no such claim.

Rationality is the product of the airing of many different views.

There aren't many skeptics about the Singularity -- in the sense of being skeptical of its conclusions. Ray and Hans Moravec, Kevin Kelly, Ralph Merkle (nano-technologist), Bill Joy(chief scientist at Sun Micro), Frank Ray (SETI), John Collum (invented genetic algorithms) and John Posner (sp?) (inventor Genetic programming).

(editor of wired)

Bill Joy: Commandeered 0.5 of the time, but talking about the idea that we'll be smothered by self-reproducing technodust. This is the low-IQ end-of-humanity scenario.

I want to talk about the high-IQ scenario, that we'll be made obsolete.

[[ MW on obsoleteness: So, an obsolescence scenario looks like, maybe, space robots on Mars, or intelligent spaceships flying off to colonize Alpha Centauri with glistening metal men grown in vats, with nary a carbon-based life-form in sight. Now, couldn't one argue that human beings have made chimpanzees obsolete? Somehow, I don't think the chimpanzees mind: I don't see them holding conferences about "has the Chimpanzee Singularity arrived? The threat of Homo Sapiens and the development of the Chiquita Banana as a fundamentally incomprehensible agro-economic system. So, even if we are made obsolete by technology, *we won't care* ... because, well, there will always be frat boys trying to get laid, and no matter how forcefully we argue that Space Robots have Co-Opted the Human Galactic Destiny, they'll continue to ... try to get laid.

I guess we get worried when humans try to extinguish chimpanzees; but humans aren't geared to handle that rationally, only viscerally. Our natural response to such a situation is, in fact, the right one: to continue to try to get laid. John Connor fights SkyNet, but John Connor also has babies. ]]

Only two people voiced any skepticism about the high-IQ scenario: the genetic algorithmists. They were skeptical about genetic programming/algorithm to evolve high-level intelligence.

In Ray's new book -- which I highly recommend -- he's done a great deal of impressive research of all sorts, from nano/nanotubes, etc. Ray is a past-master at displaying exponential curves as straight lines by taking logarithms. The book is filled with those things. They're interesting to contemplate.

It also interestingly refutes critics, especially John Searle. There's a lot of technical hand-waving though. We don't really know what will happen in 10, 20, 40 years. The Law of Accelerating Returns is meant to be analogous to Thermodynamics, like an ideal gas.

But in physics, you use statistics to derive the laws of thermodynamics. But Ray's law is more like a tendency or trend, albeit with a lot of evidence to suggest its continuance. It's an extrapolation.

What demarcates one human from another once uploaded in cyberspace? Ray talks about merging and unmerging. But it's not clear what a human would be in such an environment.

I doubt that computers today compose music as well as Bach.

[[Comment BJS: Why does the entire environment have to be simulated if humans become software uploaded to cyberspace - computing machines interact with the existing environment already.]]

Our most powerful computer can only simulate one protein -- if intelligence requires simulating lots of them, we're not clearly within sight of simulated intelligence.

Some cartoons to illustrate my POV:

Cartoon 1: Prehistoric rocks argue that they could never gain life, a recently created life-form mocks them: "Ribo! Ribo!"

That transition is far more profound than mere AI.

Cartoon 2: "Sentient life on land?! Don't make me laugh! You couldn't even breathe -- the idea is self contradictory. Don't worry." A nearby land-life-form amphibian mocks: "Ribbit, ribbit!"

Life can change substrates!

Cartoon 3: John Searle: "Sentient life on silicon? Chinese room! It's self contradictory! Don't you worry!" Computer: "Robot! Robot!"

Cartoon 4: Ray Kurzweil: "Hit a wall! Don't make me laugh! Recursion! Faster smaller cheaper smarter! Nothing will stop exponential growth! Limitations? The idea is self-contradictory, don't you worry!" A wolf eating an Australian rabbit: "Rabbit, rabbit!"

One of Ray's themes is the end of mortality. "Live long enough to live forever." This is about the survival of our essence forever.

Cartoon: 5: People in wheelchairs: "Live forever, don't make me laugh! The only certainty is death and taxes! It's self contradictory! DOn't you worry." Kurzweil: "Ray bet! Ray bet!" He doesn't bet money, he bets his life!

Once we're all in cyber-heaven, how long can we stay there?

Cartoon 6: AI: "Crash? It's self contradictory -- nothing can go wrong! Go wrong! Go wrong!" "Reboot, reboot"

These issues should provoke you into thinking about this.

Life can exist in many substrates -- Robert Forward's "Dragon's Egg" -- life could exist in a neutron star. It's an interesting speculation and there's no reason in principle to believe that such extremes forms of life could exist -- nor life in nanotubes.

It's plausible.

I don't believe in the timeframes, despite the exponential curves.

What I'm concerned about: How realistic is this? I've asked a number of highly informed friends from different disciplines. "Nutty" "Scary" "I don't know" "Reasonable and probable" -- but none of them have read the book!

This is strange: the scientific world doesn't take any of this seriously. It's not chatter among physicists, who pooh-pooh this. Most scientists have a skeptical attitude. Having read Ray's books, I'm less skeptical.

Why am I skeptical? I'm a conservative, it's emotional.

Also: the ideas are marred by being blurred with sf. Hans talks a lot about time-travel. When computers are sufficiently complex, they universe's history will be recreated uncountably many times. These kinds of scenarios are fantastic and are blended in with the rest of the book, which contaminates the book. How sane is the author?

Ray does the same thing: It's hard to take Ray's immortality seriously. It's hard to take utility foglets seriously. Sub-quark engineering, galactic-scale computers. These are wildly beyond any degree of speculation that I'll accept.

Ray also mixes up "the genetic code" and "the genome." These aren't the same thing.

Exponential curves don't have "knees" -- there's no such thing!

But I believe that some of the ideas are possible. There are a lot of partially true things in Ray's book -- they're blurry, there's a lot of hand-waving. Multiplying the partial truth of a lot of things available, you end up with a very small probability.

Listening to Ray is like listening to one side of the divorce. I want to hear serious scientists to deliver a serious skeptical response. I want a real debate.

[[CD: We're running late, so Hofstadter was cut off]]

--

Nick Bostrom – Artificial Intelligence and Existential Risks

talk ends at 11:07 am

More "existential risks" than AI

Risks:

* Scope

* Intensity

* Probability

The worst disasters are but a ripple in the "sea of life"

Types of Risk: Anthropogenic and "all the rest"
We have survived "all the rest" for thousands of years

Anthropogenic existential risk is the real issue. It is of the greatest concern.

Anthropogenic risk is "in theory" containable.

