Waiting..
Auto Scroll
Sync
Top
Bottom
Select text to annotate, Click play in YouTube to begin
00:00:01
this is a kind of a fortress set up of kind of the professor against the students but the cable is short so I have to duck back there periodically
00:00:15
thank you all for coming out I believe I'm the second prize in the raffle this
00:00:26
this talk is is about some of the ways I think about our field it's not particularly technical it asks a lot of questions it's an attempt to be critical
00:00:40
without being a rant and it won't be successful unless I can get you to interact periodically there'll be a couple of opportunities early in the
00:00:55
talk and the title of the talk deals with two properties of frogs one of them is that their nervous systems like all
00:01:10
nervous systems are mainly interested in differences but unlike us they are able to habituate very rapidly and so it is
00:01:24
amazingly true that if you put a frog in a pot of water and heat it up very slowly they will not get out because the the difference is that their nervous system are detecting are too small over
00:01:38
the unit of time that their nervous systems work on and they will just stay there until they cook and of course we aren't like that and the other
00:01:52
interesting property of flies which i think is one of the most amazing properties and it was first discovered in a landmark paper called what the
00:02:05
frog's eye tells the frog's brain and you may or may not know that the retina in your eye is acts part of your brain is actually extended out from your brain and your eye
00:02:19
actually does a little bit of thinking right in the eye for efficiency reasons for frogs their eyes do a lot of thinking thinking that actually triggers
00:02:33
behavior and so one of the most interesting things about a frog this if you take its natural food which your flies paralyze them with a little
00:02:46
chloroform but peep them alive and put them in front of a frog it will starve itself to death because its conception of a fly is not that if you take oblong
00:02:59
shaped pieces of cardboard and throw them at the Frog it will eat them until it's stuffed because it's built in pattern for food is an oblong thing
00:03:14
that's moving and there's no limit to the amount of cardboards that a frog will try to try to eat so one of the ways of looking at this is that the
00:03:27
frogs both of these conceptions of normal that the frog has are quite harmful for it under certain conditions so it considers normal food to be oblong
00:03:39
things that move and it considers a non change in temperature as far as it can detect it to be quite safe for itself so this is the famous Pogo cartoon this
00:04:04
is true of so many things about us humans trace back some problem and very often the trace is back not to physical nature but to human nature and I think
00:04:18
this is the problem in many fields a little bit later we'll talk about how the biologists have dealt with some of the same problems that we have in
00:04:27
computing so first so here's our first opportunity for interaction who's that first guy Newton who's the
00:04:45
second guy Darwin and the third guy the most famous face on the planet who's
00:04:59
that first guy who's the second guy nope who's the third guy who's the fourth guy
00:05:17
and who's the fifth guy pardon nope so first guys John McCarthy never heard of him so these guys are all magic people in
00:05:38
our field so John did many of the most seminal things to this day in the field of reasoning and programming languages
00:05:50
second guy is Ivan Sutherland he only invented computer graphics and he did it in a way more interesting than most computer graphics today probably if you
00:06:01
had to pick the most powerful groundbreaking distance from zero thesis that has ever been done in our field it's his thesis and if you don't know
00:06:14
what he looks like then I'm sure you don't haven't read that thesis and Engelbart so what did angle Bart do okay what else did he do
00:06:29
okay so when I was giving talks in 2004 this is the question I asked every audience at every University everybody had heard of angle Bart in the mouse but
00:06:41
most people don't realize that Engelbart was the prime conceived of what we call personal computing today and the prime conceived er of what we think of as
00:06:54
hyper linking ideas together over a network with collaboration between people and the prime conceived ER of what collaborative work would be like and by the way he built all of this with
00:07:06
his research group in the 60s and it was demo to 3000 people in 1968 41 years ago so an interesting thing here is to just
00:07:18
ponder the fact that even though you've heard of Engelbart's name if you had taken the trouble to type e ngle ba RT into google within the first three hits
00:07:31
some of them you will be doing it right now you'll find the bootstrap Institute which have the 75 seminal papers that he and his group wrote and you'll also find
00:07:42
the videos of that what is called the mother of all demo in 1968 and yet in fact the internet and personal computing were invented just so
00:07:55
you could do that and so now ponder the fact that you do know the name of a famous person in our field from the past and yet you didn't have the curiosity or the energy to do that simple little
00:08:06
typing job so if this were physics you would all be dismissed from the field because physicists would not tolerate somebody being in physics and not knowing quite intimately what the
00:08:21
pioneers and developers of physics including Newton and Einstein and so forth actually did it's not an optional idea but in computing seems to think
00:08:32
that it is so part of my thing to get you to ponder is I don't think computing is a real field it acts like a pop culture it deals in fads
00:08:45
and it doesn't even know its own roots and worse than that it does not know about the really good things that were done in the past for instance Engelbart's conception if you were to look at it you would see far exceeds
00:08:58
what Tim berners-lee was thinking about and his demo far exceeds what you can even do on the Met and the web today and why can't we do it today because the
00:09:10
