On the Nature of Things | Forum

How long will the first go?

Will it go far enough that we get the second?

I don't understand. What is the Singularity? How would we know we've passed its event horizon, so to speak?

Taboo wrote:I don't understand. What is the Singularity? How would we know we've passed its event horizon, so to speak?

When the production of true statements regarding a particular object exceeds the capacity of a human at current (unaided) levels of cognitive speed.

So with regards to Literature and Science, we have passed the Singularity event horizon? I'm pretty sure that there are more books (and articles) coming out in most subjects each year than a human can actually read.

How narrow must the Object in question be for us to accept the advent of singularity?

yeh this is why i havent really had much of an opinion on this particular subject, many fields have already grown beyond one persons ability to understand them and drawing a line in the sand about when this accelerating complexity gets called something else is probably best done from a few centuries distance.

for example, computer cpu's which are at the heart of this networked billion monkey machine have long since been step-by-step debuggable, the parallel pre emptive threading with out of order execution and caching make them too complex to follow the pathway of a single piece of code, even for the brightest of sparks.

for me, it will be the theory of everything that represents it, when all the seperate fields and mini languages of science can all be represented by a single system that can describe it all.

moores law is hitting the wall with current tech, we are getting as close as you can to atomic size with silicon and their are plenty of articles on that if your interested but the next techs (optical? carbon? organic?) are already being played with so perhaps it will continue at current speed with a few hiccups and pauses along the way.

The Singularity isn't about a field of subject matter becoming broader than an individual's ability to understand it. It's when the computation of a machine exceeds the IQ of about 150, i.e. when a machine has the ability to upgrade itself, its intelligence will increase exponentially.

Moore's law really has nothing to do with it. Because there are other ways to increase computation than increasing the clockspeed of a CPU.

I would argue that the event horizon has already been passed because the internet is an extelligent singularity, human beings are components within the system.

Enki wrote:The Singularity isn't about a field of subject matter becoming broader than an individual's ability to understand it. It's when the computation of a machine exceeds the IQ of about 150, i.e. when a machine has the ability to upgrade itself, its intelligence will increase exponentially.

How do we know how smart a machine must be to be able to successfully upgrade itself? Even mindless systems (like Evolution) can effect change on a working pattern, but how would we (or it) recognize a successful upgraded version of itself from the failures?

Moore's law really has nothing to do with it. Because there are other ways to increase computation than increasing the clockspeed of a CPU.

I would argue that the event horizon has already been passed because the internet is an extelligent singularity, human beings are components within the system.

Or not:
http://www.edge.org/3rd_culture/lanier0 ... index.html

And that is part of the larger pattern of the appeal of a new online collectivism that is nothing less than a resurgence of the idea that the collective is all-wise, that it is desirable to have influence concentrated in a bottleneck that can channel the collective with the most verity and force. This is different from representative democracy, or meritocracy. This idea has had dreadful consequences when thrust upon us from the extreme Right or the extreme Left in various historical periods. The fact that it's now being re-introduced today by prominent technologists and futurists, people who in many cases I know and like, doesn't make it any less dangerous.
...
The collective is more likely to be smart when it isn't defining its own questions, when the goodness of an answer can be evaluated by a simple result (such as a single numeric value,) and when the information system which informs the collective is filtered by a quality control mechanism that relies on individuals to a high degree. Under those circumstances, a collective can be smarter than a person. Break any one of those conditions and the collective becomes unreliable or worse.

Taboo wrote:. . mindless systems (like Evolution) can effect change . .

And how do you know that Evolution is mindless?

Taboo wrote:How do we know how smart a machine must be to be able to successfully upgrade itself? Even mindless systems (like Evolution) can effect change on a working pattern, but how would we (or it) recognize a successful upgraded version of itself from the failures?

It is measurable in terms of computational power.

Moore's law really has nothing to do with it. Because there are other ways to increase computation than increasing the clockspeed of a CPU.

I would argue that the event horizon has already been passed because the internet is an extelligent singularity, human beings are components within the system.

