I'm sure you I'm sure you would definitely but in the form of the
arrangement actually I have a release form for you the the thing that we're in
in order to first ISP and there there was NSF there's DARPA and NSF and
regionals but the idea of having the net take off for people that weren't DARPA
contractors I thinking machines we were able to get on the DARPA net because we
were DARPA contractors very difficult took six months of a full-time person to
get us on the DARPA net in 1985 and it was hard as hell and but the little
garden basically made it so that any old person and more than that not just
themselves but the way Ben Franklin always tried to make things that weren't
just you had to go to Ben Franklin but he made it so that you could go to other
people and enabling other people to create their own ISPs and I don't know
there are 400 ISPs now in the Bay Area in large part because of the little
garden the two things that were the what was special about the little garden is no
restriction of resale which was just basically forgotten everybody was
concerned you know the whole place will crumble you know you'll just start
reselling it and they'll just max out these connections and what we want to do
is we wanted to originally it was kind of a cooperative I don't know if you
heard that but it was what happened is John Romkey and John Gilmore and Steve
Crocker actually with pretty much three original people and Steve Crocker is
trusted information systems John Romkey FTP software and John Gilmore you
probably know him through Cygnus and EFF and such and they wanted a connection
at that point it was like $2,000 for a connection from alternate at that point
and in order for them to be able to do like kind of like resale it was going to
cost them $5,000 they said that's bogus for a T1 so they actually sort of they
knew Rick Adams real well and said hey listen can we do this cooperative thing
yeah sure and when when was this was 80 88 89 around there and Rick said yeah
sure don't just don't let expand too much of course Gilmore being Gilmore
said ah sure no problem well at that point he he ran a 56k up to what his his
house called Toad Hall and he called Mia I was working with him because I wrote
the software that hooks FIDO up FIDO net up to the internet and he said I got
this connection we really like to be able to get more people on it and so we
I got a copy of what was called Ke9 Q at that point which is this kind of TCIP
software that's written by Phil Karn and we set up this little 80 88 or was it I
think it's a 286 or something like that and put a couple serial cards in an
Ethernet card and all ran under MS-DOS hacked it up so that my machine at home
could call into it you know I had like another version of Ke9 Q at home and so
we were able to start selling slip connections which was something that you
really just didn't have the we didn't have the hardware the or the money to be
able to do so we were making these little boxes out of $400 you know at the
most to be able to do this so it got bigger and bigger and bigger and at that
point we said gee we really can't manage this on our own it was supposed to be
again it was supposed to be kind of a cooperative and not that many people
were helping out with the co-op so we hired Governor Tom Jennings who I had
known again from the FIDO days because I did the FIDO network and he had
just quit Apple and he was like pissed off the corporate world and just didn't
want to deal with that whole scene at all so he said yeah this sounds cool I
can just sort of work this tiny little company and you know make a couple of
bucks and you know do my skateboarding and all that other stuff and we gave him
a cut of how much people would do you know pay into this thing it's like ten
bucks per person who plugged in and it got bigger and bigger and bigger and
bigger like how many people actually had slip accounts at that point from you
guys this at Toad Hall what happened is we outstripped the NOS boxes we had
about probably about six NOS boxes each with four serial cards into them so
there's about 24 lines that were going into to John's place which got a little
messy you know the the phone company guy and it was a woman as well who came in
and said oh yeah another phone line for Toad Hall and they got to know us real
well and we had like you know punch blocks and telephone lines going through
his basement and the whole thing after that we we graduated to like a
Livingston port master which does about 30 and we far outstripped John's little
poor little 56k link so as soon as we the next step was co-locating with a
with a company called MFS metropolitan fiber systems and where we put in a
bunch of our equipment at 444 market and then we just moved everybody at Toad
Hall over there and that was kind of the start of the real business we were
actually signing three-year leases and getting really scared like gee what
happens if our company collapses in six months but where was Toad Hall what city
it was in the Haight Ashbury district yeah that's where John's John's place is
but it's named after a restaurant in Palo Alto no no the little garden is
named after a restaurant yeah oh but I mean the little garden is a later
incarnation no the little garden was kind of the co-op and see what happens
all of our you know Steve John and I and everybody else would meet at the little
garden to sort of talk about how we were going to work on the net and get it
going and do what we need to do and things like that and our last meeting
there we basically took over the whole restaurant we just showed up and the
people there just went what are we gonna do with all these people and so we just
ended up all these tables so we kind of like outstripped the restaurant as well