people who set up the web did not know about Engelbert and did not take the trouble to see if anybody had been thinking about this so this is the mark of a pop culture this guy is the world's
00:09:22
greatest computer designer designer among other things of the Burroughs be 5,000 and that machine implemented in hardware what we would call a byte coded interpreter for a higher-level language
00:09:36
today today we can only do it in software because Intel and Motorola don't know this is a good idea the bharden knew it was a good idea in 1961 and designed and built with borrows
00:09:48
a computer that could do that and a lot more so you can't be in computing and not know what Barton did and not know who he was and this guy
00:09:59
was part of the for funders at ARPA that funded the fundamental research in the 60s and he was also the guy who set up Xerox PARC his name is Bob Taylor and as
00:10:15
they used to say in the old days no bucks no Buck Rogers so part of the reason the past actually was quite a bit more successful in the present is
00:10:26
because the funding was done better and people who knew how to run research groups like Bob Taylor you had to do it so I knew you were going to give that
00:10:37
response because okay how about this picture parted yeah so both of you were
00:10:52
right so iliac one was a johniac the von neumann princeton computer and was built here at the university of illinois and they actually got done before the
00:11:07
Princeton people did so there are several of these machines built all of the same plans and the Princeton people are a little bit late about that I just
00:11:24
love this computer so this is one of the first supercomputers to ever be done happen to be done right on this campus called iliac 2 and it had the register
00:11:40
pipelining idea even before the IBM stretch did so this was a really serious computing machine done on this campus by the professors and the students kick ass
00:11:57
how about this play-doh what was played up yeah then it wasn't exactly the first
00:12:12
graphical interface but so this is a system that was built by dom Bitzer and colleagues here at the university of illinois is one of the largest funded
00:12:24
computer-aided education projects in history there are a lot of people worked on this and give you the scope of it 2000 terminals connected to large
00:12:37
control data mainframes and the flatscreen displays on these terminals were actually invented here at the universe of Illinois so they could make flat screen display terminals and why
00:12:52
did dom Bitzer want to do it because he wanted to back project colored slides they wanted to have something that could do multimedia education so when you use one of these terminals in the top of
00:13:03
these things was a Kodak slide carousel and so when you went to learn something you get a slide carousel that would have the high resolution images that were going to be part of it put it in there and then you had a 512 by 512 plasma
00:13:17
panel display this is the first flat panel display in the world that was actually practical done right here at the University of Illinois and with cooperation from Owens Corning shouldn't
00:13:33
this makes you proud to be in the University of Illinois computer science department right you should be because this place is one of the leaders of the world but guess what you guys haven't
00:13:47
done a supercomputer for a long time you haven't done a project like this for a long time you're using vendors computers and vendors software and guess what you can't invent the future by using vendor
00:14:00
stuff it's all looks back in the past the best of it and most of it is exactly the wrong way of looking at the future so Illinois like most places like UCLA where I'm on the faculty
00:14:14
most of the major universities in the country have given up the future in favor of something else most places in this country when it was really hard to design and build computers built around
00:14:27
yes well no I don't think so because this assumes that there's something good about what you can buy from vendors and I think when that's the case it's
00:14:54
worthwhile buying from them but in fact the vendors are struggling like mad themselves because Moore's law marches on and they don't know the first thing about doing parallel computing for
00:15:07
example somebody needed to be doing that a university is the right place to do it so not enough has been done and I think most of the great inventions come out of a kind of a something as more like a
00:15:20
university culture and by the way there is a technology called fuel programmable gate arrays in which regular students like you and me can sit down and make a
00:15:31
thousand processor supercomputer by writing not a lot of pages of code because the state of FPGA is of in the last year and a half or so has gone
00:15:45
through a knee of the curve and is now a serious contender for allowing people to do huge multi processing projects by themselves if you're interested we can
00:15:55
talk about that a little bit later so some years ago my wife who's a writer and an actress and an artist and so she
00:16:11
had one of these huge Mac cinema displays on her desk and she showed me this display and was kind of odd over
00:16:24
here she had all the applications that she used that were WYSIWYG she really liked them and here all the app the stuff that she does on the web and none
00:16:36
of it was WYSIWYG at that time and so she said in these apps I can see and do full WYSIWYG opera authoring and the web browser has all these modes and I
00:16:55
have to type through a keyhole and I have to wait to see whether what I did was ok and all these stuff and and why is that she asked me and I said well it's because the stuff you like was
00:17:09
invented in the 70s and the stuff you don't like was invented 20 years later what do you think about that how could that happen it's not that people don't
00:17:23
make mistakes but you like mistakes to go away yes yes and it's quite
00:17:35
interesting of course where was this invented web browser was done by University of Illinois students a little
00:17:49
bit after that age of people that made playdough terminals and supercomputers so in 1993 here's mosaic and one of the interesting