Or not:
http://www.edge.org/3rd_culture/lanier0 ... index.html

This is kind of a straw man based on a sense of idealism. It isn't a matter of whether or not it is IDEAL that the so-called hive be considered the highest form of thought. It is simply a matter of fact that information is shared, and the cognition is not a singular function performed within a computer known as a brain. We share information and as such, we as individuals make decisions based upon that information, but at any given time the individual acts on only partial information.

And that is part of the larger pattern of the appeal of a new online collectivism that is nothing less than a resurgence of the idea that the collective is all-wise, that it is desirable to have influence concentrated in a bottleneck that can channel the collective with the most verity and force. This is different from representative democracy, or meritocracy. This idea has had dreadful consequences when thrust upon us from the extreme Right or the extreme Left in various historical periods. The fact that it's now being re-introduced today by prominent technologists and futurists, people who in many cases I know and like, doesn't make it any less dangerous.

The bolded part is the straw man.

...
The collective is more likely to be smart when it isn't defining its own questions, when the goodness of an answer can be evaluated by a simple result (such as a single numeric value,) and when the information system which informs the collective is filtered by a quality control mechanism that relies on individuals to a high degree. Under those circumstances, a collective can be smarter than a person. Break any one of those conditions and the collective becomes unreliable or worse.

No, this isn't necessarily so. It doesn't matter whether or not the collective is smarter than an individual person. That's not even wrong. The collective is an amalgamation of every person plugged into the collective. Right now there is an economic imperative to plug every human being possible into the collective. This imperative is not being carried out by any individual, but by billions of individuals furiously laying fiber-optic cables and cell towers all over the world. Take any individual out of the system, including Steve Jobs, and the process will march on without them.

We are Borg already.

I would say that Moore's law is useful in the same way that whole numbers are: for example, when the number of transistors in the world exceeds the number of neurons, we have an actual measure of when things have begun to turn. Not that we know that is singularity but that singularity is pretty well inevitable for even the most skeptical at that point.

I was also using M'sL to stand in for similar laws that look at pipes and storage systems.

Enki wrote:

Taboo wrote:How do we know how smart a machine must be to be able to successfully upgrade itself? Even mindless systems (like Evolution) can effect change on a working pattern, but how would we (or it) recognize a successful upgraded version of itself from the failures?

It is measurable in terms of computational power.

It is a commonly held opinion. If you would have asked me 15 years ago, I would have said that computers with terrabytes of storage and 8 Gigahertz would be just about bordering on intelligent, after all, they'd be thousands of times more powerful than the already powerful machines I was using back than.
Now I don't think so anymore. A machine with enormous computing power and terrible software can produce worse than useless results, while a slower machine with a brilliant programmer can solve ridiculously complicated problems.

In other words, who has been more successful: mammoths or cockroaches?

It is simply a matter of fact that information is shared, and the cognition is not a singular function performed within a computer known as a brain. We share information and as such, we as individuals make decisions based upon that information, but at any given time the individual acts on only partial information.

We've been doing that for ages, using things such as markets. We're simply doing the same thing, but a bit faster. Maybe too fast for our own good, since having a million people believe and act on the basis of some half-digested factoid can be worse a million people thinking about the problem on their own.

No, this isn't necessarily so. It doesn't matter whether or not the collective is smarter than an individual person. That's not even wrong.

It does matter. OF course it matters. If the collective result is the basis of decision-making by individuals, it matters a damn lot.

The collective is an amalgamation of every person plugged into the collective. Right now there is an economic imperative to plug every human being possible into the collective. This imperative is not being carried out by any individual, but by billions of individuals furiously laying fiber-optic cables and cell towers all over the world. Take any individual out of the system, including Steve Jobs, and the process will march on without them.

I don't understand what you mean by that. Perhaps you can explain in a bit more depth. I mean, I use the internet, but merely because it saves me time. I never feel any "collective" or such. Just a bunch of relatively dumb programs, performing very simple tasks. In fact, I found that using the internet less actually makes me more productive and even happier with myself, not less.

It's not persons you plug in, but shallow and superficial avatars. The "collective" seems to be a world dominated by trolls, lolcat designers, mentally unstable schizophrenics and pornographers. It's the triumph of the loudest and most persistent, of the lowest common denominator, the most base aspects that often come through. If ever in doubt, just read some Youtube comments.

Demon of Undoing wrote:We are Borg already.