so but that was the name of it that's what we took it
right the East Coast was all very stodgy you know you had to have official
contracts and the West Coast you come out here and there's the little garden
that was just taking over the Haight Ashbury flat to be able to get the
internet to everybody it was just that sort of grassroots thing that the West
Coast does beautifully I think that's why you're seeing the internet happen in
the West Coast well we invented the PC why not the internet no I mean the home
brew computer club and the whole and the community community memory and all this
yeah no no I mean we you know we fell in signs and everybody else you know or are
just involved as in this stuff as well you know he got his community memory
stuff going and he started using this sort of infrastructure he didn't use us
but he was using his own internet infrastructure Wired magazine their
first connection was through us I don't know if you've talked to them and they
mentioned that at all well Brian Bounder I know I knew Will Kretz for a little
bit here this is before they moved this is an original
location people brick warehouse like place they moved actually a couple more
times after this yeah yeah they they went to us and they had one T1 that went
into us and of course once they put up hot wire or Wired magazine wherever the
site was originally called I can't remember but it was like just hose in
that line that in Ayuma I don't know if you remember they went through screws
net they connected to screws net which was one of our clients and that was
another line that was just that was our two highest usage lines as I am and the
Wired magazine site and finally they just kind of outstripped that and they
connected directly into sprint what dates give some landmark dates for some
oh jeez you moved to the city must be around 1990 I have I have actually we
have archived all the history is the web came later than that your you did a ways
database in your basement what was that it'd have to be 1991 because the ways
code hadn't been put out until April of 91 and that's when you ported it to a
PC you know what my dates me wrong yeah because I'm thinking that we've been
doing this for about four years now it's 96 now so it must yeah it must have been
91 92 because why it started pre web and they must have had some interest in the
internet at that time I mean they you know I remember they had some stuff they
had an FTP site and gopher originally I don't remember them having a website
original I remember them moving this well that the web didn't exist when when
they started
why yeah I mean the web sorry it didn't exist commercially or in any well I
mean even by 92 you could say the web existed but 91 it's gleam gleam intends
I have some demos or whatever at high telnet
tell that for me was one of the major turning points because that got us out
of NCSA telnet when they did their own TCP but it allowed you to make
applications that would run on the Mac platform and so it could get you out of
just being unix only then you can start to get to windowing and it was quite a
while before the PC had TCP IPs that really were they had real it's not
usable yeah they had really expensive stacks like FTP softwares which were
$400 PC for the stack and they made their money that way but and they had a
really nice stable stack also it just took up a lot of TSR space and that was
something that DOS machine just kind of handled very well there was what was the other thing there was water
low TCP which was a stack that you could use under MS-DOS there's Phil Karnes
thing but it was basically not a stack that you can run other things on top of
and then there was Novell's land work place. So is it fair to say that the little garden was the first modern ISP that really helped not only itself to bring lots of users but also brought hundreds of other companies into I remember John
during saying Brewster you know there may be three or four companies doing this
stuff now next this time next year there will be a thousand yeah he was
absolutely right if not low this was in 92 going from 92 to 93 there was a
there was an internet world that the first internet world was very interesting
to me and that the real people were there one person's comment is there are
more CEOs in this room than anything else these are CEOs where they didn't
have any other employees right and the booths almost all of like science
projects you know this is you know the table with the with the tablecloth right
the little stand-up you know this is who I am people that had printed out a
business card because they had to name the company before they went there and
it felt what I you know heard about the homebrew computer club and and we were
one of the more established companies at that point this is early ways but ways
ain't got already existed so 93 but the little garden was from my point of view
the first that was trying to bring it outside of the research and big company
networks at that point all the big high-tech companies had the net and
they're all doing intranets because they were Unix the Unix based places but to
try to get it out to the wild skateboarders and the like it took the
little girl the companies before that well basically around that time to see
imps the MCI no sprint wasn't around at that point they were doing they weren't
doing internet MCI we were their first customer and first what's called
commercial customer we got through their government division because we were
partially NSF funded up in Portland we were the company that well indirectly
actually rain that was one of our business partners got South Africa and a
bunch of other companies or countries online and the NSF was sort of funding
