Future of Coding - As We May Think by Vannevar Bush
Episode Date: July 22, 2025A classic "80-year predictions" episode. How did we do, humanity? Links $ Patreon We're considering changing The Name of the FoC community, podcast, et cetera. Transatlantic Accent, and all the ships... at sea. This episode's advertisements: Ex-Lax, Mifflin, et cetera. The New Media Reader is a good read(er). Blue Prince is surely everyone's 2025 GOTY. You'll find ample discussion of its twisty design in our bonus feed. Oppenheimer was probably my least favourite David Nolen movie. ENIAC was, for a time, the computer. Engelbart The Polaroid instant camera dates back to the 1940s! Google Glass and Ray-Ban Meta, sigh Wikipedia's List of Existing Technologies Predicted in Science Fiction Apparently microfilm and microfiche are both instances of microform. Boy Milk, no explanation needed. Sean M. Carroll, physicist. Claude Shannon, information theorist ;) AI Is Not Your Friend appeared in The Atlantic on May 9th, 2025. Amazon Mechanical Turk is "artificial artificial intelligence". Awesome GitHub lists Daring Fireball and Kottke.org are two long-running blogs that helped popularize the "linkblog" form. The Wirecutter was our previous top pick for product recommendations, but their history of shilling junk and our growing concerns about e-waste have led us to downgrade them in our rankings. Devine (the other Lu) maintains a pair of wikis, one for hundredrabbits and another for research and projects. Alexander Obenauer also does a lovely job publishing research notes. Unison and Dynamicland are two strong visions for the future of computing. Music: lemon (tussed/screwed because everyone listens at 2x) ! Send us email, join the FoC community, and find us on-line: I: 🐘 🦋 🌐 J: 🐘 🦋 🌐 K: N/A L: 🐘 🦋 🌐 See you in the future! https://futureofcoding.org/episodes/077Support us on Patreon: https://www.patreon.com/futureofcodingSee omnystudio.com/listener for privacy information.
Transcript
Discussion (0)
Spanking won't be necessary.
Welcome back to the Feelings of Computering podcast.
Wait, what do you mean the Feelings of Computering podcast?
Wait, is it actually, is it Feelings of, Feeling of Computering or Feelings of Computering?
I think it's actually Felt and Korderoi.
The Felt on Kordoroy podcast.
That doesn't sound right.
How many of these have you come up with by yourself,
the FOC, and how many have you asked ChetGBT
to generate for you?
100% by myself and I don't come up with them in advance.
Okay, okay, so yes, there's been a big thread
about renaming the podcast on Slack forever ago,
and then there was a new re-thread about renaming the podcast.
And I think it would be good to talk about, like why, for those people who are not, you
know, forever online, why would we want to rename the podcast?
Now I'm going to correct you already, Jimmy.
Oh.
It's not renaming the podcast.
It's renaming the whole kit in Kaboodle, possibly excluding the podcast.
Maybe the podcast just keeps the same name, but the Slack community and the meetups that
I, you know, I and Jimmy have no affiliation with, but Lou, you are affiliated with the
meetups. Maybe those get renamed.
I'm not sure about that.
I think the Meetups are going to keep the same name probably.
Yeah.
Travesty.
I mean, I am currently involved in them, but I wasn't, it was this guy Anselm, right?
It was sort of doing maternity cover for Maggie.
And I said I would like support that and help that.
And except I haven't heard from Anselm in a while.
So I need to go hunt him down and say,
Anselm, when's the next future of coding meetup going on?
But maybe if Anselm is AWOL, I could just try and arrange a Feelings of Computering one
instead or whatever it is, whatever it is that it's called.
What, what would you, what's, what's your favourite, Lou?
What's your, what would you want it to be?
My favourite?
Yeah.
Okay.
Well, here's the thing. I think people have been talking about renaming the podcast and community, etc.
I guess the whatever this is for a while now. I think that's because of a few reasons, right? It's all evolved quite a bit over time and future of coding doesn't necessarily capture
what we're all doing anymore, especially since all we do is look at things from the past.
So it's less about the future.
I mean, you might say, oh, you need to know the past to know the future, but I don't think
you could even
go that far here. I think the other thing is, future of coding for many people, including myself,
has this weird vibe to it. I guess future of coding, it sort of sounds a bit like marketing. You know, it's what you might expect some Silicon Valley
startup company say that they're bringing to the world
the future of coding, we're reinventing coding.