Types of existential risk
* Bangs
* Crunches
* Shrieks - limited form of post-humanity
* Whimpers - Gradual disappearance of things we value

Bangs:

* Nanogoo

* Rogue AI

* Life is a simulation, someone powers it down

[[MW: See Greg Egan's Permutation City, http://www.amazon.com/gp/product/006105481X/ ]]

* Nuclear holocaust

* Biotech holocaust

* Grey goo caused by non-weapons accident

* Etc

Crunches:

* Resource depletion

* Misguided world government or another static social equilibrium stops technological progress

* Dysgenic pressures: Evolution that leads us in the wrong direction

* Technological arrest

* Social collapse

Shrieks:

* Flawed super-AI

* Repressive totalitarian global regime

* Takeover by transcending upload

[[MW: aka. SkyNet, see Terminator with Arnold Schwarzenegger! ]]

[[MW: sidebar: 2004: The future of human evolution in Death and Anti-Death, ed. Charles Tandy (Rita University Press: Palo Alto, Ca 2004) pp. 339-371

Whimpers:

* Ability to self-modify erodes our potential

* Killed by ETs

* Loss of human fertility/escapism

Challenges:

Biases Galore?

* Good story bias?

* Scope neglect

* Calibration and overconfidence problems

* Bystander apathy

One experiment:

How much would you pay to save 200 birds, 2000 birds, 200000 birds?

* 200: $80

* 2000: $84

* 200000: $88

People aren't good at valuing risk and mitigation. This is a challenge for philanthropic orgs to keep in mind.

Reducing existential risks even by tiny amounts is worth more than anything else.

What to do?

* More research into methodology

* More research on risk-categories

* Build institutions focused on evaluating risk

* Specific efforts to reduce specific threats:

* Foster peace

* Differential tech development

--

Sebastian Thrun – Toward human-level intelligence in autonomous cars

associate prof at stanford.

talking about the DARPA challenge of neato vehicles going from one place to another on their own.

http://www10.giscafe.com/nbc/articles/view_weekly.php?run_date=22-Mar-2004

These things were supposed to be smart, but What came out of it was just a disaster.

They had GPS and whatever. Some of them drove back to their home state.

Most of them didn't make it 5% of the course.

The heaviest is a 15 ton [[ MW: what looks like a bright green garbage truck ]]

[[BJS: It's an airport firetruck.]]

Motorcycle gets out of the starting gate and falls over.

Evidently riding a motorcycle is something that requires human level intelligence.

This is a story of how we are at a critical juncture in driving. Everybody says 2005 is when we invented autonomous driving, but there's actually 30 years of history that has come to a head in the last 2 or 3 years; development is accelerating.

Slide: an autonomous vehicle (AV) with gizmos and gadgets strapped on the top and front and inside. Across the bodywork is a paint job: sponsored by Red Bull.

3d laser terrain acquisition and realtime mapping. as it scans the road it can see vertical elevations.

very sexy false-color animation of the vehicle following the road.

it took us months and months and months to write a piece of software to find a road. 30 years ago it was a hard problem and it is still a hard problem. the system knows what the road looks like in the near range. so we train the car on that and look for something similar farther away. as the system learns how the road looks right now it uses that model to extrapolate. so it handles transitions from paved roads to grass roads.

stanford racing team, arizona desert, september 2005.
418 flawless continuous miles.
average speed 22 mph, max speed 38 mph.



funny video: UCB's autonomous motorcycle ends up ... in a pond.

[[ MW: i wonder if the vehicle is programmed to slam on the brakes if it can't figure out where the road is anymore. ]]


the motorcycle falls over ... and then it magically rights itself using a big arm that sticks out from the side!

anthropomorphizing: sending these things out is like sending your kid to college, and it comes back without a scratch... and hopefully not pregnant.


CMU developed an engine problem.

some ooopses. SUVs drive up a hill, off the road, into the bush.

at mile 80 CMU developed an engine problem with the hummer they had flipped before the race. by mile 103 we were given the opportunity to pass the front running vehicle. in all 5 vehicles finished, but we were happy to take the money because we came in 11 minutes ahead of the rest.

the Stanford car recognizes the CMU car as a piece of metal sitting in the desert and successfully avoids it.

[ laughter ]

cliff on the left side would be a final catastrophic event.

end of talk: Does this take away from human identity?

If you identify yourself with a fast sports car, it does! [Laughter] And rightfully so.

If you care about that, go back 400 years when it was all agriculture; do you worry that not working in agriculture takes away from your identity?

Is this (finally) the advent of Strong AI? Strong disagreement with people like ... .

Have seen growth curves for RAM and disk but the curve is not there for strong AI. we're still in the flat of the curve, before the inflection point. It's still really hard for AI to tell the difference between a tumbleweed and a rock.

Will this advance human life? Yes, even though it's not Strong AI. [[ MW: Kitt can drive the car, Kitt doesn't have to pal around with David Hasselhoff. ]]

The average American spends over an hour a day stuck in traffic. Wouldn't it be nice to take this time and make it more productive; instead of sitting in traffic, you could delete spam from your inbox.

* Improve quality of life

** The 14 year old kid can "drive" himself to soccer practice.

** Blind people can get around.

* increase highway throughput.

8% utilization of freeway area during traffic peak


** [[ MW: yay, we can burn more fossil fuels. ]]

* change urban real-estate.

Predict: soon, something will come out that will be disruptive. People have asked me to put a timeline on this. I'm not good at timelines, but I think by 2008. That means that in a 5 year period we have made a significant advance.

2010 may be a target for reliable urban/highway driving

2030; 50% of miles driven autonomously.

[[ MW: it'll be 2010 before we can do it, and 2060 before regulators approve it. Just as science advances by the death of one scientist at a time, public policy advances by the death of an entire generation of lawmakers ... and the funny thing about life extension is that it imposes a natural slowing on how fast we allow technologies to come out! ]]


--

Cory Doctorow – Singularity or Dark Age? How the copyright wars threaten technological progress

How did 3 million pictures of cats get on the Internet? So much investment poured into tools that make great stuff. What causes human beings to do this?

Tools aren't the only thing to accelerate themselves; diseases do too. Yet we don't have a law of accelerating ebola.

[[ MW: yes we do, it's called epidemiology and Malthusian limits, etc. ]]

This narrative give us a way to control the world in which we live. Agency. In the greek plays, they had Deus Ex Machina. That was satisfying to the ancient greek. there are a lot of different theories. if the world is not moldable, where you have no agency, where your fate is not in your hands, then a god in a box solving your problems is as satisfying as any other ending.

But nowadays, in the Modern post enlightenment age, agency is primary. if the character does not take hold of her fate, and just gets buffeted about, the reader gets bored. this happens in amateur fiction. you do not see it because it does not get published.

technologies allow us to better control our world.

[[ MW: for technocrat, educated, rationalist, geek-type values of "us", but not for the folks who work at Walmart, methinks. for them, technology is a tool others use to control them. Plus ca change, plus c'est la meme chose. ]]

[[ CD: No way -- these people buy microwaves, VCRs, cars, etc in droves -- all technologies to let them better control their universe. No one buys a subscription to 300-channel cable services if they don't want to be in the driver's seat. ]]

Nowadays, though, if you try to refill a toner cartridge, or service a car engine without permission from the manufacturer ...

DRM is the means by which someone else controls your personally owned devices.

agency is the difference between frankenstein and i, robot.

if you're in charge, you're happier; if you're not, you are sadder and madder.

[[ MW: unless you believe Barry Schwartz, Paradox of Choice. try asking the Amish what they think. though maybe by opting out they are actually neither controlling nor controlled. ]]

Bruce Schneier: (premise of movie "Firewall")

and nowadays it is in fact unlawful to tell people how DRM systems work, thanks to DMCA and UN and WIPO.

all over the world, we are requiring that users not be allowed to understand their technology. the horror, the horror!

