Placeholder Image

字幕列表 影片播放

  • MALE SPEAKER: Hello, hi.

  • Welcome to Tech Talk by Kevin Kelly.

  • This talk is going to be available on Google Video.

  • And so we ask that you hold questions that are proprietary

  • at all to Google until the end of the talk.

  • So Kevin Kelly is well known around the world as the

  • editor-in-chief of the Whole Earth Review, and is the

  • author of the book Out of Control: The New Biology of

  • Machines, Social Systems, and the Economic World.

  • Today Mr. Kelly shares his ideas on the future of science

  • and the scientific method.

  • He's come to the right place because Google has an

  • important role to play in the future of science.

  • Join me with a warm welcome for Mr. Kevin Kelly.

  • [APPLAUSE]

  • KEVIN KELLY: Thank you.

  • It's a real pleasure being here.

  • In addition to being editor and publisher of an obscure

  • magazine called the Whole Earth Review, I was also one

  • of the co-founders of Wired Magazine.

  • And for the five or ten years that Wired was the center of

  • the universe, it really felt great.

  • And being here at Google feels like that again.

  • It feels like I'm at the center of the universe again.

  • And I use that in a kind of metaphorical way, but also in

  • a very real sense.

  • I define the center of the universe as that place where

  • there's the least resistance to new ideas.

  • And I feel that, in some ways, this is the

  • center of the universe.

  • And it's a real privilege to be talking at a place where

  • there's probably the least resistance to new great ideas.

  • And what I want to spend a few minutes this afternoon talking

  • about is something that I think is very important but

  • gets very little attention.

  • And that is the nature of the scientific method.

  • And I'll start there and I kind of end up at some cosmic

  • level, and hope you kind of go along for the journey.

  • I think that it may have some bearing to what you're doing

  • in the long-term.

  • So in addition to also co-founding Wired Magazine and

  • being editor for many years, I also am in one of the small

  • group of people who have been pioneering long-term thinking.

  • We have a foundation called the Long Now Foundation.

  • Among other things we're trying to build a clock that

  • ticks for 10,000 years.

  • And the idea of that clock is to serve as an icon in

  • encouragement to think long-term.

  • So what I'm trying to do a bit in this talk is also taking a

  • very long-term view of things.

  • Rather than just thinking of the last five years, the next

  • five years, I'm trying to stretch our perspective and go

  • back 1,000 years and maybe look forward to 50 years.

  • That's kind of asymmetrical, but it's the best I can do in

  • terms of the future.

  • So what's interesting about science is that science is the

  • only thing that generates news.

  • Science is really the only news.

  • If you pick up a paper, most of the stuff that's happening

  • is not really news.

  • It's just kind of repetition, a little of recycling.

  • But science is the thing that generates real news.

  • What's interesting about the news that it generates is that

  • it's often novel.

  • And when we tend to think of the long-term history of say

  • the scientific method or science, we tend to think of

  • the inventions that science generates.

  • And as an exercise, we sent a questionnaire around to a

  • bunch of scientists around the world that said, what's the

  • most important scientific inventions in the last 1,000

  • years or the last 2,000.

  • And you come up with a list. And you can make your own

  • list. But some of the nominations for the most

  • important scientific inventions or the most

  • important inventions over the last couple

  • thousand years was hay.

  • The hay was actually the most important.

  • This was actually Freeman Dyson's idea, that it actually

  • encouraged cultivation of domesticated animals, which

  • led to protein, which lead to a longer life span, et cetera,

  • civilization.

  • So that's one.

  • Another one is antibiotics.

  • That was a really interesting, great, and important invention

  • in the history of the world.

  • Paper, who can argue against paper?

  • And, of course, this auxiliary technology of printing, that

  • was obviously a very important thing.

  • The rudder allowed navigation of the world, and sense of the

  • whole world and encouraged trade.

  • So that was also very important.

  • Electricity, obviously another very interesting things.

  • Once we harnessed the electrons, all kinds of other

  • great and wonderful things happened.

  • What's important about that is that you can make your own

  • list, and that these are different inventions.

  • But I'm actually not really concerned about that because

  • science makes news, but we very rarely look at science

  • itself, the method by which these things come about.

  • And that's what I'm really trying to do.

  • So if you actually suggest to someone that the method that

  • we know of as science today if still changing, and that, in

  • fact, the scientific method may, in fact, be very

  • different 50 years from now than it is today, this is the

  • kind of reaction you normally get.

  • Because when we talk about the long-term trend of the

  • scientific method, I find that people are very puzzled.