00:18:09
things about mosaic was was it remotely is nice at what angle Bart had done in 68 and it was full documentation of course they didn't know anything about it Tim berners-lee who did do the web at
00:18:23
CERN was quite chagrined when it was pointed out to him just the sweep of Engelbart's ideas he was very upset that they did what they did the way they did
00:18:35
it and even more interesting five years six years before the web browser there was HyperCard at Apple which is the perfect model if you think those of you are familiar with HyperCard it's the perfect model for
00:18:51
what a web browser should actually be it had full WYSIWYG editing right in the thing and anybody who understand this at all would have realized oh yeah this is the perfect
00:19:04
thing it already has hyperlinking in it there's an editing model it's been tested on four million users we know it works nope a bunch of hackers got together and
00:19:17
one of the ways to think about this is Engle Bart invented the wheel and Bill Atkinson with HyperCard invented a better wheel that's really great and and
00:19:31
unfortunately I can't even give the web browser the flat tire award and or even the square wheel because you can imagine improving both of those to be something
00:19:43
good but you know when you invent a broken wheel there is it isn't obvious what the wheel is or it doesn't even work at all flat tire you can you can
00:19:55
run it at the expense of the tire and figure out what the tire should be in a square wheel you can run it and figure out it should be rounder but a broken wheel won't work and so attempts to fix a broken wheel
00:20:08
produce more variations of a broken wheel and we've had 16 years of them ok
00:20:21
so let's a man here you are today let's suppose you're at a computer science those you had an undergraduate degree in computing from back then and you're
00:20:35
faced with this idea that yeah there's this internet that's going all over the world and it's going to be not just a consumption mechanism for stuff that's
00:20:48
already there but everybody is going to be an author and a publisher and so how would you solve that problem anybody got an idea what would your approach be to it knowing what you know today or even what
00:21:01
you would have known back this is a complicated problem right there six billion people in the world there now a couple of billion nodes in the internet there is zillion everybody
00:21:17
wants to do something yes the question is how much work do you want it you want to solve this problem really well but how much work do you want how much do you think your browser should actually
00:21:31
know in order to solve this problem really well because think of everybody has a different interest in the web there's many kinds of media you don't
00:21:45
even know what kind of media is going to be invented and for any given kind of media there like video there are dozens of different codecs and just think of the what the problem is so how would you approach
00:21:57
this problem yes well but suppose they want to write machine code because they know how to do something really really fast like
00:22:20
suppose somebody wants to implement their own codec yeah but you're you're partly partly there but I'm talking about letting people exercise their
00:22:32
creativity and sometimes they're going to need the most every resource on the machine to be used but of course we can't allow them to take over the
00:22:46
machine so what would you do yeah so it's closer it's almost a good idea after 16 years including see Google has some real interest in this because
00:23:19
Google did v8 because they couldn't stand to have the normal conception of a slow JavaScript and they're doing the thing that you just mentioned because
00:23:32
they realize that people who have ideas also so ok so this is an interesting
00:23:44
thing so it really should be like an operating system kernel right the operating system kernel controls address
00:23:58
spaces that can confine computations completely therefore you can actually allow any binaries to come down and be used you can completely control what goes in and what goes out and all of a
00:24:13
sudden your browser is simply something that allows bitmaps canvases that you give to these things to write on to be displayed so this it's there's nothing you have to do because you don't want
00:24:27
it's like the the the good idea in UNIX was hey we don't want a big operating system we want the tiniest kernel we can have and then we want to use address
00:24:41
spaces in order to protect everybody from everybody else so this is this is operating system 101 this is well known all the way back in 1965 and 1966 it
00:24:55
should be the first thing that occurs to anybody so I figured somebody young so this occurred everybody in the 60s and 70s knew this is the way to do it we're all surprised it wasn't done this way
00:25:08
but I figured somebody in the current generation would have figured this out and so I went looking on the net using Google looking for downloadable binaries and stuff like that
00:25:21
and I found this grad student in Cornell in 1997 Oh far Erlich son who is from Iceland wrote this paper in this paper if you just type in the title of that
00:25:35
paper you'll find it's an HTML document that he put on and on the net in 1997 and his problem by the way was hey I want to run a Kodak I want to run a Kodak and I don't want to have a system
00:25:48
bin worried about whether it's an executable or not so it's this thing should be like an operating system so of course I wrote an email to him he's at the University of Iceland now but on
00:26:00
leave to Microsoft up in Mountain View and had a long correspondence with him on the pathway of these ideas and he
00:26:11
started a company that Google acquired and is using a couple of the ideas but interestingly not all of the ideas in this paper which actually solved the problem very nicely so this is an
00:26:26
interesting thing where it wasn't that somebody didn't understand what the what the problem was and what the right solution was it's that the larger mass was quite happy to plunge into a de
00:26:39