"compare AOL to real Internet email". there was a time when AOL had no spam problem. There was a time when it was a closed world, a walled garden. And AOL could kick you off.

Then federation and opening happened, driven by the power of network effects.

Kathryn Myronuk likes to say: All complex ecosystems have parasites. If you render the system simple enough to not have parasites, it becomes uninteresting and nobody will use it.

[[ MW: I ranted about this stuff at http://www.circleid.com/posts/internet_governance_an_antispam_perspective/ ]]



You can't solve the problem of parasites by making the system simple.

Comparison of CDs vs DVD. The crucial difference is that media on DVDs are protected by DRM. Circumventing DRM is illegal. CDs have no DRM.

[[ MW: both ways, though, violation of copyright law is illegal; the DRM is just a nicety. ]]

This is interesting because CDs spawn new inventions like iPods; DVDs have not done the same thing.



Maybe the film/music/recording/publishing industries will go through a Singularity of their own.

[[ MW: I think this will happen, because the Internet is helping us return to many pre-industrial models of society; and pre-industrial music was not a commercial enterprise. Musicians did it for fun (garage bands) and if they made any money it was because they had sponsors (paid gigs to play the piano at a fancy restaurant in the background) or because they went on tour -- troubadors -- and today bands go on concert tours to make real money -- the album sales are just a promotion channel. ]]


The US was a pirate nation for its first 100 years.

Copyright regulates technology. Making derived works is a technological act; therefore, law regarding the making of derived works regulates technological development.

We need new rules and regulations that see to it that new art is created and creativity flourishes.

... but to say that today's copyright law is helping, rather than hurting, is patently untrue. [[ MW: "patent"ly, ha, ha. ]]

if you could reproduce hot lunches at will with no incremental cost, nobody would call that a disaster except the delicatessen industry.

[[ MW: disagree; the folks who design lunches would complain that they weren't getting rich off licensing. ]]

Elsevier has a model where university libraries that cancel their subscriptions will see their "holdings" vanish.

nobody knew that saving the whales and eating health food and stopping pollution were connected until somebody invented the word ecology. [[ MW: did he mean "environmentalism"? ]]

list of Good Guys to Join: EPIC, EFF, etc. Get involved! It's a movement! Let's have more control over our environment instead of our environment controlling us!

[[ MW: so, um, what does this mean? Just as a vegan declines to eat meat, I am declining to watch Hollywood product. ]]

--

K. Eric Drexler – Productive Nanosystems: Toward a super-exponential threshold in physical technology

Let's talk about manufacturing.

We gotta give something for the farmers to do, I guess.

We're talking about information-driven manufacturing.

The process is much like software development.

We're creating atoms the way we create bits.

Nanorex is developing open-source, freely available designs for molecular level thingies. BGPL?

Productive Nanosystems

* Will treat atoms like bits

* Enabling research is advancing rapidly

* Development will be super-exponential

* Will transform physical technology

* Will advance far, then hit physical limits

Slide: the famous STM slide of the "IBM" text made with 35 atoms.

That was 13 years ago.

Today, we can make stuff with 1 million atoms.

Rothemund Paul W K. "Folding DNA to create nanoscale shapes and patterns", Nature.

Using DNA as a structural material. Build this stuff with a turnaround time of 1 week.

Protein folding simulation is hard. It takes forever.

By contrast, protein design takes hours. It is routine.

Slide: a productive nanosystem. It is a ribosome. It takes instructions in the form of RNA, and builds proteins.

Scale comparison: the stuff we're making is about the same size. We can engineer things to approximately the same scale. So, woot.

People think of proteins as icky soft things.

Meat is mostly water.

This Stuff is a protein; inedible and deadly. Cow horn.

Plastic: plastic is protein, not meat. Polymers.

Here's a sheet of graphite. C222 H42. Made by solution chemistry.

Simpson CD et al. Synthesis of a giant 222 carbon graphite sheet. Chem-Eur J. 2002.

Here are some nanomachines we want to build. They do familiar mechanical operations on a molecular scale. They transmit power. etc.

Standard mechanical approaches are where we'll start. If you want to do something more radical with biological approaches, go for it. [[ MW: he is openly recognizing paradigm inertia. ]]



Moore's law says: we'll have a billion transistors per CPU!

Nano fabrication will give us a billion CPUs in an air-cooled laptop!

applications in medicine: something like a white blood cell. targeted. with fast sequencing, you can build an instant vaccine. [[ MW: this was predicted in a science fiction novel somewhere, where a skiing accident was instantly cured and the patient given a drug to forget they ever had any pain. ]] [[PS: Cory wrote about something like this too, didn't he?]] [[CD: I think you're thinking of 0wnz0red]]

[[ MW: an Elixir product would be an orally administered nanomedication which state sequenced the patient's genome, calculated the desirable biological state, analyzed deviations, diagnosed the disease at fault, and synthesized a vaccine which was itself a nanomachine that went around and ate all the bad bugs. ]]

Macro scale: can make 50X the strength:weight ratio of the space shuttle.

Trends don't tell you what the limits are:

* Second law of thermodynamics

* Quantum measurement limits

* Light-speed limit

Technology is about exploring limits.

Science can revise limits up or down.

While nanotechnology is advancing rapidly, it's unfocused.

However: bogus criticism of nano has fallen out of fashion (e.g., you can't turn lead into gold)

Foresight institute is making a nanotech roadmap

Non-US journos ask better questions

R&D agencies seeking planning advice on how to take their research and move it around a little bit.

National leaders now referencing the concept of nano

People are paying attention to more than "Engines of Creation" -- now interested in "Nanosystems: molecular machinery, manufacturing, and computation," a more technical work.

It will change everything. From the bottom up.

[[ MW has been hosting the subethaedit thing, but will have to shut the laptop to go forage. y'all save and somebody else please host this afternoon. I'm running out of juice. 20060513-12:23 ]]

--
Max More – Cognitive and Emotional Singularities: Will super-intelligence come with super-wisdom?

http://www.maxmore.com/

[[ MW's buddy JT: Of course not. You roll separate 3d6 for each. ]]

[[ MW: I wonder if the transhumanist extropian types have groupie chicks with tattoos and old copies of Mondo 2000 in their basements. Perhaps Cory could enlighten us? ]]

[[CD Not me, but maybe I missed the gold rush]]

[[ MW: Max More is the ultimate name for an extropian. The rest of us should just go home. ]]

Cognitive intelligence = Wisdom
Machine intelligence will have these problems as well. When will machines have ≥ human wisdom?