  • Science changing?

  • Science will be different in 50?

  • The scientific method?

  • And I think it will.

  • And so what I want to do is kind of take a quick tour

  • through the evolution of the scientific method in the past.

  • And interesting, when I went to look at this, I found that

  • basically there is no literature on this.

  • There is no book about the history of

  • the scientific method.

  • And for something so fundamentally you

  • would expect that.

  • But there isn't.

  • So I sort of cobbled these together and I'm going to go

  • through very quickly just to give you a sense of some of

  • the things I think are important.

  • And one of the patterns that you'll see, I hope, is that

  • this is a lot about information.

  • We have the idea of indexing and cataloguing.

  • Books have been around for a long time.

  • And then you have this idea of indexing or cataloguing the

  • information.

  • Or you have a book that is the index to the other books.

  • And so actually some of the great inventions in the

  • history of scientific method was the idea of an

  • alphabetical index, was the idea of having an index in the

  • book, was the idea of having a catalog to all the books in

  • your library.

  • Then there was the collaborative encyclopedia

  • where you had more than one author, more than one expert

  • coming together, sharing information, and in such a way

  • that it was all catalogued and indexed.

  • There was, later on, several hundred years later, there was

  • the first laboratories where they begin to do experiments

  • actually trying to observe, and measure, and record, those

  • observations.

  • We had the invention of observational tools like the

  • telescope and a microscope which greatly enlarged the

  • scientific method because I, along with some other people,

  • believe that the tools of science actually move science

  • more than any other discoveries of science, that,

  • in fact, the way in which you have progress is by the

  • creation and progressive improvement of tools.

  • Then we had the controlled experiment, which was of

  • Francis Bacon's idea that one needed to have controls.

  • To do an experiment you have controls.

  • There was Society of Experts that began to share the

  • information and the observations

  • that we were making.

  • So they began to actually have kind of a peerage.

  • And then there was--

  • oops.

  • They got cut off.

  • I had a problem here.

  • This is [? Bolton's ?]

  • idea that real science required they be able to

  • repeat an experiment, that someone else should also be

  • able to do it at the same time that you are.

  • Peer review where they began to actually send out their

  • observations and have them peer reviewed by people who

  • knew about it and could either give them credibility or

  • verification.

  • And, then this is Newton hypothesis/prediction, which

  • actually he was making a hypothesis about something and

  • making a prediction that you should find

  • data at some point.

  • Falsifiable testability, Popper's idea that what really

  • counts in terms of a prediction is whether it can

  • be falsified, that that's the sort of criteria by which you

  • judge a hypothesis.

  • Then Fisher and the randomized design, he was involved with

  • statistical analysis, bringing statistics into the design of

  • an experiment.

  • Placebo, somebody actually had to event that.

  • That was not an obvious idea.

  • It took many years.

  • It was actually fairly late in coming, 1937, before the first

  • placebo experiment was done.

  • And computer simulations--

  • I'll talk to a bit more about that-- began very early.

  • Almost as soon as computers were invented they begin to do

  • simulations.

  • And then the double blind refinement, where you actually

  • have both the observer and the patient not know about the

  • experiment.

  • And finally, I think, a very landmark moment was where

  • science began to study itself, where the scientific method

  • became a subject worthy of the scientific approach.

  • So one of the things that's interesting about this is that

  • I began to think about, what about China?

  • Now the thing about China is that it did

  • not discover science.

  • But if you take a list of all the technological inventions

  • that China made, it's phenomenal and actually very

  • unknown in this culture.

  • Most of the technologies that China discovered, they

  • discovered not just a few years before the West, but

  • often, on average, maybe 500 years to 1,000 years before

  • the West discovered it.

  • So it was independent, and yet they did

  • not originate science.

  • And what was missing?

  • Here are some examples of the things that actually were

  • discovered in China: paper, printing, gun powder, compass,

  • the rudder, the stirrup, row crops, iron plow, wheel

  • barrow, cast iron, vaccinations, the chain drive,

  • suspension bridge.

  • They were using petroleum and natural gas as fuel, again,

  • thousands, at least hundreds of years before the

  • West came to this.

  • And actually going back to Francis Bacon, Francis Bacon,

  • when he began the scientific method, believed that four

  • inventions had more impact on science at his time-- this is

  • around the 1600s--

  • than anything else.

  • He said these four inventions which were paper, printing,

  • gun powder, and the compass, he said transformed our world

  • far more than any religious belief, far more than any

  • political or war worrying, conquering nation.