facto standard even though most of the original people to use the web we're computer scientists they didn't protest that this was a very bad way non scalable way of doing it they just plunged in and sixteen years
00:26:53
later Google is and other companies are struggling to actually use the internet the way it was intended to be that's pretty interesting to me now another
00:27:03
story is 35 years ago the kind of personal computing we think of today was done at Xerox PARC in 1973 and much of the paraphernalia we have today
00:27:21
was done in about ten thousand lines of code Smalltalk code written for this purpose and so another question you
00:27:31
could ask is well jeez the browser is a terrible design but it does have JavaScript in it and it turns out JavaScript even though it's slow by today's standards it's faster than an
00:27:45
Alto was back in 1973 so hey why doesn't somebody just do a whole personal computing system in Java like everything Windows interface the
00:27:57
overlapping idea is two and a half D graphics like every damn thing in it and so basically what you're doing is just drawing a line and saying this stuff is
00:28:09
crap but it runs and I should be able to write 10,000 lines of code so the basic idea take the rubble that's the browser
00:28:22
make an arch out of it get the guy who did the job originally at Xerox PARC Dan Ingalls to do it in make stronger classes in JavaScript make
00:28:40
a better graphics system make the widgets and stuff to make an application development system and and an interactive development environment and
00:28:51
if you want to try this out I don't have time to do it but this is what it looks like running completely in a browser took about five people six months to do
00:29:07
could have been done any time but nobody knew how to do it has it take some guy who had done it before in the past because nobody knows how to do a system like that in ten thousand lines all
00:29:22
right so this is a craft that was lost because people have been used to just building on the stuff that's already there so when this new thing came along it wasn't anybody around who was working on it that could just say okay we want
00:29:35
to have something like the WYSIWYG interface and authoring tools and everything else that we enjoy on the our regular personal computers and we want to be able to do all of the stuff in the browser so this is called the lively
00:29:50
kernel it was done while dan was at Sun you can look that up so here's another way of looking at this is if you take the minimum bar I would say the general
00:30:03
practice today is below that minimum and take a qualitatively better bar I'd say that the best that was done by the ARPA park community like the Internet and the ethernet and personal computing and
00:30:17
stuff like that was better I think the best knowledge of today is better than that it's not that everything is back slid absolutely not it's not that the
00:30:32
people are not as good as they used to be I'm convinced that there are more people of high ability available today than there were 40 years ago there's not
00:30:45
a complaint about the people but I'm complaining about the outlook of the that produces such poor results now the
00:30:56
problem is that the real bar today is above the best of what we did a long time ago and it's above the best knowledge today this is the problem this is why companies are writing millions of
00:31:08
lines of hard to deal with code because the ability to scale is potentially there yet you have a question okay the ability to scale is potentially there
00:31:22
but the knowledge of how to do the scaling is not there by anybody so the real way of looking at this is we need something like that today and we have to think figure out
00:31:35
something that's a little bit above the best knowledge today that has a stair-step built into it to get up to where we are also things are just going to get worse than they are now so this
00:31:47
is a very very difficult problem it's been fomenting for years I think many of you are aware of it in one way or another I'm just I'm pointing it out not so that you believe my point of view on it but just so you think about it
00:32:00
yourself so back to a couple of ideas so here's the printing press appearing in 1450 and whenever a new idea appears you
00:32:18
get two things you get news news is stuff that's incremental to what we already know this is why you can tell a news in five minutes hey a train just
00:32:33
crashed we all know what that means so the reason we want to know is we're excited by changes like that but we don't have to have it explained to us there's no epistemological hurdle we
00:32:47
have to go through to understand that and so the original idea for the printing press was to do what monks do but do it cheaper and there are a couple
00:32:57
of other side things but when ideas are really interesting is also do and knew is by definition not like what
00:33:12
we already know so there's no news about new there's nothing you can tell somebody in five minutes about what new is and new for the printing press was a
00:33:28
huge change in Outlook for our planet and one of the changes was from believing to arguing in a new and special way and it took 150 years to go
00:33:41
from the Gutenberg Bible to Galileo two hundred years to go to Newton and three hundred plus years to go to the American Constitution but one was the driving
00:33:58
force for the these other things now interestingly about 50 years after the printing press there were people who were writing in Europe like Erasmus who indicated that they understood completely what the printing press was
00:34:10
about and what was going to happen from it but the larger mass of humanity took 150 to 300 years to actually gradually change their thinking around to start
00:34:21
thinking that way ok Leonardo had a high IQ but he couldn't invent a single engine for any of his vehicles so if he
00:34:38
had an IQ of 500 and you were born in 10,000 BC you're not going to do a lot you can outwit everybody before they burn you at the stake so Henry Ford was not nearly as smart as
00:34:55
Leonardo but he was born in a better century so he revolutionized transportation so knowledge almost always Trump's IQ and the big problem
00:35:11