that to people help that happen so we were able to get sort of like
introductions to get get connected MCI but the the companies around there was
like the NSF sponsored regionals which were things like Barnett you know not
BBN yeah sir nice or not a couple others but and then then the only other sort of
like commercial net at that point was alternate or unit and they were they
weren't interested in trying you know doing slip connections or they weren't
interested in doing like low-speed connections that would be affordable
yeah the dial-up idea really hadn't really occurred to people experiment and Rick Adams
wrote the slip code you know on the Berkeley Unix so you figured you know
he'd be one sort of like interested in pushing this no alternate was like you
know big companies they want like you know the HP's and the apples to connect
a lot easier to make a business based on those yes we've seen but we still wanted
to make sure that that any Tom Dicker Harriet could plug in at a reasonable
cost and our first prices were the way we set it up is you would buy the phone
line on each end this is the way so this way we didn't have to incur the cost and
billing handling everything like that and we would charge you $70 a month for a
14-4 connection which was something that was still a little high I mean you know
it's like two or three times what cable TV costs but we had a lot of diehards
particularly with people who just couldn't get connected you don't have
these $25 slip accounts you know back then and with our our connection we would
give you an 8-bit connection so you could you had all these people who'd
like we're like serious hackers and have like three or four maybe ten machines at
home and put that behind that one connection now to the internet as
opposed to today where people just are plug-in and serve this is the era when
code was distributed on the internet in source code when we first started
putting out ways we put it out in source code not in binaries the idea of putting
out binaries was it was thought of as corrupt because mostly was wasting
bandwidth I think mostly is there was this club of sys ads that their job was
to bring things down compile them and get them to run well my issue about
binaries is that we didn't know who compiled them and what viruses are in or what back doors and such
so you never even you always compile from source so you can say I know what
went in that code and I know that you know so-and-so is not gonna screw me
over by compiling a binary and me just sort of automatically plugging into my system so that was the era and then Mosaic started
distributing code in binary forms Mac stuff had always been distributed in
binary but the Unix code not nearly 90s the all sources news groups are always
bountiful and yeah and full these days it's like a desert I put out one message
to one email list which was ways talk and about a hundred people were on that
list at that point that there's this release alpha one of ways and and here's
the FTP address and there are 70 databases in the directory of servers
there's a directory servers 70 databases most of them on a connection machine
some of them on that MIT and you could download this stuff and I just watched
and that posting was reposted on lots of different groups and the use just
started going first the first people were the people that were willing to
port because that was the first thing that had to happen it ran on Sun's and
that was about it and people ported it to everything and then people started
setting up databases and they'd send in their their registration to be part of
the directory of servers and it just spread in that sort of classic
exponential approach but it was very interesting to see the freeware
community just be alive in a well community of people saying this is cool
it's based on open stuff it allows us to do something we've never been able to do
before go for it and it and it just spread and the freeware community is
still alive and well today
Linux yes the freeware community is not corporate in that respect I guess so
it's gotten big you have a third party yes signals huge companies now they
were just tiny little companies that are supporting just free software yeah that's
weird yes but it's a really needed thing because they have such great tools like
GCC it's like one of the best compilers out there and you need something to
support wasn't that the whole idea of the free software I mean it's not
Richard Stallman's thing that the money to be made is from support not from the
code right I mean that's fulfilling exactly what and that's what what
Gilmer actually did is create a company called Cygnus and actually is making
money off of that right now is supporting that but he's supporting
things like people like NASA and and Sun and various other folks and also it was
interesting and he's doing all these really the company is doing all these
interesting little hacks and every time there's a new micro processor comes out
that company needs a C compiler immediately and the best thing out
there is GCC because it can cross compile so well and there's not all the
licensing that's that's involved in being able to get that to go but yeah
that I remember I remember probably soon after that I probably contacted you and
we reported ways to system five this is great my first meeting with Tim of
course you know in the sort of West Coast way you don't have a meeting in a
meeting room we had a meeting in a place called the right spot which was in south
of market which is this little divey diner and is completely great and then
dim came in and put this thing down on the desk and it was this telephone I
said oh a cell phone he said no it's a radio phone and it's just this guy hacks
radio and he said well I've got this connection in they don't bill me to be