Some people have pointed out that like futurism,
was it called futurism?
I don't know.
The word future is often pretty
Prevalent in some more dodgy things, right?
Like and and I don't know much about that, but I can see it, you know, I can see it
it's like
out with the old we're replacing everyone with
with new technology and
Yeah, it has a bit of a eugenicist stink on it.
Right.
There's a nice kind of historical dovetailing of futurism and certain
other political projects that are kind of gross.
Yeah.
It's like, what kind of utopia are you hoping to create?
A good one or a not so good one? And I think, to be honest, this is why there's like such a mishmash of opinions on the new name, because
it's not necessarily that there's a new name that everyone wants, it's just that people don't want
the old name, right? So it doesn't matter what the new name is so much.
It's just that the old name may need to go.
So that's why I guess, I don't know, it could just be one of those acronyms
that doesn't mean anything anymore.
Like BP, BP British Petroleum doesn't stand for that anymore.
They like, they, I'm not saying that they're like good role model as an organization.
I'm just saying that they did it, right?
They didn't want to be known as British Petroleum anymore.
Wait, you're saying we should just like learn everything about BP and follow their playbook.
I think that sounds like a great idea.
Not what I said.
They're famously ethical and good for society.
Good for the environment, yeah.
Yeah.
I think the latest one I've seen is fear of clicking.
Like, FOC could stand for fear of clicking.
Yeah, that's, I quite like that word.
It could also stand for failure of community,
which is, if we're not careful,
a possibility that is on the table.
So I'm curious why you said renaming everything but the podcast, Ivan.
I thought definitely the podcast was on the list of things to rename.
Well, I don't know.
This isn't my decision.
I reject all responsibility and accountability for this.
Yeah, see, that's the, I mean, ultimately, I think this is one of the things I'll say, like this, you know, the threads that we've had on this. Yeah, see that's the I mean ultimately I think this is one of the things I'll say like this you know the threads that we've had on this it's like
looking for unanimous agreement more or less on like everyone is happy with this
name change. No name change we ever make would be happy everyone's gonna be happy
with because that's just the nature of changing things right there's always
gonna be people who like the old better than the new.
I think if we're going to change, I have no opinions on the communities, like the meetups
changing their name or not, because I'm not a part of those. If they want to change their name,
great, do it, right? Like that's awesome. The Slack, you know, we're kind of more involved with,
obviously, admins on there, so that makes sense that we would be involved in changing that.
And then the podcast, something that the three of us do.
I feel like if we want to change the name, we should change it because it's like,
hey, changing it gives us a new outlook on the
the thing we're doing, this little collective act together.
We maybe feel it better reflects what we're doing. We feel it better. Maybe it's more inclusive
Maybe you know, it has some properties that make us feel better about the endeavor we're undertaking
So yeah, I know Ivan had put out feelings of computering as an April Fool's joke. I
Really like the feeling of computing. I think that that honestly
does a better job capturing what I care about and what I want to talk about and what I want
to do than the future of coding. If you look at like Twitter search on future of coding,
you'll find tons of people who are all these AI company shill things going, Oh, the future of coding is
English.
And before AI, it was the same people saying the same things about web three.
And before web three, it was the same people saying the same things about VR, et cetera,
et cetera.
Because there's so much money in technology, and there's so much money in technology and there's so much money available to people who do a particular kind of get rich quick scheme and play on hype and hope and try to solve problems in a very superficial way but drum up a lot of enthusiasm around their solution.
Like that name future of coding is poisonous.
name Future of Coding is poisonous.
We've talked about that in the past on the show as well.
Like people talk about Future of Coding with a small F and a small C.
And they also talk about Future of Coding, like the community, the podcast,
the meetups, and I guess that's been me trying to distance ourselves from some of that.
So yeah, I guess changing the name might be a more formal way of doing that. I think it's funny that Jimmy's version of the title, the new name, is like a sort of
more serious, sensible version of Ivan's.
Like, Ivan says,
Feelings of Computering!
And Jimmy comes with, Feeling of Computing. And Jimmy comes with feeling of computing.
For what it's worth, I was totally down
with feelings of computering,
but Ivan said it sounded too goofy
and he didn't want something that sounded goofy.