Super-wisdom (Integral Intelligence)

* mindfulness
* creativity
* objectivity
* cognitive capability
* extropic condition [[ MW: WTF ]]
* social-environmental-institutional support

[[ MW: i define wisdom as: knowing alternative outcomes, choosing the best, knowing what needs to be done to bring it about, and doing it ]]

Foundations of Foolishness


* bounded awareness

Cognitive blinders mean that, despite highly relevant, easily accessible, and readily perceivable information:

** we may fail to see or seek it

** we may fail to use the information that they do see because they aren't aware of its relevance [ the /knowing-doing gap]

** we may fail to share information with others

recommends the movie Memento http://www.imdb.com/title/tt0209144/

* Theories about the world

The "cascade of consequences." We tend to ignore possible outcomes or consequences in order to simplify complex decisions. Leads to five biases:

*** ignoring low-probability events
*** limiting the search for stakeholders
*** ignoring the possibility that the public will "find out"
*** discounting the future
*** undervaluing collective outcomes

Faulty judgment of risk --- shaped by denying uncertainty; not learning from experience (hindsight bias); poor risk trade-offs; and framing

* Theories about other people
[went past slide very quickly]

The perception of causes --- we tend to focus on people ("fundamental attribution error"), perceive events differently, and commit sins of omission.

People should write down their predictions so they can be checked in the future.

* Theories about ourselves

... often exhibit the illusion of superiority in 3 ways:

*** the illusion of favorability
*** the illusion of optimism
*** the illusion of control
*** Plus: our theories of self are infected by self-serving fairness biases, selective memory, and overconfidence. We don't learn from our mistakes because we fail to keep track of our decisions and their outcomes.
*** lack of access to emotional centers of the brain. Emotions go one way, it's hard to change them.

[[ MW: Meng did a diagram once that showed:

GENES -> communicate with emotions/feelings -> MIND
MIND -> take actions to satisfy -> GENES
]]

AI Evolutionary Paths

** Default path if not guided?
** Artifactual intellects have one (potential) advantage --- they are not (directly) conditioned by foresight-deficient evolution

** But: Artificial super-intelligence will be created and shaped by /people/ and /organizations/

!! Beware the hyper-rationalist archetype embedded in our intellectual culture!

[[ MW: aka, "are those geeks crazy or what?" ]]

Solutions!

** Cognitive/emotional error correction
** Integral intelligence
** "Friendly AI"
** The ProP as a core part of "friendly AI" programming

[[ MW: I wonder what the military makes of the term "friendly AI" ]]

** Embed the ProP/structured thinking in default social mechanisms

** Importance of openness in AI projects (a component ProP principle)

"ProP: Proactionary Principle"

[[ MR: hey hey, Cory can't see this session so I"m going to take it over to see if it fixes it...Sorry about that, but I guess it works now. I guess I have a good reception spot being on the balcony...]

Proactionary Principle is documented at: http://www.maxmore.com/proactionary.htm

The ProP uses the most well-tested, leading procedures to structure the decision-making process.

Its ten component principles embody the wisdom of structure so as to guide decisions toward integral intelligence.

Critical Thinking + Creative Thinking + Evidence-based forecasting.




1. Freedom to innovate: Our freedom to innovate technologically is valuable to humanity. The burden of proof therefore belongs to those who propose restrictive measures. All proposed measures should be closely scrutinized.

2. Objectivity: Use a decision process that is objective, structured, and explicit. Evaluate risks and generate forecasts according to available science, not emotionally shaped perceptions; use explicit forecasting processes; fully disclose the forecasting procedure; ensure that the information and decision procedures are objective; rigorously structure the inputs to the forecasting procedure; reduce biases by selecting disinterested experts, by using the devil’s advocate procedure with judgmental methods, and by using auditing procedures such as review panels.

3. Comprehensiveness: Consider all reasonable alternative actions, including no action. Estimate the opportunities lost by abandoning a technology, and take into account the costs and risks of substituting other credible options. When making these estimates, carefully consider not only concentrated and immediate effects, but also widely distributed and follow-on effects.

4. Openness/Transparency: Take into account the interests of all potentially affected parties, and keep the process open to input from those parties.

5. Simplicity: Use methods that are no more complex than necessary

6. Triage: Give precedence to ameliorating known and proven threats to human health and environmental quality over acting against hypothetical risks.

7. Symmetrical treatment: Treat technological risks on the same basis as natural risks; avoid underweighting natural risks and overweighting human-technological risks. Fully account for the benefits of technological advances.

8. Proportionality: Consider restrictive measures only if the potential impact of an activity has both significant probability and severity. In such cases, if the activity also generates benefits, discount the impacts according to the feasibility of adapting to the adverse effects. If measures to limit technological advance do appear justified, ensure that the extent of those measures is proportionate to the extent of the probable effects.

9. Prioritize (Prioritization): When choosing among measures to ameliorate unwanted side effects, prioritize decision criteria as follows: (a) Give priority to risks to human and other intelligent life over risks to other species; (b) give non-lethal threats to human health priority over threats limited to the environment (within reasonable limits); (c) give priority to immediate threats over distant threats; (d) prefer the measure with the highest expectation value by giving priority to more certain over less certain threats, and to irreversible or persistent impacts over transient impacts.

10. Renew and Refresh: Create a trigger to prompt decision makers to revisit the decision, far enough in the future that conditions may have changed significantly.

[[ MW: "Max: with that, I now embark on a career in management consulting." ]]

Four possible futures:

* No super-intelligence, no super-wisdom: business-as-usual

* super-intelligence, no super-wisdom: Fast and messy

* super-intelligence, super-wisdom: Smooth and speedy

* Super-wisdom, no super-intelligence: Smooth and moderately fast

[[ MW: ooh, a 2x2 matrix. this feels very BCG. ]]

[[CD: BCG ==?? "Bacille Calmette-Guerin"? ]]

[[ MW: Boston Consulting Group, see http://www.netmba.com/strategy/matrix/bcg/ ]]


Super-intelligence

No | Yes
|
Yes Smooth, | Smooth
modestly fast | and speedy
Integral |
Intelligence --------------------|------------------
No Business as usual | Fast but messy
|
|
|

--

Christine L. Peterson – Bringing Humanity and the Biosphere through the Singularity

Goal: Well-being of unaugmented humanity and the biosphere (augmented humanity is important, but more capable of taking care of itself)

"I'm looking for a pathway for humanity and human values."

* Three ways to augment intelligence:

* Humans

* Other species

Make sure -- if anybody's doing this, don't use chimpanzees; use bonobos. They're much nicer people. [[ MW: They use sex to get power. Chimps use power to get sex. See http://www.amazon.com/gp/product/0300099835 ]]


* Software


Here we assume the last --- software --- for simplicity in doing thought experiments.

Nothing wrong with having lots of brighter entities around -- humans are already on a brightness spectrum.

* Need protection from physical and economic coercion

* The issue is power and how it's wielded

We're already experienced with powerful entities:

* Wealthy individuals

* Governments

*Corporations

We solve this with Techniques for Managing Power:

* Constitutions

* Rule of law

* Property rights

* Mutual agreements


Securing of Property Rights

** Computational
** Non-computational

Securing non-computation property rights such as physical property increasingly will depend on securing computational property rights.

Regardless of whether we Singularity or not. More and more of the physical world is controlled by software. If the thought of that doesn't disturb you, you don't really understand software, heh heh.