able you want to dial anywhere in the world and it was a ham radio set that
you could dial that can patched into Pac Bell and this was a different kind of
guy I then got into his car and you know this we're normally people just have a
stereo radio and so there's all the ham radio sets to be able to tie into lots
of different networks went over to his house and he said quiet he went down to
his basement in a place that he doesn't live anymore said this is a bootleg radio
station we I know all of the people that are this where the different trucks are
that go and track radio stations and he knew the frequency to use such that
there was a heel in between him so he didn't actually interrupt the frequency
of a real radio station but it was a pirate radio station that broadcast at
extremely high wattage and then he said oh and here's the the PC running the
ways servers for the emergency broadcast system right because you need to be able
to get information to lots and lots of people if there's an earthquake and all
that was done through voiceover radio but that doesn't make any sense so he
had a feed of it going into a way server that could be accessed over the internet
of course over a telebit trailblazer going to a phone line to the little
garden and it was getting hits all the time and that was fighting that and so
this was the sort of theory and living living this that happened in San
Francisco when a new technology they didn't go and say oh that's the DOD it's
horrible but the San Francisco community just as it did with PCs when said this
is something that we could use for the people and went and did amazing things
so where things were done with radio and Tim was instrumental in building a lot
of radio stations is now done on the internet where that's the mechanism to
get at large numbers of people and if you are a small-scale producer and you
want to put something out it's the internet and that's I found that very
much inspiring after coming out of more or less the the DARPA world we scam so
much equipment out of till bit wiring up the third world the peace net is here
in the Presidio oh they've moved that they've moved yeah you know me
because a friend of mine works at PeaceNet I've been down to some site
South market well it's actually over here it's in the Presidio or IGC you're
gonna see the Presidio just turn into a think tank of creative high-tech this is
going to be a university type atmosphere among people in their 20s and 30s you're
gonna just see amazing things flower here because we've got this place we've
got the goal here we've got a campus atmosphere in San Francisco right now
it's fairly difficult and so yeah as it so what does the workstation do right
we really wanted to separate out and use the PCs for what they're good for the
web never has right it's used it basically as a dumb terminal and we're
seeing them incarnated in horror and in hardware it's a sort of on so the
workstations on dynamic folders find information for the user the idea is
dynamic folders are things that find information and put them on your machine
for you important right the web really doesn't work for a lot of users because
it's you have to search for it and that's just not mainstream using
information servers so that how does it do it different interface possibilities
you can use newspaper format email format geographically books on a shelf
there are a bunch of different ideas on user interface motifs the Netscape has
no metaphor I sure they do they have the metaphor of an old telecom program
advantage in remote versus local filtering right do you want to do it
locally or do you want to do it remotely what balance do you need between those
two the web doesn't have that stuff local caching of documents how do you go
about doing that such that it's robust the web wasn't really designed for
caching in line we've run into some problems time to live things that kind
of local scoring of competing servers if your computer goes and contacts several
then how are you going to know what your user should see budgeting the user's
time is money the servers role probing information servers examples of
information servers file-serving printers faxes bulletin board services
da da da newspapers movies bulletin boards archive searching banking
services you know all the stuff that you'd kind of imagine we've seen now and
pizza ordering everybody really liked it you could be that that was sort of yeah
we want to do pizza ordering right because there's a bunch of geeks at the
original but there are library catalogs and you know all the things that you'd
be able to do and that aren't basically there now navigating through the
directory servers so what is the role of the directory of servers which I still
think is an extremely important role how do you find resources what's the metadata
behind them and we'll probably we're trying to put some of that together
through the archive servers that rate other servers with speaking of has one
of the gaps that I would see is okay the inner neck which is the name service for
doesn't make any attempt to categorize that's the logical place to collect the
information but it went public too soon they take all sorts of information on
who you are and what your company name is nothing about what you do no it's a
crime it's a crime I mean you know I have to do is get them to say add this
little thing to your template driving it and making it part of the
infrastructure we've basically got a very thin piece of infrastructure out
there yeah well you know there's time so role of editors right how do you use
both human editors and computer editors right that's the the wave that will
happen then markets versus hierarchies using Silicon Valley so the idea of what