with knowledge is most of us know a lot but it's not it's not good knowledge might even never have been good knowledge but sometimes it was good knowledge and now it's bad knowledge so
00:35:24
knowledge is this double-edged sword very powerful when we have it right and the thing that gets it right for us is when somebody is able to change the outlook the fancy word is
00:35:37
epistemological stance the way we actually look at things and Newton was one of these people so he changed what knowledge actually meant and he changed what engineers and more practical people
00:35:50
could do and so the line there is point of view is worth ad IQ points you go from a weak way of looking at something to a strong way of looking at something that's like getting an extra brain and
00:36:04
Newton is a perfect example because before Newton the smartest people in the world couldn't do things that a high school calculus student can easily do so calculus is like getting an extra brain
00:36:16
and it makes you much much more able than a genius who doesn't have it so these outlook changing things are really critical so I think of knowledge
00:36:31
of silver outlook as gold and IQ as a lead weight I believe one of the biggest problems with computing is in a sense we have too many smart people it attracts
00:36:44
cleverness and you can do clever hacks but the clever hacks don't scale well and it's very hard to build a halflife into software it just stays around forever and so what what's actually
00:37:00
happening is kind of something like building a large garbage dump that makes it the odor of which makes it very hard to think about other things so if we
00:37:10
take news or normal we can think we can solve problems avoid obstacles beg every once in while we have an outlaw thought
00:37:23
but we went to school we went to church we have parents we live in a world of normal and for most human beings the world of normal trumps any weird new
00:37:38
kind of idea try having one when you're going out for tenure think of all the different reviewers who already have an idea of what computer science should be
00:37:52
about it's crazy and people start winding up gaming the system and doing lots and lots of little papers that mean nothing at all in order to get the
00:38:04
approval of these reviewers whose ideas about computing might be completely off but every once in a while maybe in an
00:38:17
unguarded moment like taking a shower getting up in the morning or something you get a Kirpal not that this kerpow is true most ideas are mediocre down to bad
00:38:32
but at least you've got a new Kirpal and these career paths are the things that need to be tested out science needs them as much as art because you have to get ideas from somewhere so when you have to
00:38:45
be in a state to get these cows and then you have to have the tools to be able to discard the Kirpal if it turns out to be a really stupid idea after all
00:38:56
so think about this is if the pink plane there is reality then what do people think the blue plane is like insanity
00:39:07
you are insane and learning a new idea something that your brain isn't well set up for could require almost as much creativity as inventing it in the first place because you have to invent it
00:39:24
inside your head and so you can say that normal is the greatest enemy with regard to creating the new in the way of getting around this is you have to
00:39:37
understand normal not as reality but just the construct and a way to do that for example is just travel to a lot of different countries and you'll find a thousand different ways of thinking the
00:39:51
world is real all of which are just stories inside of people's heads that's what we are to normal is just a construct in to the extent that you can see normal as a construct in yourself
00:40:02
you have freed yourself from the constraints of thinking that this is the way the world is because it isn't this is the way we are so changing people's minds is hard because of this so I'm going to give you
00:40:21
just two simple dimensions so this is the whole human race just 5% of us are intrinsically interested in ideas 95% of
00:40:35
us are interested in new ideas and tools just so far as they contribute to our current goals so once we have a goal we tend to be fastened on it you can
00:40:47
imagine why so when somebody shows us something new we immediately evaluated in terms of our existing goal structure it doesn't fit into our existing goal structure forget
00:40:59
about it 5% of us though are get interested in whether this is an interesting idea or tool intrinsically so let's take a different dimension this
00:41:12
is one that has been well studied about 85% of us generally do things for the approval of others who are social beings after all people use the Internet
00:41:26
very much for confirming and disconfirming what they should be thinking about by finding out what their friends think about so this is completely normal to human beings about
00:41:37
15% of us have many of our motivations be internal so if we do something that we like we don't really care what other people think and to the extent that
00:41:51
these two dimensions are independent you get something like this and so 1% of us
00:42:03
love things for their merits and we don't care what people think you can imagine what do those people do with their time and the other extreme is
00:42:17
eighty percent of us are instrumental and our goals are primarily determined by whether other people approve of them
00:42:31
so think about what these one-percenters actually do and think about what happens to those ideas when they try and get
00:42:41
society to change and then the other extremes are outer ideas and tools and a very dangerous group of people who are
00:42:56
inner and instrumental there are a lot of managers and politicians in this group dictators of countries and so
00:43:09
forth okay here's another analogy it was an old one and has been shown to be a good one that when it rains it's somewhat random as to where a little gully starts
00:43:25
but the gullies have the property of being self optimizing once you get one started they are more efficient at getting water through them and so they
00:43:36