We need secure OSes:

[[TÇ: waiting for Cory to have a field day with this one, feels like a sales pitch for Palladium etc. "Contracts" make the whole system inaccessible to "normal" people, who have no time/interest to attempt to read/understand the contracts, like today's crappy EULAs. ]]

[[ CD: No, this is more like a cryptographic "contract" -- like a three-way handshake -- I agree to send you a key, you agree to send me a key, we use another key to enforce the agreements, etc -- it's an old-time cryptographer's gedankenexperiment -- obscure now, trendy about 5 years ago]]

[[TÇ: I would like to believe this is like a cryptographic contract, but the phrase "smart contract" makes me VERY suspicious.]]

[[CD: I actually quite like this: you can easily interpret this as, "You should have control of your computer, you should not have control taken away from you, EULAs shouldn't be a means of ripping you off]]

[[CD: She's actually selling Mark Miller's "Capabilities computing" model -- not trusted computing -- presented at ETECH about 4 years ago]]

[[TÇ: I hope that is the case. Also, IMHO, "Automated mutual defense" is a very quick way to "mutually assured destruction" (and their acronyms are anagrams as well. "reacting violently is defense" WTF?. Need to start with user/consumer "rights" first. Then we can build contracts on top of that. Until we get such "rights" established, typical organizations will continue to write and assert contracts that attempt to reduce user rights as much as possible.]]


[[BJS: "Your software rights end where mine begin" type of thing?]]

[[CD: Yeah]]

* No authority for programs by default

* Capabilities computing

* Constitutional system of property and contracts

* A minimum framework of 'law' enabling voluntary agreements and enforceable contracts.

[[ MW: see Lessig's Code and other Laws of Cyberspace http://www.amazon.com/gp/product/0465039138/ ]]



* Smart contracts

Simple contracts embodied in software

Automatically enforced!

Types of rule to be enforced must be vastly simpler than current legal code. Example: typical real estate lease is far too complex.

Design contracts and rights such that system can understand and enforce

An environment where force doesn't work

Have "intelligence explosion" occur in this environment.

[[ MW: Obviously, to 0wn the system, you have to be Neo. ]]



On top of smart contracts, we want: Automated Mutual Defense

Need consensus on initial asset division, what i a violation, and what response is merited for attempted violation. Simple rules enforced.

Resources brought to bear must be greater than violators'. Must obligate self, some resources.

Filters enable a learning component.

Throw away key, large supermajority, or long cooling-off period (for stability))

Reacts violently to inappropriate violence by any entity, including AI.


PROBLEMS with this: social engineering. people will hax0r into these systems.

[[ MW: if you're running windows, you're 0wned anyway ]]

[[ MW: how is this different from a japanese dude getting sent to jail for mugging someone in world of warcraft? ]]

[[ TÇ: camera battery ran out right here. ]]

E Language: www.erights.org
KeyKOS / EROS lines of operating systems: eros-os.org

Capabilities: www.cap-lore.com

Smart contracts; szabo.best.wnn.net

Nanotech information and action

www.foresight.org
Blog blog blog! www.nanodot.org
Foresight Vision Weekend, Fall 2006
members meeting, primarily nano but addresses multidisciplinary issues (nano/ AI)
14th Foresight Conference on Nanotechnology, Spring 2007, nanotech only

www.foresight.org/roadmaps



[[ MW: One sign that we're really far away from the Singularity is that even though people produce PowerPoint presentations that could be put on a globe-spanning information sharing network, here we are in the audience using cutting edge realtime file editing technology to ... transcribe those presentations. ]]


[[CD: Why is that anti-Singularity? If anything, it suggests we can have a singularity even if the old guard don't want to cooperate -- IOW, that Singularity does not presuppose cooperation from the people whose labor is appropriated for it]]

[[ MW: because they should just ... put this stuff online so we don't have to type it in. Mmmm. If people don't change, we have to wait for them to die. That puts a limit on how fast change can go. Paradigms change by the old people dying. ]]

[[CD: Or we interpret them as damage and route around them]]

--

John Smart – Searching for the Big Picture: Systems Theories of Accelerating Change

http://accelerating.org/slides/

[[ MW: "Max More" ... "John Smart" ... are these for real or are they Singularity Porn Star Names? Can we please add to the panel: "Jane Electric", "Alex Alpha", "Peter Power" and "Ellen 2048"? ]]

Story in three acts:

* Framework, picture, painter

Framework of accelerating change:

* Intelligence: Driver of accelerating change

The universe is a physical-computational system.
We exist for information theoretic reasons.
We’re here to evolve and develop.
To create, discover, and manage.
To care, count, and act.
To innovate, plan, profit, and predict.

In a wondrously ordered, elegant, and
surprising environment.

Book - Seth Lloyd - Programming the Universe

[[TÇ: Anybody have an extra Canon SD400/SD300/SD450 battery? Mine just conked. :( ]]
[[MR: is you camera dead now? I can take a few, but not many, due to memory]]

Universe was a geosphere, then a biosphere, next a noosphere

The Earth will be "cephalized" -- get a head (a dense skin of intelligence)

Acceleration studios:

* Moore's Law

[[ MW wants to know where the foresight institute gets its funding. ]]

[[CD If they're a 501(c)3, some of that will be in the public record]]

[[CD:
http://www.foresight.org/about/VPdev.html

Understand the actual and prospective funding communities for the Foresight Institute; develop strategies to secure ongoing significant sources of private/unrestricted funds to support the organization; plan and execute innovative ways of raising increased funding from Foresight members, foundations, corporations, private donors, government agencies and other sources, while nevertheless ensuring that the Foresight Institute maintains its independence of any one funder.

]]

[[ MW: Coming to a Stanford Summit and explaining Moore's Law is like going to a medical conference and trying to explain the germ theory of disease. ]]


Dickerson: Solved protein structures are a Moore's-dependent process

[[ MW: I'd like to know what happens to Moore's law if, say, Avian Flu comes along and kills 800 million people worldwide. ]]

[[ MW: I'd like this guy to make a disprovable statement at some point. This line of talk must work great for picking up early-20s chicks at bars, but I'm not getting his message here. Maybe my mind has officially been blown and I just didn't notice it. ]]

[[ MW: "Smart's Laws of Technology" ... bwaha. what 1960s sf movie did this guy come from? Oh, wait, I get it. He thinks he's Dr Baltar on Battlestar Galactica. ]]

[[ MW: in 1975, I weighed 7 pounds and 4 ounces. In 2005, I weighed 175 pounds. In 2035, I will weigh 1234561 TONS! ]]

[[ MW: I call BS. there was more technological change in the 19th century than there was in the 20th. ]]

[[CD You are on crack -- the key measure for me is how often you have to re-learn something fundamental to function in society -- that's been happening at much more regular intervals in the past 106 years than in the 100 years before]]

[[ MW: Standage's Victorian Internet, http://www.amazon.com/gp/product/0425171698/ got the world used to the idea of instantaneous conversation. We invented electricity, the plane, the car, the dishwasher, the flush toilet, medicine, in the 19th century. in the 20th century we invented HDTV. ]]

[[ CD: Yes, but you didn't need to reinvent your understanding of the world nearly as frequently -- that's like saying that inventing the abacus is the same as inventing the computer]]

[[ MW: That's a futurologist-centric point of view, most humans today don't need to reinvent their view of the world at all during their lifetimes, self-improvement books notwithstanding. ]]

[[ CD If you want to participate in culture, transportation, civil discourse; if you want to raise kids or travel, if you want to renew a driver's license or buy real-estate or get a phone, you need to know things that you didn't know a year before or the year before that ]]

[[ MW: how many rednecks do you know? it's *useful* to know these things, but you can participate in society perfectly well without a computer. what things do i *need* to know today to renew a driver's license that i didn't know last year? the last substantive change to my driver's license was the organ donor checkbox, and that appeared how many years ago? frankly, i don't think the average human gets as future-shocked as a boingboing reader does. speaking of future shock, i'd like to hold a Toffler conference to talk about how he was basically right about everything. ]]

[[TÇ: Metaverse! drink! Attention Economy! drink!]]