Oh, so I've got it the wrong way around then.
Yeah, I was down with it.
I was like, oh, let's do it.
That's fine.
And then Ivan was like, oh, it sounds too like, uh, uh, well, I can't what I would have said. It sounds
too like doji too doji. I know what you mean. Yeah. But I can't use doji anymore. I mean
what it used to mean. It's too gauche to say doji. Too gauche. It's bougie to say doji.
Yeah. Too millennial, I guess would be the way of putting it.
Oh, it is very millennial-y, isn't it? Oh, God. Oh, no. I'm completely blind to the ways that I'm
like a lame millennial, you know? Yeah, I do think feelings of computering sounds distinctly...
It's very millennial-y. Well, actually, you know, millennial.
We need to get an actual Zoomer on the show
to help us with this, I think.
That's my feelings of computing.
I mean, I think that that's one of the things that, honestly,
really connected me with computing, with programming,
with doing things on the computer.
It is an emotional experience for me. It Like, it is an emotional experience for me.
It's something that was an emotional release for me.
I, you know, was, I started programming when I was in a not great state in life as a teenager.
Family life was a little messed up and it was kind of an escape.
And I still see it that as like a therapeutic thing for me.
So, I think, but I think the things that like the papers
and the things that we talk about that are most interesting
are people's experience computing,
it's people being able to express things with computing,
like that's the kind of stuff that I find fascinating.
So like that's why I would like to do,
I would love that name change personally,
because I think it also can reorient us on like, well,
what are we wanting to talk about? And I think it also brings in these elements that are not
computer science, these elements of social sciences, these elements of other perspectives
on computing. It kind of opens it up to say, hey, this isn't the hard science mathematical thing is the only thing we can talk about.
We want to talk about computing more broadly.
I like it too. And I think I think part of why I like it is that it's not necessarily disavowing what's happened so far.
It feels more like a progression, you know? Like this... it started out perhaps with more of this forward-looking, science-y, HCI-y, UX-y outlook.
And I guess through exploring a lot of that, the three of us have realised what we find interesting and useful to talk about,
and by switching the name it's just kind of an acknowledgement of what's happened naturally.
I mean, already it's made me think of different papers that we could look at,
which I think would be really good and worthwhile to look at, and I just didn't think about them
before and it's... I feel silly for not thinking of them before but absolutely I am sorry I
forgot to give lemon her treat that she gets at this time every day totally
forgot and now she is right here asking me for this treat so I'm sorry to
interrupt lemon it's good to get a lemon break so early. Lemon
Lemon
Lemon
Lemon
Lemon Lemon Lemon
Lemon
Lemon
Lemon
Lemon Lemon Lemon
Lemon
Lemon
Lemon
Lemon Lemon Lemon
Lemon
Lemon
Lemon
Lemon How do we even do a... what's it called? Oh my god, what's it called?
I'm not helping. What's it called? I know what you're thinking.
It's that thing you ride around on, right? No, no.
You personally?
No, no, no, what's it called?
Where you do like, oh shit, it is what you ride around on.
It's a segue.
Yes, it is.
That was what you were trying to think of.
How do we do a segue?
Yeah, we can't do a segway from the show to the show.
Well, I'll just talk about something totally not relevant.
It's about Lemon still.
So Lemon has been getting a little chunky.
Oh no.
Which is pretty normal for Beagles.
Beagles have a tendency to balloon up.
And she also has gotten way more food motivated.
So I have two little like I'll have like one little food things here.
So when she was a puppy she would not she didn't care about treats at all.
But now like that's all she cares about.
And so there was at this time every day she gets this like little no hide, rawhidey thing.
And we were like trying to figure out how do we take off calories? So she can lose a little weight and that one turns out that that treat was 48 calories
Which for a 20 pound dog 20 something pound dog is a lot
The vet recommended no more than 30 calories a day in treats and that one treat was
48 calories. So it turns out these other treats
that she likes so much more, takes way longer to eat,
is so much happier with, are only 14 calories,
which is crazy.
So that's what she's having now.
And the reason I'm mentioning this is
she also gets super excited that she says
she probably, she starts running around the house
with the treat and then comes back and eats more of it.