  • That those four inventions which, he said, whose origins

  • are unknown were the transforming

  • powers in his world.

  • And he died before he discovered that all four of

  • those inventions came from China.

  • Now so why didn't the Chinese invent science?

  • And there is actually no easy answer this.

  • A man named Joseph Needham devoted his entire life to

  • trying to answer this question, and it's now called

  • the Needham question, which is how could science and

  • technology which was so far advanced in China for so long,

  • how come China did not continue going that way and

  • actually come up with the scientific method?

  • And the answers, as I said, are very complicated.

  • There's numbers of strands to it.

  • Part of it has to do with an inability to separate the

  • political and the inquiry, the religious and the inquiry, and

  • a sense of maybe valuing other things besides a quest for

  • truth as a proper form of investigation, all those kinds

  • of wrap together.

  • What happened was that there was one discovery that China

  • didn't make.

  • And that discovery was the

  • scientific method of discovery.

  • You can go a long way with creating novelties, actually

  • having progress by developing new ideas without the

  • scientific method.

  • But the scientific method is the best way to do discovery.

  • And once you make that discovery, you're off on a

  • different course.

  • I tend to think of science as a structure of information

  • that allows discovery, in fact, as a

  • structure of discovery.

  • And this is just a kind of a mapping of

  • the citation indexing.

  • This is a mapping of what they used to call in library

  • science citation indexing, which we would now think of

  • almost as a type of page ranking

  • of scientific discovery.

  • So they would look at the whole of all of the citations

  • that were referenced in the footnotes of the paper.

  • And if you extract out a map of those backward links, you

  • would get something like a clustering of those journals,

  • or those individuals, who were cited the most, basically that

  • were linked to the most. And this work was done by Eugene

  • Garfield in the '50s in Philadelphia, and it was

  • somewhat the inspiration that Larry and Sergey looked at for

  • their page rank.

  • And if you take this structure of scientific discovery just

  • as an example, and think about it as a way of structuring

  • knowledge and information as a way of knowing, that's the

  • view that I want to bring into what science is.

  • It's a way of knowing.

  • More importantly, it's actually a way that we change

  • how we know.

  • And I hope to return to that idea that, in fact, it's not

  • just the process of discovery, it's the process of how we

  • change how we discover.

  • Because there's a second order of delta in science.

  • It's not just studying things, learning new things, it's

  • changing how we learn new things.

  • It's learning how we learn new.

  • It has that recursive sense in it.

  • And I think that's really crucial because it's that

  • recursiveness that actually gives us generative power.

  • Almost all the things that we find interesting int he world

  • have a sort of paradoxically recursive

  • nature at the bottom.

  • They are self-organizing.

  • Their self-referential.

  • When you come right down to it, they make this sort of

  • paradoxical pointing back to themselves.

  • So the question is what's the next 150 years of science, I

  • mean what is it really?

  • What is it really?

  • What people really want to know is when are we going to

  • have flying cars?

  • They're not interested in the science and information.

  • They want to know when we're going to have flying cars?

  • When are we going to have robots that talk back to us,

  • that can tell us they're smarter than we are?

  • And when are we going to have virtual reality and things

  • where we can just interface by thinking or by gesturing?

  • Well, I'm not going to tell you about those because I

  • don't know.

  • My point is that those are just individual novelties.

  • Those are just examples, like the Chinese, of just making

  • new things, like hay or penicillin.

  • They aren't really about the structure of

  • how we discover things.

  • And so what I want to try to talk about is some

  • speculations on the ways in which the scientific method

  • itself may change, and not so much the particular ideas or

  • inventions that it might throw up.

  • The first set, well first of all, I'm suggesting that

  • science will actually change more in the next 50 years than

  • it has in the past 400 years.

  • If we know anything about the kinds of trends and curves

  • that we see, then the actual structure of the scientific

  • method will change more in the next 50 years than

  • the last 400 years.

  • And if you believe that, then you have to go along with me

  • in trying to think about ways in which that might happen.

  • And I think-- and basically this will be the context of

  • what I'm starting to suggest--

  • most this will be in the realm in which you all work, which

  • is in the structuring of information and knowledge.

  • That's what this is going to be about.

  • So I might say that Google is about changing the structure

  • of the scientific method in t he next 50 years.

  • So two, one thing that we know about is it's going to be a

  • bio century.

  • And here let me explain what I mean by that.

  • Right now, biology this year, has the most funding, the most

  • scientists, is generating the most results that are

  • published, and has the most economic value.