do you use Silicon Valley for and so that's why you know basically the things
that have come about you know work really well here that they don't work in
Europe or in the East Coast so how would you go about using Silicon Valley how
server companies can make money the protocols role so here's how why do you
need a protocol and then there are different pieces why open protocols
for both wider acceptance hardware independence protecting the users
privacy and including in that Steve Jobs way why ways is going to change the
world turned out to not be ways turned out to be gopher and then web but the
basic structure I think is still in place and it didn't come from me none of
the stuff is new by me you synthesize related projects there are related
documents that all sorts of really interesting documents from that time
catalyzing a market for wider information servers that was the one of
the original blurs that came out in 88 that was the pitch to thinking machines
we need we need to catalyze a market so people making money to for these
information servers wonder if I could find that how to do searching on you
know supercomputers that now become PCs so copy serve many tell 16,000
information services phenomenal success netlib the Switzerland system the only
thing I've got for the web is Switzerland system still assessing this
system that was all I knew in at that time Switzerland system still assessing
this system Lotus information brokerage companies and hypertext so to what was
going on in the Xanadu world and the different hypertext systems that were
all bouncing around help systems and the like so this was after I got to Apple
and pulled together all the companies yeah kind of fun did Tim see this
so well yeah no it went out with every ways distribution it just yes you know
there's this there was a few things that I thought were important to put in
every ways distribution one was the ethics of digital librarianship now that
lots of people are becoming librarians and they're seeing usage logs how
should they ethically deal with that responsibility there is proposals for
the payment structures the URL structures and so the idea behind what
ways was about and that just went out with the yeah I recall there's an awful
lot of stuff in the ways distribution much in the same there's an awful lot of
stuff in the emacs distribution so that I think you might find fun just you know
when you're driving around or you know on a bus or something I think you might
find it fun it's it's interesting to see how you know it's true people thought
through all of this so clearly decades ago well I mean other than other than
Nelson who really does so in such a vacuum yeah oh yeah I've got this
wonderful signed copy of computer lip from you know it's just so ancient book
with just all the hand scrolls Ted Nelson you know he wrote to me when they
night announced the the archive he said oh another piece of Xanadu being done
but you know in this sort of lamenting way and I said of course Ted we're just
fulfilling all of the things that you went and suggested so many years ago so
should be no surprise so that I think would probably be in terms of early
stuff that might be interesting or useful to you and you say you did do a
history of some sort or I've got different history files that most of them
are home okay yeah but a history of ways no no if you don't really just the
different segments like the Apple segment different company selling it to
AOL integrating we're all the same building it's right yeah they moved into
the Waze building yeah he came over here he's giving us some hints on how to do
the crawler for just for your video you might want to just you know there's the
robot and they're sort of going and archiving the web which I think is
somewhat interesting towards your web history it's a different form you're
doing an anecdote history the idea is to be able to collect the source materials
so that historians and scholars will be able to do a different kind of job than
what you're doing no no but you realize that they should be interconnected in
other words ideally the anecdotal history like ours has links that rich
links that lead you to all the source material I mean it's just another view
of the source material basically basically a Sun machine donated by Sun
that's basically connects to several hundred web servers at any particular
time gets a page waits five seconds get another page and just goes through all
of the websites and it's an ultra just running underneath somebody's desk that's
able to basically keep a t1 busy sucking about a million pages a day the ultimate
robot that's well it's just like Alta Vista or a web crawler but we're trying
to get everything including all of the images everything so you can recreate
the web what way it was at any particular time so you can dial the way
back machine and experience what the web looked like not but with all the CGI
scripts and there's certain Java things but you won't be able to do that but
you'll be able to try to use the web the way it was or find a copy of record or
what did a particular person say at a particular time or how can you cite a
document if the documents go away every 44 days the mean life of a document on
the World Wide Web is 44 really that's what they say it based on one study we
don't have our own numbers on that yet but it seems to be about right but if
you're not picking up a lot of CGI stuff I mean most virtual documents will not
happen then right we'll get whatever you could get to at that time but we don't
know how to go and type anything right into a search string right so anything
that's in a database behind we cannot get so you're grabbing all this stuff
down the pipe what kind of bandwidth you have here t1 and we're we'll be bumping