So I just, it's gonna take longer
and maybe
be another interruption of her barking or something. But then I'll I just have to mention this other
one. So beagles are not famous for fetch. They cannot fetch to save their lives. They were bred
to go straight and keep going straight because they're for fox hunting. They don't come back,
they just keep going. But she has this little toy
that you can put a treat down inside and Janice threw it out in the yard
thinking that she would just go out in the yard and have fun trying to get the
treat. But she really wanted to run back in the house to get the treat. So then
Janice realized if she just keeps throwing this treat out she'll run from
the back porch, go get the treat, run back in, try to get inside and now she's playing fetch.
You know, um, no, no, there was a, there was an advert with a good little dog on the reading
you gave us, Jimmy.
The fact that the adverts-
Milk bone ad.
The adverts were incredible.
Absolutely incredible.
I can't believe we finally read a PDF with adverts on.
Like there's one that's stand out the best.
The best advert.
You know which one I'm talking about, right?
I actually might not,
cause I kind of ignored the adverts.
I just focused on the article.
Oh, how could you ignore these?
I don't know.
Like, oh, he's lying.
You know which one's the best.
I actually don't.
I read the one about Milk Bone.
I read the one about garters.
I read the one about Smoke Smudge, and that's it.
Can I read this advert to you?
Yes, please.
Okay.
Can you do it in like a 1950s radio announcer voice? Can I read this advert to you? Yes, please. Okay.
Can you do it in like a 1950s radio announcer voice?
So you gotta do the transatlantic accent.
I don't know what it means, but let's see.
Yeah.
Do you believe in spanking a child?
When he throws a tantrum, a good walloping won't help
if the tantrum is caused by fright. If your child shows signs of
fear when you try to give him a laxative, be careful. It may be that the medicine tastes bad
or that it upsets him afterwards. You'll have to admit yourself that some laxatives are too strong.
some laxatives are too strong when he sulks before you reach for the hairbrush remember that sulking is often a child's way of avoiding something unpleasant
like the sissy laxative some parents some parents give their children
unfortunately some of these supposedly mild medicines may
upset a child without giving the relief that's needed. The fact is that some
laxatives are too mild. Spanking won't be necessary. If you treat the youngsters to
X lax when they need a laxative, Children really enjoy taking Xlax because it
tastes so good, just like fine chocolate, and mothers like the way Xlax
acts. Effective but oh so gentle. Not too strong, not too mild. Xlax is the happy medium. As a precaution, use only as directed. X lax the happy medium lax...
X... X... X lax the happy medium laxative. 10 cents and 25 cents at all drugstores.
Seriously, how messed up. Yes, I knew this was the bad for sure.
How messed up?
Is this like the post-war diet in America?
Means that everyone has,
make sure you take your daily laxative.
Is this, was this a thing?
Yeah, I have no idea.
This is-
You were around then.
Yeah, hang on, hang on.
Wow. Wow.
Wow.
Damn it, I was about to make that same joke.
Oh boy.
And now I can't.
Darn.
Back in your day, Ivan.
This ex-lax changed my life.
You don't understand.
Robin, relax with Mifflin. 15 minutes to her date, no time to shower, she'll be late. Smart
girl, she knows a pleasant trick. A Mifflin rub! She'll feel fresh quick. Mifflin means
the highest standard of quality in lilac, pine, lavender, wintergreen or unscented.
Mifflin. the national rubdown.
Yeah these are pretty good. So the paper we're reading today is not a paper. The thing we're
reading today is As We May Think by Vannevar Bush. Am I saying that correctly?
I believe so. Vannevar Bush? Am I saying that correctly? I believe so. Vannevar, Vannevar Bush.
Vannevar Bush. As we may think. And I've read it before and I read it in a nice little textbook that I picked up used for like 30 quid Canadian. And it's this collection of essays. It's got a
little bit of Borges. It's got this, it's got some other stuff I can't
remember, but it's called the New Media Reader. And it's basically like, here's a little collection
of things that if you just want to start dipping your toes into this kind of, what could the
dynamic medium be? What could computing be that's more than just spreadsheets and whatnot?