  • It's the most ethically important or relevant in terms

  • of the kinds of questions that it generates, and it has the

  • most to learn.

  • Let me explain that last one a little bit.

  • The living world, the biological world, basically

  • has had four billion years of learning.

  • Every day there's information being generated by biological

  • systems that's being recorded in genomes of these organisms

  • throughout the world, not just the single ones, of course,

  • but even the individual ones.

  • It's being embedded into the very ecological systems that

  • we have. It's a vast amount of knowledge and information.

  • If you compare that to physics, physics is the same.

  • It's pretty universal.

  • There are deep mysteries about the physics and understanding

  • of what's underneath the hood in our universe.

  • But the amount of information in physics is

  • actually very small.

  • The amount of information that's embedded

  • in biology is huge.

  • And that's one of the reasons why this is at it's point

  • right now where it's become the biggest science that we

  • have, and will continue to be at least

  • for the next 50 years.

  • There is so much to learn.

  • It's so deep.

  • It's so structured.

  • It's so complex.

  • And so the biggest mother lode of information and data in the

  • next 50 in science in terms of

  • understanding it is in biology.

  • Third, computers are leaving the third way of science.

  • And I'll take a few minutes to tell you what I mean

  • by the third way.

  • Traditional understanding of science has two parts.

  • There was hypothesis a measurement,

  • observation and theory.

  • Those two went hand in hand, and you kind of tried to keep

  • them in balance.

  • You might have a weird idea, and you go out looking for

  • data that was OK, or you might start with the data and try to

  • come up with a theory.

  • But the two of them together were kind of like the two feet

  • of science.

  • And this is how the normal understanding of science

  • worked, that you needed to have theories that predict,

  • make hypotheses that would predict data that you would

  • find, and you had to have theories that were based in

  • some ways that would explain the data that you had.

  • And you try to let neither one outrun the other.

  • But had a kind of a two-faced aspect of science that

  • required both.

  • So in the measurement side, what's happened is, is that

  • nothing, really, nothing that I can think of, nothing that

  • anyone else has been able to think of, is growing faster on

  • this planet than information.

  • It is the fastest growing entity on this planet.

  • Hal Varian, whom you're familiar with, did a study of

  • how much information in the world.

  • And he and Peter calculated that it was

  • growing at 66% a year.

  • Physical production, there may be spikes.

  • The iPods may grow at the rate of 200% for a couple of years.

  • But if we talk about anything on the terms of a decade or

  • longer there is nothing that we're making that is anywhere

  • near the growth rates of information.

  • So information, right now, is the largest growing thing on

  • this planet.

  • And if you take a chart of of the kind of data volume over

  • time, in the beginning most of the advances were coming

  • through increased precision.

  • And then after awhile we had increase in the spectrum in

  • which we could measure things.

  • We had in the beginning just what we could see.

  • Then there were things that we could see through a microscope

  • or telescope.

  • Then we increased the sources.

  • We could measure things through a thermometer.

  • We could measure other things that we couldn't see.

  • And then, over time, we had durations in which we could

  • measure them all the time.

  • For instance, we could measure them longer periods,

  • constantly, day and night for years.

  • And what's happening now, I think, is of course we're

  • adding these technological senses around the world.

  • So we're basically generating huge, huge volumes of

  • information all the time, in real time, constantly,

  • everywhere around the earth.

  • So I call this zillionics.

  • Zillionics is the field in which you're dealing with

  • zillions of things.

  • Once you get into zillions a lot of the tools that we have

  • break down, and we're trying to invent ways in which to

  • deal with [? zillion ?]

  • And science, of course, is generating

  • zillions of bits of data.

  • These are just some, actually at this

  • point, kind of old amounts.

  • But terabytes, petabytes, exabytes, we're headed there

  • very fast, zillionicbytes at some point or other.

  • On the other hand if I hypothesize of how we learn

  • things, one of the things that's happening right now,

  • right now this is done kind of with human minds, very, very

  • handcrafted, but, in fact, we now have ways to do a one type

  • of science that's called multiple competing hypothesis.

  • Instead of trying to go through, I have a hypothesis

  • about what may happen.

  • I'll try that out, try some things.

  • If it doesn't work, OK, I'll make another one.

  • You actually make multiple simultaneous hypotheses in

  • which you try to apply data.

  • So hypotheses become something that is becoming kind of a

  • mass phenomena rather than just a single

  • little thing you handcraft.

  • And there are ways in which you can actually try and

  • manage multiple hypotheses as a way to deal

  • with scientific discovery.