erode faster and so pretty soon you've got something human memory is rather like this basically once we get comfortable about something
00:43:49
we are extremely homeostatic about it we love our gullies and they take a long time so a little bit like the Frog in the pot occasionally our gullies will do
00:44:02
us in because we can get awfully comfortable and then get surprised by something like AIDS which takes a long time to really show up and we humans have a hard time forgetting
00:44:16
things we don't have the men in black stick for when we have a bad idea or learn something that isn't actually not good for us we can't erase it and so we
00:44:28
have this interesting this is something that's been noticed in programmers there very large percentage of programmers think like their first programming language for the rest of their life no matter what programming language they learn afterwards if you look at their
00:44:42
code it looks like that you can tell what first language they had because that was the first time they got fluent those were their paradigms for thinking about problems so human beings are
00:44:53
actually quite anti learning and in fact most of these things are built into our genes for example we are a coping species not a product not a species that
00:45:09
likes to progress we're basically social we all have language so more than three thousand cultures have been studied by anthropologists and every single one of them has about 300 things in common and
00:45:21
so the determination is that these things are not things that were invented but are actually built into the genes a very human so once you've and by the way if you
00:45:35
look at these things you can realize if you want to make a lot of money build an amplifier for any of these using technology and you'll win see why so you
00:45:48
know when television came out the movie people didn't think it was going to sell but they didn't realize it was a fantasy amplifier that was going to go in everybody's home of course it's going to win but they couldn't understand it for
00:46:01
what it was and once you've done that you can take a look at things that have not been found in every culture and a couple of the interesting ones are at the top progress as it was called in the
00:46:22
18th century is an invention it's an idea was first articulated not so many hundreds of years ago because back then was the first time when people realized that the world that their children are going to die and was going
00:46:34
to be different than the world that they were going to die in and so the people who wrote the American Constitution even suggested that there'd be a new Constitutional Convention every 50 years because things are going to change fortunately they didn't carry that out
00:46:47
they were unusual compared to our current puppet politicians and the things in blue they're the non universals are harder to learn the ones in the pink because they're not as
00:47:01
strongly built into us they all had to be invented and so an accurate picture of us is that we're pretty much cave people with briefcases and there's
00:47:14
actually a group of people who study us as cave people with briefcases called behavioral economists so if you've looked at so one of the books is called nudge by Richard Taylor another great
00:47:27
book is called predictably irrational by Dan Ariely these are people who study us as we are and why we actually make decisions even against our best interests and most of these decisions
00:47:42
are made because of things that are built into us and this would be a humorous picture except that the cave guy with the briefcase has an ICBM in his briefcase and a lot of other things
00:47:55
that are too easy to invent for technology and the wrong kinds of things for K people with briefcases to be carrying around so
00:48:10
so a lot of what's going on today in technology in general not just in computing is basically for marketing reasons we automate the pleistocene so if you look at what is being sold and
00:48:23
used on computers there's almost nothing that wouldn't be completely recognizable to a cave kid right there's hardly any new media on computers it's all
00:48:35
imitations of old media and most of it has to do a storytelling of one kind or another very little a little bit has to do with those powerful ideas in blue
00:48:48
okay here's another thing I think you're all familiar with that in a short period of time change appears linear but often that's just because you're looking at an
00:49:02
exponential curve at two finer grain so these are big surprises for many people so for example with silicon in round 65
00:49:15
we had Moore's law and that allowed us to have a very different look at what was going to happen and we use the Wayne Gretzky theory of hockey anybody know
00:49:27
what that is yes that's right so they asked him why are you so much better than everybody else and he said well everybody just goes where the puck is I go where it's going to be so this is a
00:49:42
good idea because it takes time to invent things so if you invent things for where the puck is you're going to be behind by the time you get done so what
00:49:57
you'd like to do is to know what that technology is going to be out there and the wonderful thing about computers is you can get that by just spending money
00:50:12
that's all it is that's the easiest thing for Americans to do we just love money and we make tons of it so for example that's what the University of Illinois did when they did iliac 2 they
00:50:25
wanted to get five or ten years in the future in computing so they built this supercomputer neuter enabled them to probe into the future and it cost millions of dollars so what if they
00:50:37
built something that was not as a compute supercomputer at that time all they'd learn is obsolete stuff there was not going to be usable later so a park
00:50:50
we looked at about 12 years or so to the late 80s and said we want a computer in 1973 that will compute like the computers are going to compute somewhere
00:51:02
between 1985 and 1990 those cost $22,000 for us to make we had to invent them and build them we built about 2,000 of them and so 22,000 in those days was worth
00:51:17