[[CD: Arse! Feck! Girls! Drink!]]

[[TÇ: is having fun watching Cory on stage type and seeing it appear here. It's like watching continuous partial attention live. ;)]]

[[ MW: *nominalization* *hype* *somebody else's book cover* ]]

[[ MW: In the 20 years ago, everybody said Strong AI was 20 years away. Today, we're still saying Strong AI is 20 years away. I bet in 2026 we'll be saying The Singularity is 20 years away. Sorry, just being a skeptic. ]]


--

Eliezer Yudkowsky – The Human Importance of the Intelligence Explosion

Term intelligence explosion coined by J Good in 1965

* The smarter you are, the more creativity you can apply to the tasty of making yourself smarter

* Positive feedback cycle yields a super-intelligence


[[ MW: People will use improved intelligence the way they always have: to GET LAID. where do you think our big brains came from? http://www.amazon.com/gp/product/046500802X/ ]]

[[CD That's a very Darwinian view!]]

This doesn't require/imply technological progress on a predictable curve, or AI



"Book smarts" vs cognition:

Intelligence is not being a good "Jeopardy" contestant.

Book smarts =
* Calculus
* Chess
* Good recall of facts


Cognition:
* a bunch of other stuff that non-SAT-high-scorers do just as well

We think of intelligence as a continuum between village idiots to Einstein, instead of from amoebas to humans. All of humans fit in a small slice of that spectrum.

Intelligence is a human super-power that we all share: "Intelligence can't beat a gun" -- where do guns come from?




[[ MW: more interested in measuring adaptation to the present, and adaptability to the future; and also there are coping skills. i have a spectrum that goes

( mental illness ---- low coping ---- high coping )
]]

[[TÇ: Thanks to Paul Schreiber for a freshly charged battery.]]

[[ MW: "intelligence is the most powerful force in the known universe -- you see its effects every day". wank, wank, wank. if we were a bunch of idealists rather than rationalists, we would say that "love is the most powerful force in the known universe". if we were religious types, we would say "god is the most powerful force". we're just worshiping at a different altar. ]]

[[ MW: he just said "Intelligence moves in mysterious ways". ]]

[[ MW: why don't we just look at what really really smart people care about and extrapolate that? ]]

There's lots of knowledge about how intelligence works, but it's scattered across many different fields.

There are lots of things we don't know.

[[TÇ: Is this gazing into the navel of intelligence? ]]

Ignorance about a phenomenon reflects on the observer, not the phenomenon. Confusion exists in the mind, not reality. There are mysterious questions, but not mysterious answers.

[[ MW: unless the answer is schrodinger's equation.... ]]

"(Or google \"Singularity Institute\")"

[[TÇ: Note lowercasing of "google"]]

If your processor ran at 100 mhz, you would need a million processors to run your spreadsheet.
undocumented spaghetti code

The human brain is not user modifiable, alert Cory Doctorow.

[[TÇ: acoustic vibrations are *very* noisy]]

Respect the power of creativity and be careful what you call "impossible"



Fallacy of the giant cheesecake

Major premise: a super-intelligence could create a mile-high cheesecake.

Minor premise: Weak intelligences will be able to create more strongly intelligent versions of themselves.

Conclusion: The future will be full of mile-high cheesecakes! Oh noes!

Making predictions about AIs is a mug's game because there could be lots of different kinds of AIs.

Futurists make more confident predictions about the actions of AIs than they do about the future of actual contemporary humans.

[[ MW: AIs will be like rabbis. ]]


AI isn't a prediction problem; it's an engineering problem

Some people say that the best way to find an answer is to hold off as long as possible. Others say the first solution proposed is likely to be best.

- [ ] Do not propose solutions until the problem has been discussed as Norman R. F. Maier
thoroughly as possible without suggesting any.
- [ ] I have often used this edict with groups I have led - Robyn Dawes
particularly when they face a tough problem, which is when group
members are most apt to propose solutions immediately.
1988 Rational Choice in an Uncertain World: San Diego, CA: HBJ

--

DISCUSSION of Eliezer Yudkowsky's thing



[[ MW: here's why I don't think increasing intelligence means an increase in wisdom. how many really smart obese people do you know? ]]

[[BJS: This is the difference between intelligent "smarts" and wise "smarts". I know many people (increasing all the time who are intelligent smart, but wisdom deficient... they are obese.]]

[[TÇ: It may just be that physical fitness is not a priority for them compared to the other things they choose to spend their time on, even if that means compromising their likely life expectancy (and numerous other factors). I don't necessarily agree with that personally, but there are also people who are at the other extreme as well, working out for hours per day.]]

[[TÇ: The other panelists look so bored, it is very sad. The collective intelligence*man-hours being wasted idling on stage right now is astounding. At least Cory is using his spare attention bandwidth. Ray appears to be taking a few notes. As is Jurvetson. Everybody else is VERY bored looking, arms crossed, wrists limp. ]]

[[ MW: agrees with TÇ. this is why, even if we have super smart minds, what are we going to do with them? they're going to get hax0red and then send increasingly intelligent spam. ]]

[[TÇ: I think this format of one broadcaster, many receivers is inefficient for this many smart people in a room]]

[[MR: What model are you proposing instead?]]

[[TÇ: http://barcamp.org/ inspired by Tim O'Reilly's FOO Camp is *dramatically* more productive/efficient use of that many intelligent people's time. Universities need to evolve, FASTER.]]

[[MR: Universities are actually an example in that if you believe that most learning occurs from student to student.]]

[[ MW: the model is actually Open Space, which has been around for 20 years. http://www.openspaceworld.org/ ]]

[[TÇ: MW, Tim has introduced specific innovations which made key things happen. The *talk* (theory etc.) has been around for 20 years or more, probably invented by the Greeks originally ;). It didn't explode in adoption until BarCamp "open sourced" it and templatized it with an open wiki. Rough consensus and *running* code as it were.]]

[[MR: Well, the advantage to barcamp, is you can always have one, so we could have had one over lunch, or afterwards at 5pm]]

[[ MW: i think we should have one afterwards, but i worry that the downside of an open space format in a field that's this divergent is that we'll spend a lot of time educating the newbies. ]]

[[TÇ: Openspace stuff is mostly theory. Barcamp is mostly practice.]]