that up but that that's enough to give you a million pages a day and if there
are 50 to 100 million pages and that's how long it'll take and once we get
better at it we'll run it at the San Diego supercomputer site and download
quicker so I'll show you with the storage mechanism
oh so there is putting a tape away you see where you are in the process now
we've gathered 600 gigabytes of material and it's sitting on this robot and a
robot that's down in San Diego and you do have a backup of currently we've got
only one copy though you know of course the you know as we're getting going more
and more when I stand by this thing more and more of this thing will be backed up
in multiple places for faults in terms of technology but also political faults
the library of Alexandria was burned so we need to be able to preserve our
digital history in multiple places under different regimes so basically it since
these things require attending every ten years you've got to move it forward to
the next generation of technology or you it dies so you need an organization to
keep it going and I will endow the organization with enough money to keep
the bits alive forever so any bits that are donated to the internet archive I
will make sure has the money for every ten years will keep it alive and when
it comes time to work with the National Archives of the Library of Congress so
they can deal with terabytes and then we will provide a copy to them but right
now there are no institutions that are doing large-scale digital collections
and so we're I think the first that is starting to build the core collection
that can become a digital library at least of this type of material and this
is all it takes this is a another Sun computer donated by Sun 20 gigabytes of
hard drive space that's used as a cache so a bat machine in the other room
gathers web pages and puts them in a hundred megabyte chunk transfers it to
this set of disks and after they've been here for a while they basically get
transferred to tape and that way we can cost-effectively store the whole net
which is estimated between one and ten terabytes and that's well and growing at
what rate we don't know we don't know um you know there is estimates that it's
doubling every seven months but frankly we just don't know so we have not
finished the first complete crawl yet but it's getting there
some donations of all the HTML from early 1996 and we'll get some donations
from what one of the crawlers open text had from 1995 that'll be the earliest
web we have our crawler has just been running for a couple weeks but it
gathers about 10 gigabytes a day so it seems like a fair amount but there's a
lot to go but you're sure that that's above the rate of increase yeah no at
some point everybody's gonna have their camcorders on the internet and at that
point we can't keep up but right now we think we can be comprehensive so we had
a request for some of 1992 net news and so we've been and that we've got from
CD-ROMs on our robot and so people come in through FTP contact this machine if
it asks for a file that's not on disk it's on tape then it takes it from tape
puts back on disk and then serves it out and we're building services to be able
to data mine this we don't think that this is a crypt or a time capsule it's
to create new types of services that if you had the whole net in a place where
you could do active manipulation of it to find trends and new ideas in it or
changes over time then you can be a smarter organization and that's what
we're trying to build
I don't know, through the smoke glass. Oh yeah. So somebody requested a file.
I like this thing. It gives a visceral notion how big the net is and there's
something working. Although the fact that this thing is smaller than most
refrigerators is also an amazing feat itself. And it can store two terabytes
which is about one-tenth of the text in the Library of Congress is in this box.
Ten of these. Ten of these and you have the text in all of the books of the
Library of Congress. So something very big has happened in the sense that
computers now are able to manipulate all published material inexpensively in a
way that kind of basically put computers in the library and help where you can't
read every book. Your computer can to go and figure out what it is you should be
reading next and find this thing just amazing and exciting. You'll be able to
pull off new services that never been possible before. But do you think though
that I mean we're at a funny juncture where the technology of course this can
grapple with all paper material. Ten of these would take in a good portion of
what people want but as you say as soon as everyone has their camcorder on the
net we were right at that funny threshold before the flood of digital
information right? Before richer media types and large part the net the
internet is got such small files because the internet is so slow. Which will keep it
honest for a while. For a little while but the flood bandwidth will come whether
from cable through wireless through phone companies through every which way
bandwidth will come. But text everybody can only be typing all the time and that
is small. Oh yeah text is a fractional portion of it. And it's a very dense
representation of knowledge so you can manipulate it. Computers can this will be
smaller and faster. Isn't it neat? Yeah both CERN and NCSA they have. Where these things used to just be in big
shops supercomputers cost hundreds that millions of dollars. Now you can do it in
a little house in the prairie here in the Presidio. You know where it took a
hundred and forty million to go and build a library and new library in San
Francisco. How many times could you get that? It would take a couple of these.