about 80,000 today right so it costs 80,000 bucks to have a 1989 Mac but if you had in 1973 just think of what you could do that can convened all the stuff
00:51:30
that the 1989 Mac was going to run 1989 Mac would have to wind up being like what you did because nobody else would have the time to invent this takes years to invent good things so when I come to
00:51:45
a university and I see students using current day laptops it drives me crazy you're in the past you're not even in today and just think
00:52:00
about the scope of the stuff that's already been done you need a lot more leverage the way to get leverage is to get a ton of computer power and write very simple non-optimized languages that
00:52:13
run very very quickly and you should be able to do them every couple of days to get power you should be able to just throw them away use them like Kleenex because it's rather easy to write things
00:52:27
like that if you don't have to optimize and the way you don't have to optimize is make hardware that runs a hundred times faster than what you've got today so unless you do that you can only do
00:52:38
incremental stuff today so a lot of the incremental ISM today is the lure of a cheap computer when it's actually standing a lot of people's way okay
00:52:52
and so I think the last set of ideas here is about problem solving and one of my best friends who died just a couple of years ago it's one of the great
00:53:05
problem solvers in fact was awarded the mechanical engineer of the century award in the 20th century it's one of greatest awards any engineer can do it this guy was completely magic
00:53:20
about solving problems so BAM powered flight is an idea that's been around for a long time and so Henry Kramer's idea was well I'm going to offer a large
00:53:34
prize like $100,000 but people tried to win this prize for 40 years and couldn't
00:53:44
do it Paul MacCready his brother-in-law wound up with $100,000 debt and Paul
00:54:02
being a nice guy he decided to assume it to make his wife happy all of a sudden Paul was a hundred thousand and dead and he was driving across the Arizona desert with his family one day and he had been
00:54:13
the world sail a sail plane champion a number of times it was a fantastic pilot this is driving along looking at a hawk circling in the sky across the Arizona desert and he suddenly realized that he
00:54:27
had seen the exchange rate of the British pound that morning in the paper the current exchange rate the 40,000 pound Kramer prize was worth exactly a hundred thousand American dollars so he
00:54:41
started thinking boy if I could win that prize I could play off my brother-in-law's death so he started thinking about how could you actually do man-powered flight he said the problem
00:54:58
is we don't understand what the problem is really good people have been failing for 40 years so this must be hard so his idea was forget about man-powered flight
00:55:10
first thing we have to do is to try and understand what the problem is you look to see what does why aren't these people doing this so what happened is people were building very elaborate complex
00:55:21
designs for man-powered flight and they would take it up and they'd have a crash it would take them a year to rebuild the thing so they're getting like one flight every 8 to 14 months and Paul said well
00:55:34
we need to we need to be able to do 12 crashes a day so his first design was this contraption here that was literally made out of baling wire and some plastic
00:55:46
tarps and some aluminum struts and stuff but you could fix it in a few minutes so in just a few months he had flown more had more flights and more crashes than
00:55:59
all the rest of the people in this prize put together they started getting an idea of what the problem actually was and less than six months after Paul took
00:56:13
that drive across the Arizona desert with his family they won the Cramer prize it was easy just easy and much more impressive than that
00:56:28
was the second Kramer prize which is flying from London to across the English Channel and he won that a year later one of the most thrilling videos you can ever see is called the gossamer
00:56:42
albatross you can order it every child should see it it's one of the great romantic videos of our time it's done beautifully but so this was a
00:56:55
mile-and-a-half on land and this was 22 miles over water and when asked about it Paul said well everybody else was trying
00:57:07
to make an airplane he said we were trying to do a human powered flight so he didn't start off with an airplane as the idea and started off with the fact
00:57:23
that we had a third of a horsepower at max to run this thing and we had to do this and that and the other thing we had to find out what that meant and that wound up with something that looked extremely different than all of the
00:57:36
other things and bingo let's take a look at one way of looking at this so here's the effort and we can think of reward
00:57:51
going up that way and there's kind of a threshold there so here's what tenure committees and NSF loves they love problems you can do and
00:58:10
you know what they don't care whether the problems are worthwhile or not because Congress doesn't give a damn and can't understand the difference between any and Congress oversees NSF I'm on I can say this because I'm on to several
00:58:22
of the NSF advisory committees but basically if you want to get an NSF grant you don't write it for science you write it as an engineering project here is how we're going to do it here's where we're going to be successful this is why
00:58:34
almost nothing funded by NSF over the last twenty years has been interesting just hasn't been and these big inventions that we use today we're not funded by NSF at all they were funded by
00:58:45
our PI so the first thing we have to get rid of is this idea of looking for the keys under the lamppost which is not
00:58:57
where we've lost them all right so you can think of there being a curved
00:59:10
something like this that you put effort in if you're lucky it might have a peak like that and the McCready principle is
00:59:22
spend a lot of time finding that tiny little place there it's just above the threshold of being interesting but not way above the threshold of being
00:59:36