[[ MW: I feel like standing up and saying, OK, guys, can we pretend we've heard all the talks and we agree that OMG the future is going to be EMAZING and then move on to some informative content? remember, information is a difference that makes a difference ... and none of this preaching to the choir is making a difference to anyone. ]]

[[ Round table, questions from the audience, panel discussion that uses the speakers as a PANEL and not just as a bunch of audience members who happen to be facing the wrong way and can't see the screen. ]]

[[TÇ: Another better model would be some interaction between the panelists. Would rather have seen 10-15 min. *lightning* talks by presenters in first two hours, and rest of day seeing back and forth discussions - Ray vs. Doug throwdown match etc.]]

[[MR: Someone was talking about a LISP talk, and how the audience would interject discussion, but that clearly is better for smaller groups. Has anyone done a barcamp with 1700 people?

[[TÇ: Or maybe we need to establish more dominant social protocols which encourage the idling panelists to all have their own laptops and have them participate in the #singularitysummit IRC backchannel. Seriously, this is way too low bandwidth for the attention span of most people here (IMHO).]]

[[MR: We were discussing that over lunch, note how many of the panelists brought / are on laptops even in this elite crowd]]

[[TÇ: MR, I think some of them may avoid having laptops due to anti-laptop social biases, and peer pressures, which is quite ironic for a "transhumanist" crowd etc.]]

[[BJS: Or give them all Macs with SubEthaEdit to take notes and comment as we are doing. This commentary during the summit is one reason for doing this and saving the commentary to a browsable file afterwards.]]

[[TÇ: BJS, that too. I think both subetha and IRC have functions. Arggh. Gazing into artificial intelligence navels.]]


[[MR: There's no SubethaEdit solution for Windows, right? We tried a bunch of web apps at MashPit, none were great]]

[[BJS: Not that I'm aware of, but the Bonjour for Windows driver download is only 11MB... mostly due to the size of the installer. Bonjour has been implemented in less than 8K.

MW: http://www.moonedit.com/ amusingly supports only Linux, BSD, and Windows, but not OS X.

[[MR: For all we know there's a concurrent moonedit session, but I consider this very unlikely]]

[[TÇ: Unlikely, because at least one of them would have also showed up in IRC.]]

MW: Also, windows users are tards.

[[MR: So back to the main topic, can you have a 2000 participate bar camp?]]

[[TÇ: I would be happy to observe a 10 person BarCamp of the people on stage. Let's start small and simple (hey, I am into microformats principles in general;) ]]

[[MR: Ok, well we should start forming/using distinctive terminology for different formats. Some exist, but we're mushy with the usage round table discussions, debates...etc. Is a 10 person BarCamp different than a round table discussion?]]



--
Bill McKibben – Being Good Enough

Appearing via "Tele-Portech"; a life-size video well integrated into the backdrop.

[[CD: Note to introducers: don't try to summarize the speaker's points -- if you get it right, the speaker is redundant, if you get it wrong, you're wrong.]]

Doubt will dominate my talk

I'm a voice of a different idea of the future


My doubts are not technical; I'm more than willing to grant for the sake of argument that what Ray and others have been called the Singularity may be in the offing. That we will see an intelligence that is greater than that of the human and human life will be transformed and everything will be transformed, including business and life and even death.

The default assumption in our life is that More Is Better [[ MW: hence Moore's law, ha ha ]] ... more stuff, more whatever.

And by a default assumption that's what I mean.

We call technological prowess "human nature" and that has developed most in our culture, it has produced a political and PR machine that has defied all attempts to restrain it.

[[ MW: when we have supersystems like "the military industrial complex" isn't that a greater-than-human Thing? if so, then isn't Google? ]]

[[CD: This audio is really hard to hear onstage, is it better in the audience?]]

[[BJS: It's audible, but fuzzy.]]

I had 2400 hours of video -- a day in the life of the information age. I settled in an watched.

[[TÇ: Audio is like AM Radio in the third row, but that video image (if you can even call that "video") is surreal. *mind blown*. I took a photo, but it looks so real that it doesn't look impressive. Yes, holographic.]]

[[BJS: Video looks almost *holograhpic* WOW!]]

Audio is vastly different in different spots in the room

When you watch all that video, it's like you're the center of the universe.

[[CD: I can't make this audio out, sorry]]

[[TÇ: MemAud acoustics leave a lot to be desired. I'm surprised they don't have the audio come direct from where the image is or something, and then pump it into the same feed as the other participants so that you get that same live+speakers feeling.]]

[[MR: Audio is great from the balcony, but what's that video technology? I thought that was a person for quite a while]


When we make ourselves smaller, we are closer to who we are. We define ourselves in relation to the natural world. Buddhism. The spiritual world.

The biggest thing in the 20th century was the notion of ecology!

mediate these two poles: hyperindividualism vs community.

maybe more is not always better. advocates of transhumanism will point to human memory as something that can be "more"d. marvin minsky explains why "he doesn't much like how people are now" -- he can only learn 2 bits per second. everything we learn in a lifetime fits on 1 CD.

arguments can be absorbed by osmosis at http://www.billmckibben.com/books.html

michael west answering on immortality -> overpopulation: the answer was clearly to limit new entrants to the human race, not to promote the death of people enjoying the gift of life today.

[[TÇ: won't that (no babies/turnover) result in the static humanity whimper scenario outlined earlier?]]

yes, many SF books have "i was 127 years old when i received my birth license"

[[TÇ: "birth license" is not SciFi in China. it's been around for 10? 20? years]]


If you look at Barry Bond's record breaking attempt at home runs with disdain, then they singularity will be great fun for you.


"compact disc brains"
[[TÇ: good thing they're not DVD brains, or else we'd all be owned by Hollywood and licensed for temporary self-use ;)]]

the average european uses .5 the energy of the average american.

[TÇ: and is the average european .5 as productive as the average american as well? i hate when anti-growth/consumption folks only compare costs/resource-usage, and totally omit any comparisons of output/productivity. such crappy reasoning. totally misleading framing. .5-picture thinking. it's like saying, investment A was only half the price of investment B! (uh, but what was the return? what if investment B brought 4x the return for 2x the cost? what then? duh.)]]


statistics from http://www.grist.org/comments/dispatches/2005/01/25/mckibben/index1.html



people are unhappier now because they have less community now.
http://www.amazon.com/gp/product/0743203046/

[[ MR: Do people's happy/unhappiness re community vary on introversion/extroversion? ]]

people at farmer's markets have 10 times more conversation than at supermarkets.

[[TÇ: Survey must not have included San Francisco Marina Safeway, or Pacific Heights Whole Foods stores - as far as conversations go ;)]]

And [farmer's markets] use 10x less energy to produce their dinner.

MW: For you peak-oilers, according to End of Suburbia, each calorie we eat takes 15 calories to produce (10) and transport (5); and note that commercial fertilizers and pesticides are all made from petroleum! Hence the term "the 3000 mile salad".


"bow to community judgment" - yes, because of course, we want the 68% of folks, who are literal creationists (see CNN etc. surveys), to be the community judges.



--


Kurzweil Follow Up

Doug says I'm describing 10 technologies with 50-50 probabilities, yielding a cumulative probability that's much lower.

That's a mischaracterization -- the components are unreliable, but they backstop one another, like the redundancy of the semiconductors in my laptop.