What about hypertext? What about making your own things in this meta medium? It's a good primer
for that. And so if anybody's new to this community or new to this space, that book is a really good
entry point. It's a textbook. It's like a textbook you might have had in grade school if you're about my age, which is 150. But unlike most textbooks,
it's made by people who are hypertext-pilled. So the margins are all full of weird kind
of interlinking metadata between the different essays that are collected. And it's a fun
thing to have. But it starts off with, as we may think, I believe that's the first one,
or if it's not the first one, it's one of the first ones. And it's short and it's, you know,
this little slice of the history of our field. But unfortunately, this textbook version does not
include these adverts. And these adverts are really something. So the version we'll probably
link in the show notes is not a physical book that you have to go in and buy It will be a digital PDF that includes the adverts. So if you're if you're at all curious and
Want to see some some pretty grotesque displays of what constituted popular culture circa the end of the war
It's there for you This print was just so fun.
It's from some magazine excerpting this at the time.
Life.
Life, okay.
It's from Life magazine.
Yep.
And the original article appeared in Atlantic.
Yes.
It's got all these ads that are just hilarious.
This is probably one of the shortest papers we've done other than the Dykstra.
Yeah. I don't know, it's like eight pages. It's kind of like average length, I think, for us.
Pretty little words. Little, little letters.
I had to get my blueprints magnifying glass out and go over in detail looking for clues.
You know, I sat down and read it this morning just to reread it and it was, you know, 10-15
minute read, something like that, while like even taking notes and stuff.
Nice and short, but we're going to cover it in such vivid detail that you won't need to
waste your time consulting the source text.
This is one of the reasons I wanted to talk about it.
I feel like for this one, I don't know that I want to just like sit and go through this
paper.
Maybe I am crazy.
Maybe you all do want to just like go through this paper. Maybe I am crazy, maybe you all do want to just like
walk through this paper, but I feel like this paper is something that's envisioning a future
that you know never happened. Oh, or did it? In some ways, but also well yeah, but also like some
other things may be replaced I guess I'm saying in its details right it like very specifically
looking at some technology choices
and saying, here's how we could use all this current day technology to do this thing.
And spoiler alert, computers do a lot of those things instead of the things it said.
And so I don't know as much that the details of like paragraph by paragraph,
it's not making quite an argument. I just thought in the same vein that it's imagining a potential future, it's imagining,
you know, what do scientists do after the war.
I thought it might be useful for us to, yeah, go through, summarize some of the paper, but
I would love to talk about what it proposes, see if we think that it's good and valuable, but then also talk about our own ideas for
solving similar problems, the same problem, what problem do we think we should be solving
essentially, you know, maybe make this a little bit more of a speculative episode than a history deep dive
So it is quite interesting though, because this is kind of a prediction piece, you know? It's like one of our New
Year's bonus episodes that you can listen to on patreon.com slash future of coding.
Patreon.com slash future of coding.
We have to throw in ads, because you know, the paper had ads, and sadly they didn't
sponsor pay us for them. So, though, you you know what? My Ex-Lax has been going down very well recently.
This episode sponsored by Ex-Lax.
None of the three of us have been spanked yet.
So, so far, so good.
There's been no spanking.
So it is a prediction piece.
It's making predictions about the future
and we're in the fortunate position
or unfortunate position of being in the future so we can say whether they came
true or not. Who is Vannevar Bush? Why is this article being written? What's
the sort of the setup? And correct me if I'm wrong Jimmy because I know you know
and I'm just pulling this off the top of my head. Vannevar Bush was the head of
the Office of Scientific Research and Development in the U.S.
And he sort of oversees a big core of scientists and coordinated their efforts through the war.
And the war here is World War II.
World War II, yes.
If you've seen, what was that movie about the atomic bomb that came out recently?
Oppenheimer.
Oppenheimer. Oppenheimer.
Oppenheimer.
Oppenheimer.
There we go.
If you saw Oppenheimer, Vannevar Bush appears in that film.
And so at the end of the war, here he is writing this article basically saying, okay, what
do we do next?
Where does science go next?
And one of the problems that they are finding is that the rate that science is progressing,
the rate that new papers are being written and the rate that new ideas are being forwarded,
is becoming so fast that it's hard to keep up. That there's these new advancements in
science arriving at such a clip that people are having trouble staying abreast of progress.
The summation of human experience is being expanded
at a prodigious rate, and the means we use
for threading through the consequent maze
to the momentarily important item is the same
as was used in the days of square rigged ships.
But there are signs of a change
as new and profound instrumentalities come into use.