  • Another way is combinatorial sweep.

  • And this is what Stephen Wolfram did.

  • OK, the way you explore something is we'll generate

  • every possible variation of it, and we'll explore through

  • that space.

  • And by exploring that space of possibility we'll know

  • something about what that thing is.

  • So he takes the CA, the cellular automata, and they'll

  • make every possible CA and make that whole space of CA,

  • and then go through that trying to discern the

  • characteristics and the behavior of the CA by

  • exploring that possibility space.

  • Well, we can do the same thing with other things like with

  • combinational libraries and exhaustive search.

  • They're using it in biology right now.

  • They'll make every possible variation of a protein.

  • They'll make every possible variation of a ceramic or

  • chemical compound.

  • You just explore the possibility space.

  • And again, using robots to make up them, to test them,

  • this is something that was not possible before.

  • But we're basically exploring this possibility space of

  • hypothesis by doing it exhaustively in combinatorial

  • exponential expansion.

  • So chemistry compounds, synthesis methods, you make

  • something new.

  • You make every possible way of making it new.

  • And you also generate hypotheses in the same, every

  • possible hypothesis that could apply to the situation.

  • Those are the two standard pillars of science,

  • observation, hypothesis, theory.

  • But there's a third one, and that's what I

  • want to talk about.

  • The third one is a simulation or synthesis of those two.

  • You synthesize something.

  • I call it "nerd culture," and I did a little piece in

  • Science Magazine talking about the usual division of cultures

  • that C. P. Snow talked about, the two cultures, humanists

  • and humanities on one side and the scientists

  • on the other side.

  • Humanists would explore the human condition by creating,

  • by kind of probing an expression, by, I guess you

  • might say, examining the human condition.

  • And the scientists would do the same thing by

  • measuring or probing.

  • The third way of knowledge is the nerd way.

  • And the nerd way is that you make something.

  • So the way you study democracy is not by getting in and

  • reading a lot of books or by trying to measure it, the way

  • you study democracies is you make an artificial one.

  • You make a virtual democracy.

  • The way you study a mind is not to contemplate it, to

  • examine it from all sides, or to try and measure the little

  • data points in the brain cell.

  • The way you study a mind is you make an artificial mind.

  • That is the nerd way.

  • It's a way of doing things, studying things, and

  • discovering things by creating them.

  • Well, simulations has a little bit of that in it.

  • It generates a lot of data.

  • Here's a gamma ray simulation.

  • This is huge spinning off huge amounts of data.

  • And so what this suggests to us is that in the data volume

  • that most of the data in the coming years is going to be

  • generated by our simulations of things.

  • Actually simulating it into things that we make is

  • actually going to generate more than the natural world.

  • Their efforts in science, science itself, will be

  • generating more data than the natural world measurements do.

  • And so these three, I think science now has a triad.

  • It's now a tripod of three things.

  • It's data, measurement, and simulation.

  • And those three things together are becoming the kind

  • of core of all sciences.

  • So you have data, which is being fed into the simulation,

  • and the simulation is creating new data itself.

  • And then you have the simulation which also is kind

  • of a theory.

  • Basically simulation is a theory that's active.

  • It's an interactive theory.

  • So theory and simulation has become something that feed

  • upon each other.

  • And, of course, hypothesis and data measurement is the same.

  • And I think that--

  • the intersection of those three is

  • what I call deep science.

  • This is this new science that has increased data.

  • We can see this happening three different--

  • [UNINTELLIGIBLE] complex adaptive science where you

  • have continuous real time measurement, which is also

  • being simulated in real time feeding back to the real time

  • measurements.

  • The simulation itself is creating a continuous search

  • in real time for our hypothesis which is going on,

  • and the hypothesis, of course, guiding the creation and

  • measurement of data.

  • You can imagine the same thing happening in, say, health

  • sciences where you have a person.

  • You have sensors in the person and they're generating in real

  • time all kinds of information which is being

  • simulated in real time.

  • And that simulation is, of course, itself is feeding into

  • the hypothesis.

  • And so the three of them working together becomes this

  • new science of simulation, hypothesis, and measurement.

  • And, of course, we can imagine it happening say in the

  • ecological natural world where we're looking at the real data

  • from nature.

  • We're simulating it at the same time in real time.

  • We're generating real time hypotheses which are, again,

  • guiding our measurements and our simulations.

  • This is deep science.

  • So this is the third way of science where the computer

  • becomes very involved in our method of discovery.

  • It's the third way of science, or not.

  • It's scary because it has computers at the center, and

  • we are always careful about that.