So it's a good size library. Turns out the Library of Congress isn't that much
larger than the big libraries in terms of the number of volumes. Big
libraries are five six million books. Library of Congress is 20 million. It
just basically paper doesn't scale. But computers do. It's fun. They had a Cray
which they finally got rid of when they switched to linked workstations virtual
mainframes. The Cray spent its last days as a tape server. All it was doing
serving tapes from the big underground library. It cost $500. That was the fastest
computer in the world ten years ago. Cost $500 and it's not worth turning on. So
just imagine what terabytes are going to mean to our kids. You know it's like dad
that's just a terabyte. Hopefully we actually have some some of the old stuff
to be able to put on it. I love that statement. I wonder which one's
going to be true. If it's not digital it won't it won't be saved. It's one
statement and I read another one. If it's digital it won't be saved. Well
but without people doing what you're doing clearly the just lost early nets
lost. Yeah there's a there's an advantage right. You could reuse the tapes. Why
would you why would you want to keep it? Well when we go into space we can
recapture some of them from there. Yes we'll have to hurry though right. Have to
go a little faster than the speed of light to be able to catch up with the
old I Love Lucy's. I Love Lucy was archived but the early Tonight Show all
gone and you know you say ah good riddance but Ted Turner bought the early
archive a bunch of early TV programs for one billion dollars recently I've heard.
So there's some value in the stuff and he's been doing very well by by using it
but also the early stuff is got people's dreams. The early people that sort of say
you know the early people in the web like you doing your dinosaur exhibit
you've got this ability to be able to do something new and different and at some
point people will just say oh well you know I can't do that and they'll just
sort of focus in on all it can do but in the beginning you have just amazing
things so I always kept the early logs of ways I thought those would be the
most valuable things is one of the questions that people ask the net and so
I kept all of those logs. Oh you have those. Oh yeah of just what do people look for
what are people dreaming that the technology can finally answer and right
now already the web is sort of you know stagnated you know people's idea of it
is really focused down to sort of the stuff that the technicians have given
but it's the early dreams that is what we should be trying to live up to and so
trying to keep some of that alive. So that was an exciting period when you
was it any Newton and the apple for you with ways when did this when did you
decide this is gonna be my project rather than just a random thought? 1988
when I decided what context were you in the bath driving in the car? It wasn't a
Eureka. The question I had was how do you find a
field supercomputers basically in by 1988 it looked like we'd come up with
most of the reasonable ideas and so it took started ten years at least from my
point of view for the phase that I'm usually involved in. I wanted to find a
field that would have interesting problems to solve for 25 years because I
didn't want to have to change fields again and the only one I could come up
with was helping answer tough questions for people. And what's my bank balance?
Little computers will do that but answering the hard questions should I go
back to graduate school? Should I marry this woman? What book should I read next?
Da-da-da-da-da. Hard questions want to be able to help make computers answer those
that was the goal that I started with and Waze was the first step which was
making open protocols work. Turned out to not be the Waze protocols but
that's fine we won anyway and thanks large part to Tim Berners-Lee. Then so
the open protocols were necessary then is building an infrastructure for
collection of the content and the usage of the information. That's step two
that's what we're involved in now. Step three is to help answer the hard
questions. Tracking people's histories so you can learn from other people's
mistakes or other people's examples and basically by having computer records of
what's going on and how other people won and lost, how their careers turned and
twisted, what worked for people, what didn't. Hopefully we're building a
resource that can play some of the augment the role of mentor or priest in
small towns to answer the hard questions. We're just in phase two so I'm just
surprised that we're in the end of 1996 and this is as far as we are. We haven't
got a long way to go. The open protocols worked but unfortunately we screwed up
some of the distributed nature of the net and so we're trying to do step
number two here. Then I'm gonna move on before I'm you know when I'm 40 then I'll
start something else. Then what are you gonna do? You don't want to know.