Speak for yourself, tin-face. I don't even have piercings.

Milo wrote:I would say that Moore's law is useful in the same way that whole numbers are: for example, when the number of transistors in the world exceeds the number of neurons, we have an actual measure of when things have begun to turn. Not that we know that is singularity but that singularity is pretty well inevitable for even the most skeptical at that point.

Creating useful human-level AIs requires 3 things:
Hardware capability (what you speak of)
Software design (or evolution?) to create an artificial consciousness - a dark mystery for the foreseeable future.
Capability to control said AI over time (i.e. making sure it keeps providing true and useful answers to our querries) - again uncertain, given the enormous difficulties our program designers have controlling current software, nevermind software millions or trillions of times more complex.

Taboo wrote:It is a commonly held opinion. If you would have asked me 15 years ago, I would have said that computers with terrabytes of storage and 8 Gigahertz would be just about bordering on intelligent, after all, they'd be thousands of times more powerful than the already powerful machines I was using back than.

You miss the point. There is an empirical observation. If the machine starts upgrading itself then it is proven that it has achieved the ability to upgrade itself. It really isn't about the clock speed, as I said before, it's about computational power and how that power is used.

Now I don't think so anymore. A machine with enormous computing power and terrible software can produce worse than useless results, while a slower machine with a brilliant programmer can solve ridiculously complicated problems.

Sure, the Singularity is not a notion of hardware by itself.

In other words, who has been more successful: mammoths or cockroaches?

Whichever one does a better job of upgrading themselves I imagine.

We've been doing that for ages, using things such as markets. We're simply doing the same thing, but a bit faster. Maybe too fast for our own good, since having a million people believe and act on the basis of some half-digested factoid can be worse a million people thinking about the problem on their own.

Yes, and the process of building the machine has been going throughout that time.

It does matter. OF course it matters. If the collective result is the basis of decision-making by individuals, it matters a damn lot.

Except, I am not talking about collective decision making. I am talking about collective process sharing. But our decisions are made somewhat collectively. Look at how we use polling data. Polling data is as prescriptive as it is descriptive and people will make electoral choices, market decisions, etc... based upon the polling data.

I don't understand what you mean by that. Perhaps you can explain in a bit more depth. I mean, I use the internet, but merely because it saves me time. I never feel any "collective" or such. Just a bunch of relatively dumb programs, performing very simple tasks. In fact, I found that using the internet less actually makes me more productive and even happier with myself, not less.

You are participating in the collective right now. As you read this you are thinking MY thoughts.

It's not persons you plug in, but shallow and superficial avatars. The "collective" seems to be a world dominated by trolls, lolcat designers, mentally unstable schizophrenics and pornographers. It's the triumph of the loudest and most persistent, of the lowest common denominator, the most base aspects that often come through. If ever in doubt, just read some Youtube comments.

Depends on who you are talking to doesn't it?

Taboo wrote:

Milo wrote:I would say that Moore's law is useful in the same way that whole numbers are: for example, when the number of transistors in the world exceeds the number of neurons, we have an actual measure of when things have begun to turn. Not that we know that is singularity but that singularity is pretty well inevitable for even the most skeptical at that point.

Creating useful human-level AIs requires 3 things:
Hardware capability (what you speak of)
Software design (or evolution?) to create an artificial consciousness - a dark mystery for the foreseeable future.
Capability to control said AI over time (i.e. making sure it keeps providing true and useful answers to our querries) - again uncertain, given the enormous difficulties our program designers have controlling current software, nevermind software millions or trillions of times more complex.

I think that we will continue to make very little progress on the design side of this. But I also think that we are able to ratchet up the fast idiots at this geometric pace without limit.

In other words, I think that we will realize some grotesquely overbuilt solution initially and then reverse engineer our way into doing it smart; think ancient bridges compared to modern.

Milo wrote:

Taboo wrote:

Milo wrote:I would say that Moore's law is useful in the same way that whole numbers are: for example, when the number of transistors in the world exceeds the number of neurons, we have an actual measure of when things have begun to turn. Not that we know that is singularity but that singularity is pretty well inevitable for even the most skeptical at that point.