  • So the last thing I want to talk about science is that it

  • actually will be creating a new way of knowing.

  • WikiScience is one possibility.

  • The editors of Nature told me that they're expected to get

  • their first 1,000 author paper the summer.

  • So that's in traditional science.

  • We can imagine WikiScience where there is an ongoing

  • document and an ongoing scientific journal article

  • that's never done.

  • It's constantly being updated.

  • It has thousands of contributors around the world.

  • It's open-ended.

  • And it's constantly in flux.

  • That's the nature of adaptive knowledge.

  • Another suggestion is compiled negative results.

  • Right now negative results are thrown away for the most part.

  • There is one obscure journal that actually has just started

  • to try and report negative results.

  • But negative results can be far more informative than the

  • positive results.

  • Normally they're thrown away.

  • They're not revealed.

  • And there's actually a move right now to require in

  • medical studies to require that the

  • negative results be reported.

  • And the way they're doing that is they're saying that the

  • journals will not published the medical article of your

  • final results unless you register with them to indicate

  • that you've done the early negative results.

  • So they're hinging the final results of your study being

  • published on the fact that you actually are going to

  • catalogue and report your negative results.

  • AI computer proofs is another example.

  • This is a packing hypothesis.

  • I think it was Kepler who did it, suggested at first. It was

  • not proved until recently, and it required AI help to

  • actually make the mathematical proof.

  • And we're going to see more and more of that as well.

  • The triple blind emergent trials, this is the idea that

  • you can actually do scientific studies ad hoc by

  • taking real time data.

  • Imagine if you have a large population of people.

  • You have 24-hour sensory information from their body,

  • temperature, all kinds of things.

  • And you basically extract out of that large amount of data

  • afterwards.

  • You sort out the controls afterwards so that neither the

  • observer, nor the researcher, nor the scientists actually is

  • aware that there's an experiment going on.

  • The experiment happens after the fact but taking large

  • numbers of variables and extracting out the ones that

  • you want to use and consider as controls.

  • So finally distributed experiments is another example

  • of [UNINTELLIGIBLE].

  • Other examples where we're taking very small amounts of

  • information and measurements distributed over time, not

  • centralized, and using those measurements to actually

  • correlate and aggregate them.

  • And that's another way.

  • And then the final way is the return of the subjective.

  • This has to do with science as long-term trend in terms of

  • becoming objective and refusing to consider the

  • subjective.

  • When we get down to fundamental questions like the

  • origin of the universe and things like this, and quantum

  • mechanics, we begin to find in the fact you cannot easily

  • remove the observer, and that you actually have to account

  • for the observer as part of the experiment.

  • And we don't have very good tools for dealing

  • with that right now.

  • And so what we're going to do in the next 50 years is also

  • learn how to tolerate and manage the

  • subjective in science.

  • So those are some speculations on the

  • next 50 years of science.

  • But I'd like to end by talking about just one last thing,

  • which is what I think science means.

  • I think sciences actually creates a

  • new level of meaning.

  • And one of the interesting things about emergent systems

  • is actually what's happening is that a new level meaning

  • comes out of the small parts that were at the lower level.

  • And so if you could go back to my diagram of the scientific

  • clustering of information and citations, we can also imagine

  • this web of the Internet, which is one of the more

  • current versions of the traffic on the Internet, the

  • different countries are different colors.

  • I think North American is blue.

  • This is a recursive map.

  • This is a map where things are touching back to itself.

  • This is a structure of information.

  • we can imagine each of those nodes on there as being

  • different species.

  • We can imagine them being different technologies.

  • We can imagine them being different methods of

  • investigation, different styles of discovery, and we

  • can also imagine this as one machine.

  • And so what I'm suggesting is, is that we have

  • science right now.

  • The way we're hooking up stuff is that it's actually one very

  • large machine.

  • And I decided to actually treat this as one machine.

  • So I said if this was really one machine, what

  • would it look like.

  • I'm sorry for this bad formatting.

  • But it has a billion PC chips.

  • This machine is right now sending one million emails per

  • second, one million IM messages per second, 8

  • terabytes per second of traffic, 65

  • billion phone calls.

  • It's a huge machine.

  • And, in fact, if you were to spec this machine out and try

  • to sell it on Amazon, the processor has got a billion

  • chips, has got one megahertz email, one megahertz web

  • search, ten kilohertz instant messaging, and one kilohertz

  • SMS. And that's a very big machine.

  • That's a very, very powerful machine.