Any one of the ten approaches will suffice -- it's 999/1000, not 1/1000.

[[CD: My jetlag is killing me]]

Progress is exponential -- our hard problems will get easier. Things that were R&D in 2000 are widely deployed and quite impressive.


We actually hallucinate what we see. Vision is based on 7 different "images" converged with memory to produce what we think we see

IT does have a wall -- it's based on the physics of computation. But even based on known paradigms, the capabilities are vast. Add to that all the new computing paradigms, and you get more capacity.

Utility foglets are not used to build the Taj Mahal, they are used to produce VR environments in the real world.

The Internet is democratizing, and it creates community based on interest rather than accidents of geography. Equalizes doctor-patient relationships.

Kids can now command $1MM recording studios, $10MM film studios, for very little money.

[[ MW: The future .. it's so .. amazing. Damn. It's so much better than the past. Really. By the year 2000 we will have an amazing computer network around the world that is very powerful. And ubiquitous. And globe-spanning. With nuclear weapons. And orbital brain lasers. And Jennifer Lopez. Wait, it's 2006 already. Hmm. ]]

Bridge 2, (from his "immortality" book) is mastering the technology of programming biology.

"You're likely to live longer, the longer you live."

MW: "If you want long life, try to make it to a hundred. Very few people die past that age." -- Woody Allen.


--

Panel Moderated by Steve Jurvetson (VC)

? there are a lot of scientists philosophers
if you could physical copy the brain without the body and physical world, (phantom
if one could upload the brain,

[[ MR: What was the actual question?]]

[[ MW: i'm falling asleep, and the roundtable seems dispirited. i'm going off to look at high resolution photography at the Spring Faire, http://springfaire.stanford.edu/ ]]

Hofstater predicts (uncomfortably) "human-level" AI by 2100. He dislikes the term "strong AI"

Anybody else have numbers to give?

2101. It will look good in hindsight.

MR: 2050

--

DISCUSSION AREA (sort of most recent at top so it's next to current insertion points) -- some discussion from above has been cut and pasted into this area as well.
(I don't think we need brackets in the discussion area, it's obviously not part of the source material)


MR: Who's going to the reception at 7pm?


MW: I feel like asking the question: So, we all know that the first enthusiastic adopter of any technology is the porn industry. How will Porn approach the Singularity?

BJS: Great question!

MW: Hentai that draws itself, i think. interactive porn! autoteledildonics!

TÇ:Would that be HentAI then?

MW:tantek++



MW: I think that when we have superintelligent computers, we'll have many, many, very high bandwidth, but equally long, speeches made by one computer to others who aren't really listening.

MR: That's the overall theme you think would be better, more interactivity?]]


TÇ: For those Oxford debaters out there, I really wish this follow-up was panel-interactive, and not just a "2A/1AR".
MW: let's fast forward to 3


BJS: When I look back on the past from my galaxy spanning, bush-bot body, and wonder why I thought what I was doing then was important


OK, then, what would make you say "Yes, the Singularity Has Happened"?


MR: I thought the Singularity was a point, that was infinitesimally small, that we pass through. You would have to say, Yes, the Singularity occurred.

TÇ: It may be akin to passing through the event horizon of a physical singularity, where it is theorized that you won't notice anything different about crossing that threshold.

MW: Okay, first question. Today, in 2006, what criteria would you set for saying "Yes, the Singularity Is Here"? What about if you were from 1986?
Discuss.

MR: that kinda happens afterwards, where the speakers hang out together, we just generally don't see it unless you go too

MW: i wish we could get the folks on stage to talk amongst themselves rather than at the audience individually. let's do this as an unconference next time, eh. bring on the Open Space. When is the roundtable discussion?

a lot of singularity-related talk still sounds very divergent, like it's in the brainstorming-with-a-beer stage, where we sit around and talk about "wouldn't it be cool if".

Bill McKibben: "pardon me for saying so, but these are the sentiments you might hear in the parking lot after a Phish concert."

MW: u r welcome.

MR: my camera is working again, Thanks to MW

MW to MR: you can grab a 1 gig SD card from me if you need it, I'm upstairs in the front row of the back half, not the front half where you can spit on people.

MR: I'll skip this slides cause they're online right?

MR: OK, I'll grab it next break

MW: i have a blank 1 gig SD card, lemme know here ____ if you want it.
Tantek, do you have blank SD card?

TÇ: Thanks, yes I have plenty of RAM. Battery was the problem

It would be nice if someone could improve the mic/sound

moof.


Meng thinks: this sounds like a "framework" for saying "I'm worried about global warming". people have been worrying about global warming for a long time. ditto nuclear holocaust. or Deep Impact The Movie.

MR: I think people talk about global warming more than anthropogenic risk

BJS: Isn't global warming currently thought of as anthropogenic risk?

MR: well, it should be, i don't think people think of human extinction when they think of melting icebergs...

meng thinks: i propose a definition for artificial intelligence: the ability to convince 1 to N people to do what you want them to, for arbitrary N. The leader of the Roman Republic was called the Imperator for a reason. Passing the Turing Test is only the first step, it doesn't mean much until the AI becomes a Player. In fact, unless you have some effect on my life, you're a nonentity to me, whether you're human or silicon.

Crap. for large N, this means we've got a global computer tyrant, ie. SkyNet. hm.

--

REFERENCES

The Age of Spiritual Machines: When Computers Exceed Human Intelligence by Ray Kurzweil

Robot, Hans Moravec

The Spike: Accelerating into an Unimaginable future, Damien Broderick

--

SCHEDULE

9:00 AM: Welcome by Symbolic Systems

9:05 AM: Intro by Peter Thiel and Tyler Emerson

9:10 AM: Ray Kurzweil – The Singularity: A hard or soft takeoff? Keynote

9:50 AM: Douglas R. Hofstadter – Trying to Muse Rationally about the Singularity Scenario

10:20 AM: Nick Bostrom – Artificial Intelligence and Existential Risks

10:40 AM: Break

10:55 AM: Sebastian Thrun – Toward human-level intelligence in autonomous cars

11:15 AM: Cory Doctorow – Singularity or Dark Age? How the copyright wars threaten technological progress

11:35 AM: K. Eric Drexler – Productive Nanosystems: Toward a super-exponential threshold in physical technology

11:55 AM: Extended Break (Lunch)
12:20 AM: Real break

12:55 PM: Max More – Cognitive and Emotional Singularities: Will super-intelligence come with super-wisdom?

1:15 PM: Christine L. Peterson – Bringing Humanity and the Biosphere through the Singularity

1:35 PM: John Smart – Searching for the Big Picture: Systems Theories of Accelerating Change

1:55 PM: Break

2:10 PM: Eliezer Yudkowsky – The Human Importance of the Intelligence Explosion

2:40 PM: Bill McKibben – Being Good Enough

3:10 PM: Kurzweil Follow Up

3:35 PM: Panel moderated by Steve Jurvetson and Peter Thiel

4:10 PM: Audience Q&A

4:55 PM: Close

7:00 PM: Reception (how private?)
Who's going:

Well, if you made it this far, thanks for your patience :-)