Creating useful human-level AIs requires 3 things:
Hardware capability (what you speak of)
Software design (or evolution?) to create an artificial consciousness - a dark mystery for the foreseeable future.
Capability to control said AI over time (i.e. making sure it keeps providing true and useful answers to our querries) - again uncertain, given the enormous difficulties our program designers have controlling current software, nevermind software millions or trillions of times more complex.

I think that we will continue to make very little progress on the design side of this. But I also think that we are able to ratchet up the fast idiots at this geometric pace without limit.

In other words, I think that we will realize some grotesquely overbuilt solution initially and then reverse engineer our way into doing it smart; think ancient bridges compared to modern.

A singularity need not be conscious it needs:

1) An imperative to upgrade itself
2) Paramaters for upgrading itself
3) The external ability to upgrade itself

It is possible for a singularity to be a completely dumb machine that is able to acquire the resources it needs to upgrade itself. First by ordering the parts from Amazon, and later on by mining the materials itself. It need not be conscious at all, it is in fact theoretically possible for it to just consume all matter and resources it can acquire until it reaches some sort of hard limit.

I think physical reality pretty much fits the definition of Singularity.

Parodite wrote:I think physical reality pretty much fits the definition of Singularity.

Yes, but a technological singularity is a particular configuration of it. The notion is that it consumes everything that it can possibly come in contact with, kind of like a black hole.

In the Singularity is Near he postulates that the Singularity will emerge, beginning to upgrade itself and its capacity to upgrade itself increases exponentially and doesn't stop until it has consumed all matter in the universe.

Enki wrote:

Parodite wrote:I think physical reality pretty much fits the definition of Singularity.

Yes, but a technological singularity is a particular configuration of it. The notion is that it consumes everything that it can possibly come in contact with, kind of like a black hole.

In the Singularity is Near he postulates that the Singularity will emerge, beginning to upgrade itself and its capacity to upgrade itself increases exponentially and doesn't stop until it has consumed all matter in the universe.

I was thinking entropy doing just that. Entropyticus Rex is among us! It squeezes and consumes the universe into a dead cold and infinitly thin sheet of lavender paper.

Parodite wrote:

Enki wrote:

Parodite wrote:I think physical reality pretty much fits the definition of Singularity.

Yes, but a technological singularity is a particular configuration of it. The notion is that it consumes everything that it can possibly come in contact with, kind of like a black hole.

In the Singularity is Near he postulates that the Singularity will emerge, beginning to upgrade itself and its capacity to upgrade itself increases exponentially and doesn't stop until it has consumed all matter in the universe.

I was thinking entropy doing just that. Entropyticus Rex is among us! It squeezes and consumes the universe into a dead cold and infinitly thin sheet of lavender paper.

Actually some think that the singularity will solve entropy!

http://goo.gl/2nEn

Milo wrote:

Parodite wrote:

Enki wrote:

Parodite wrote:I think physical reality pretty much fits the definition of Singularity.

Yes, but a technological singularity is a particular configuration of it. The notion is that it consumes everything that it can possibly come in contact with, kind of like a black hole.

In the Singularity is Near he postulates that the Singularity will emerge, beginning to upgrade itself and its capacity to upgrade itself increases exponentially and doesn't stop until it has consumed all matter in the universe.

I was thinking entropy doing just that. Entropyticus Rex is among us! It squeezes and consumes the universe into a dead cold and infinitly thin sheet of lavender paper.

Actually some think that the singularity will solve entropy!

http://goo.gl/2nEn

Life solves entropy.

Enki wrote:Life solves entropy.

Maybe the self-consuming cycle never ends..

Parodite wrote:

Enki wrote:Life solves entropy.

Maybe the self-consuming cycle never ends..

That's what I think.

Life actually speeds up entropy, decreasing it locally at the cost of increasing it globally.

Taboo wrote:Life actually speeds up entropy, decreasing it locally at the cost of increasing it globally.

But with consciousness with have the capacity to re-order how the matter and energy is used. Perhaps one day we'll be able to cause a star. Then what?

Enki wrote:

Taboo wrote:Life actually speeds up entropy, decreasing it locally at the cost of increasing it globally.

But with consciousness with have the capacity to re-order how the matter and energy is used. Perhaps one day we'll be able to cause a star. Then what?

Then we'll have hugely increased entropy.