  • And this is, of course, the aggregate of all the chips and

  • all the machines in all the world.

  • We've created one large machine.

  • Bus speed of ten terabytes, ram of 200

  • terabytes, storage exabytes.

  • Of course these are all going off the edge just as we speak.

  • But the idea is, is that that is the machine that we're now

  • programming for.

  • We're not programmed for your laptop.

  • We're not programming for your cell phone.

  • We're programming for this machine.

  • That is the thing that we're doing.

  • And the other thing about this, is that if you take a

  • biological view, it's very close to the complexity of a

  • human mind, a human brain I should say.

  • It's got a quintillion transistors.

  • And if you take all the transistors and all the PCs in

  • the world, there's somewhere a quintillion of them

  • operating right now.

  • They're all live.

  • It's got one trillion on the web.

  • The web has one trillion synapses,, one trillion links,

  • 20 petahertz synapse firings.

  • That's the speed.

  • It has a 100 billion clicks per day.

  • So it's a very large brain.

  • So we are this machine.

  • Because if we add then to that web machine our own brains as

  • we sit behind, and as we guide the clicks, and as we

  • interface, if you add the collective intelligence of the

  • one billion people online right now, it's a very large

  • machine and it's a very smart machine.

  • One of the questions that people is well, are we going

  • to take off?

  • Is there a singularity about to happen?

  • I think there is, but I think it's not in the way that Ray

  • Kurzweil and other people would believe, in the sense

  • that there's going to be a rapture or people left behind.

  • I really like this.

  • This is one my favorite New Yorker cartoons.

  • It says, "Sir, the following paradigm shifts occurred while

  • you were out." I don't think that happens where we will

  • notice things.

  • In fact, I think what's happening is that we'll go

  • through this without really noticing it in the same way.

  • "Hey, did anyone notice we are using language?" There was

  • never that conversation.

  • People never sat around the campfire and said, hey,

  • actually we're talking.

  • Talking?

  • Yeah.

  • language obviously was a singularity.

  • But we passed through it without really noticing.

  • And I think the same thing is going to

  • happen with this process.

  • What technology gives us is the possibilities.

  • People often say technology creates as many problems as

  • it's solving.

  • Actually I think that's true, but each of those problems is

  • actually a possibility.

  • And what it gives us is possibilities.

  • So differences, diversity, options, these are the things

  • that technology creates, and it's a great bargain.

  • All of us will always go for that.

  • So each of these nodes in this network is actually

  • possibilities.

  • This is a possibility space.

  • And that's what technology is creating.

  • It's an infinite gain.

  • And the idea is to keep the gain going.

  • So if you can imagine, for a moment, van Gogh being born

  • before the technology of oil paints was invented, or Mozart

  • being born before the piano was invented, or maybe

  • Hitchcock being born before the technology of

  • the film was invented.

  • So there are people alive today whose technology has not

  • yet been invented, whose perfect means for their great

  • genius has not yet been invented.

  • And I think we have a moral obligation to actually create

  • those possibilities so that everybody in the world has a

  • potential to really, really use but

  • they've been born with.

  • Those possibilities, that quest for the possibilities is

  • really what technology is about.

  • And I think that's why we're here and why what you're doing

  • at Google is so great, because we're trying to play the

  • infinite game.

  • Thank you.

  • [APPLAUSE]

  • AUDIENCE: So I'll take questions if people have them.

  • Way in the back.

  • Through a lot of your talk you talked about science, and then

  • at the end you started using the word technology.

  • Now that doesn't bother me particularly, but if I think

  • about a lot of discussions on the scientific method and so

  • one, and I look at curricula in colleges, science and

  • technology tend to be kept very separate.

  • Science is a much more analytic technology and

  • [UNINTELLIGIBLE PHRASE].

  • Do you tend to distinguish the two, or to you are they all

  • lumped together?

  • KEVIN KELLY: I don't do things as much as I probably should.

  • But as I did suggest earlier, I think science has change

  • much more by its tools than anything else.

  • So the tool part of science is as important as anything, and

  • is definitely part of the scientific method.

  • So we can change science by creating tools that either

  • measure or, as science because more and more information

  • driven, the tools of information become as

  • important as anything else.

  • And so that's why a search engine I would make on the

  • list of some of the most important scientific method

  • inventions.

  • Some of the most important methods in the invention of

  • the scientific method is the search engine

  • because it's a tool.

  • And it is a technology, but it's actually not as divorced

  • as other strategies.

  • So I think there is a difference, but not much as

  • some people think.

  • Yeah?