interesting so you're looking for the minimum thing that you can do that's qualitatively better and most good problem solvers are incredible at this
00:59:49
they have a homing instinct McCready would spend a long time thinking about where he should put 80% of his effort before he did a darn thing and he never had a failure and he solved an enormous
01:00:01
number of problems in his lifetime just because he worried about this because this is where you learn about what you need to know to do the real deal and as soon as you get to that doable
01:00:15
thing you've actually changed this curve and you brought the thing you wanted to do into range and all of a sudden you've got to break through so most of these
01:00:27
things are kind of two stages effects the hardest thing is to get the funders to pay for that first stage because it's not cosmic so one of the reasons AI research has not done very well the last
01:00:40
25 years or so is because the new are Pro DARPA is not interested in funding actual progress towards AI they're only interested in funding the robot combatant of the future something that
01:00:54
nobody knows how to do but they can sell that to Congress and nobody can do that so here's a couple of punchlines Susan
01:01:10
Sontag I love this one all understanding begins with our not accepting the world as it appears this is the hardest one this is where you have to do with this
01:01:24
exercise I do it every day literally not getting normal considered harmful because of course we have a zillion mechanisms we don't want to have to think every time you take a step right
01:01:37
so you can yourself by questioning everything that you do yet so you but on the other hand every once in a while you know take fifteen minutes and instead of doing meditation on a
01:01:49
flower which is also a good thing meditate on all the assumptions you're making about the world that you're just taking for granted for efficiency reasons and try and think of which ones
01:02:01
where the world would be very different if you made different assumptions the reason even people who are good at this stuff have to do it is because we all have the same brains our brains are
01:02:15
defective nature didn't make those brains for doing great inventions it made those great brains for survival and coping so we have to get around our brain science of course is one of the great ways of getting around human
01:02:28
brains that's what it was designed to do but individually we have to get around what's wrong with our brains so it's best if we think of ourselves as actually being unsane and that we're
01:02:39
dynamically trying to be stable with respect to all that stuff out there Red Queen said to Alex to Alice try to understand three impossible
01:02:50
things before breakfast we have a blind spot in our eye which if I had many audiences I do this with I get prove this to the audience by having them do
01:03:02
the experiment where a dot this big on the paper will literally disappear before your eyes and be filled in by your brain by material gathered from the surrounding stuff because there's big area in our retina that just doesn't
01:03:16
have any optical neurons and it's the realization that we have that blind spot in our
01:03:27
brain is filling in that helps us be saner and I guess the one to leave you with is this idea that if you realize you're in the pink plane then you have a
01:03:39
chance of escaping it but if you think the pink plane is all there is there will be no thought of escape because how can you escape all that there is thank you very much - any questions comments yes
01:04:23
well I you know it's so you get old they give you gold medals and the proper response to getting a gold medal is to thank the funders because they just gave
01:04:38
you gold back before you turn the gold medal so a good funder who is willing to back something without trying to rationalize it as well as the scientist
01:04:52
is going to do is the most critical thing in our field because it takes money like I said in the old way that Buck Rogers was this guy who lived in
01:05:04
25th century so the old saying was no bucks no Buck Rogers and so what changed in the 70s the thing that forces the arc spark to come into existence was the great ARPA funding of the 60s was put
01:05:17
out of business by Congress because of the Vietnam War it was no longer possible to do that stuff in universities funded that way so Taylor decided to take one more shot at this
01:05:28
and convinced Xerox to fund about two dozen of us to finish off this ARPA dream of personal computing and pervasive networks so that's the way that worked and the funding has not been
01:05:42
good since 1980 there's a lot of money out there but the money is tied to the perceptions of the funders as to whether you're doing something worthwhile or not and that is missing the point
01:05:55
the real way ARPA did it was to say hey we're willing to tolerate 60% failure if everybody is working on really important stuff because that 40% is going to change the world so this is a baseball
01:06:09
model we're not going to fire you if you don't bat a thousand now if you bat 368 like Ty Cobb did that's good so that's the way they did it and back in the Cold
01:06:23
War Congress didn't over over C ARPA the way they do now is the Vietnam War that changed every and not for the better their so that's the simplest explanation and what but
01:06:35
once this thing winds down you start getting all of these factors that I was showing it's our tendency to get comfortable it's our tendency to say hey if I can
01:06:47
buy a compiler I'm not going to make one even though that compiler might hold back some thinking I need to do about programming languages because with that compiler I get I'm only going to compile the thing kinds of things that it's willing to compile it already has a
01:07:00
conception of what it's trying to do and that's going to tie me to past conceptions and I might have a good idea so Park literally built every bit in
01:07:11
every atom of all of its computing machinery build all the hardware and all the software and it wasn't you know it was a lot of work but we realized that it was we thought it would be a better
01:07:24
trade off if we could pull a lot
End of transcript