Photocells capable of seeing things in a physical sense,
advanced photography which can record what is seen or even what is not,
thermionic tubes capable of controlling potent forces under the guidance of less
power than a mosquito uses to vibrate its wings,
cathode ray tubes rendering visible in occurrence so brief
that by comparison a microsecond
is a long time. Relay combinations which will carry out involved sequences of movement more
reliably than any human operator and thousands of times as fast. There are plenty of mechanical
aids with which to affect a transformation in scientific records. The world has arrived
at an age of cheap complex devices of great reliability
and something is bound to come of it." And so this article is laying out these hypothetical
inventions that seem just around the corner to Bush. These things that are, you know, a natural
extension of the current state of the art of technology that might help deal with this explosion of
knowledge and might help people work with information in a new way aided by technology.
Yeah. And I think the interesting thing we see from this article in the same way that
we've seen it from some other papers we've covered is that so much of, I guess,
computer technology progress comes out of wartime and like the explosion of research that came during
the second world war. Dr. Bush is talking about, I guess, the things we can take out from all of this wartime research and like
turn into more generally useful things. Like there's an interesting bit at the start that
talks about how scientists in different fields, how their work has changed or not changed because of the war. Like physicists, and I quote,
who have left academic pursuits for the making of strange destructive gadgets.
Referring to things like the atomic bomb. And I guess, you know, in some other fields it says
like biologists for example, during wartime and during non wartime their work is
largely the same. I guess computing is in this slightly strange situation in that this field has
kind of come to exist during wartime you know like that's that's the interesting and for me slightly tragic part of computing is that it came about from war.
So like in this article, it's kind of introducing us to the idea of computers being used outside of wartime for the first time.
What I think is really interesting as well is there is this, as we'll
see through the article, there's this kind of analog bent to the whole thing. Like we don't
really see digital computers coming in here, which makes sense. This is 1945. But I was just curious
to see like Vannevar Bush's Dr. Bush as life calls him. Yes, Dr. Bush as the, as life calls him.
Yes, Dr. Bush.
I wanted to see a little bit of his background and I did find this in Wikipedia which is
that Norbert Wiener approached Bush with a proposal to build a digital computer in 1940
and it was declined because Bush didn't believe it would be completed before the end of the war. And then that same computer got
funding later in June of 1943, which became the ENIAC. And there's people who at his critics,
the critics of Bush says that he was, you know, in favor of analog computers, which
is something he actually worked on and built, rather than these, you know, digital computers
that he found to be.
So it is just interesting to see, like it's kind of a mixed, you know, while this is very
influential.
So the reason that, as we may think, I think really became such a big influence here is
that Engelbart read it and cited it as his big influence for all the things he did.
So that's the big reason why I think
this is stuck around is so influential. Can we talk about the like the cover
image? Yeah. So Ivan do you want to have, can you have a describe? Alright sure. So the
cover of this, and I don't know if this is from the original Atlantic article or
if it was an illustration done by life. I do not think it was from the original. It's a it's
a like a charcoal or pencil sketch of a close-up of a person's face and maybe it's supposed
to be Dr. Bush maybe not. Bushy I'm gonna call them for the rest of the episode. And it's a close up of a person's face and they have these sort of round spectacles and
strapped to their forehead with sort of, you know, straps going around the side of their
head and then one strap going over the top is this little camera, this little square
camera, maybe about the size of a matchbox with a little lens protruding from the front
of it. And it's got this slender cable running down sort of across the brow behind the ear.
And it's one of the inventions described in this article, the Cyclops camera,
which is this camera that you wear on your forehead, or this is, you know, what Bushy imagines,
is that you'd wear this camera on your forehead
and you'd have this little shutter mechanism
that you'd wear through your sleeve
so you could trigger it with your hand.
And there's this little square
that's imprinted on the glasses
sort of outside the normal field of view,
but if you kind of dip your head down a little bit
and look through the square,
then whatever is framed by that square
is what the camera's gonna record.
And this camera Bushy imagines would use dry photography, which I guess is this sort of
novel invention at the time where you can take a photograph and it doesn't need chemicals to
develop, sort of like we got with Polaroid cameras in the 70s. And yeah, you'd use this
camera as you're sort of working in your laboratory to take pictures of anything that seems of interest to you.
And it dovetails with the second invention very closely, which is microfilm.
So there's this idea that the things that you are photographing can be reduced to a
miniature scale and then blown up later on.