  • AUDIENCE: [INAUDIBLE PHRASE]?

  • KEVIN KELLY: So let me repeat the question for

  • those in the back.

  • The question was that science is very important, but

  • scientists themselves are not rewarded usually as much as

  • say, other people, who may be involved in

  • technology, for instance.

  • Yeah, I think that's bad.

  • I mean, I think it would be better if we

  • rewarded them more.

  • And I think some of them moved towards in universities to

  • have scientists partake in their inventions through

  • patent arrangements.

  • It's one way in which society has been trying to reward

  • scientists.

  • But I think in general they aren't as rewards.

  • My wife is a scientist. She works at Genentech.

  • She's a scientist there.

  • So I know that scientists, like teachers, don't get the

  • same reward as they should, and I should also say, are

  • often really badly portrayed in movies.

  • I mean it just infuriates me to see someone wearing a white

  • lab coat and being crazy, like [UNINTELLIGIBLE]

  • in terms of Back to the Future.

  • So yes, scientist don't get the respect they should.

  • And I don't know if there's a mechanism to remedy that other

  • than hug a scientist tonight.

  • Right here.

  • AUDIENCE: Is there an asymptote to the amount of

  • knowledge that we can reach, either as humans or whatever

  • [UNINTELLIGIBLE]?

  • KEVIN KELLY: Is there an asymptote to the amount of

  • information humans can reach?

  • Actually this is another little talk.

  • But I believe that what science really generates is

  • not knowledge, but ignorance.

  • A really good question, a really good question is a

  • question that would generate more

  • questions than it answers.

  • Every time we have a really good answer, it will generate

  • two or three more questions, and I think of those questions

  • as possibilities.

  • So, in a certain sense, while the amount of information is

  • being increased, our ignorance is actually increasing faster.

  • We have far more ignorance now than we did 400 years ago.

  • We have far more questions.

  • We have far more things we want to know about, and we

  • realize we don't know about, than we did.

  • So in a certain sense, what science is actually trying to

  • do is expand that frontier space.

  • I often think of it like friction.

  • The new economy was supposed to decrease friction so we

  • have a frictionless economy.

  • It did that, but at the same time at the frontier, at the

  • edges, it was increasing friction.

  • It was increasing the unknown.

  • And so we actually are expanding the unknown faster

  • than we are answering it.

  • And I think that's because those unknowns are all

  • possibilities.

  • Those are things that are close enough.

  • The adjacent possible is what Stuart Kauffman calls it.

  • We now are close enough to see that we don't know that.

  • And that actually is expanding faster than what we know.

  • So I don't think there's an asymptote Because each one of

  • those unknowns is more information.

  • So it's infinite.

  • AUDIENCE: [INAUDIBLE PHRASE].

  • KEVIN KELLY: Oh, individually yes.

  • But we're beyond that.

  • At the time when any individual knew everything

  • there was is long gone.

  • This is that big ball, that big machine collectively is

  • understanding things.

  • And as that happens, we will know as individuals,

  • percentage wise, relatively less every year, less of all

  • this known.

  • Yeah?

  • AUDIENCE: [INAUDIBLE PHRASE]?

  • KEVIN KELLY: So the question is, is that if opportunities

  • are expanding, could we not say that the opportunity costs

  • also expand, and maybe even faster?

  • Yes.

  • I think they do.

  • And here's what I think one of the remedies is.

  • It's that we basically want AI.

  • We want many, many minds.

  • We want every possible type of intelligence.

  • We want more things seizing those opportunities.

  • So the costs will increase, but part of what the economy

  • does is it always drives things down at the center.

  • So you have to core where the opportunity costs are less,

  • where friction is less, and at the edges which are expanding,

  • there's always friction, and ignorance, and profit.

  • If you haven't figured that out, profit

  • is related to friction.

  • There's no profit at the frictionless center.

  • There's only profit in ignorance because you have

  • information that no one else has.

  • At the center where there's friction, where everything is

  • known, there's no profit because everybody knows

  • everything.

  • So that expansion is always happening at the edges where

  • there is high opportunity cost, high friction, high

  • ignorance, and high profit.

  • Well I guess I've answered everybody's questions.

  • That's great.

  • Thank you for having me.

  • It's really been a real pleasure.

MALE SPEAKER: Hello, hi.

字幕與單字

單字即點即查 點擊單字可以查詢單字解釋

B1 中級

未來五十年的科學 (The Next Fifty Years of Science)

  • 234 20
    RexWei 發佈於 2021 年 01 月 14 日
影片單字