So it's this primitive analog form of data compression
where you could take something like the Encyclopedia Britannica and shrink it down to the volume of a
matchbox. Material cost five cents. Thus a whole library could be kept in a desk. So the camera
would be taking these dry photograph pictures and shrinking them down to this little tiny size and
then you could have a huge mass collection of source material in a really compact form.
And hey, this image really looks like, to me it reminds me of Google Glass by the way. Or Meta Ray Bans, right? It's this Cyclops camera thing. It's funny seeing it drawn here so long ago,
because I guess even now this is something that I guess appears as a product, a short-lived product,
short-lived product or like a concept piece, right? I think the Ray-Ban, so this one, the camera is like strapped on your head like you're
like a headlamp or whatever, rather than being on the glasses themselves.
And there is a little square on the glasses, but I think that's just like a printed on
square to help you frame the shot.
Yeah, yeah, rather than like, at first when I saw it,
I was like, wait, was he imagining displays and glasses?
But no, it's a little printed on square
to help you frame the shot.
But then you have a little wire
that can run down your sleeve,
so you can have a little button to press
to take the camera,
which I just think is something that Ray Vans definitely
need to add for the cool factor, you know, just having a little wire and a little thing
you hold all the time.
Oh, that would make them so much cooler.
Wired headphones are making a comeback like wired earbuds.
Oh, yeah.
There's the the the not even the zoomers like the gen alpha kids wearing wired head pods head head pods head crabs wired head crabs white head crabs that's funny I mean I
know CDs and stuff came back so it's not surprising you know everything old is
new again I find the Cyclops camera I guess prediction quite interesting
because I feel like in some ways it was so wrong like we're not walking around with these things on our
heads but on the other hand the it was the fact that it was saying that anyone
will be able to take a photo at any time using something that it calls dry
photography and is that is it called dry photography because it's like, it's not
using like traditional film if to go develop? I don't even know.
Both of you. Okay. Hang on a minute. What did the two of you completely enter a fugue
state when I read through by or when I, when I said my description of what the photo was
because both of you now have done a thing where it's like something I directly said
and addressed. You, you come back with like this thing and share it as though it's for
the first time. I put both of us.
I listened to what you said. I know what photography is. Are you talking about me talking about
the square on the glasses?
Yes. I just, and this, and the cord down the sleeve, the way you set it up was sort of
like, Oh yeah. And there's this cord down the sleeve. I said there was a cord down the sleeve.
You said, and there's a cord down the sleeve instead of saying, and that cord down the
sleeve, which would imply like, yes, the cord we talked about previously. You're like, oh,
and there's this cord. Did you know about the cord?
Do you know what it is? Both of you.
Ivan, the thing is when you do a monologue, your monologue then becomes the text. Yeah.
And then we have to unpack the text. Yes. You know what I mean?
So there's also, and I quote,
the cord which trips its shutter may reach down a man's sleeve
with an easy reach of his fingers.
So you didn't read that part.
There, there's the text reset.
See, yes, dry photography.
The idea is you're not developing the film with chemicals in a dark room,
with liquid chemicals, that it is photography that doesn't need the traditional kind of development.
I'm trying to find where he goes through, yeah, will there be dry photography?
It is already here in two forms. There have long been films impregnated with diazo dyes
which form a picture without development so that it is already there as soon as the camera has been operated. An exposure to pneumonia gas destroys the unexposed dye and the picture can then be taken out into light and examined.
The process is now slow, but somebody may speed it up and it has no grain difficulties such as now keep photographic researchers busy."
So yes, as opposed to wet, as opposed to if you had to actually have wet processes.
If you've ever developed film, you literally put it in liquid.
So wet photography.
Hot, wet American photography.
No, that's not like surfboarding with a GoPro or anything.
But this is so right.
This is so right.
I think it's good to step back though on,
because you're totally right.
He got it right.
But his use of this is so vastly different
from Google Glass or Ray-Bans.
He's not imagining people in their everyday, non-science life
just walking around with cameras strapped on their head
having fun.
And they're snowboarding, and then they're
taking all these epic screen.
This is always like the ads or whatever for Google Glass.
You're not just a glass hole.
You're just snowboarding all the time
and look at my epic thing that I was able to do.
He's imagining this specifically for the scientist.
He's imagining that, let me find this little quote of what he ultimately says,
One can now picture a future investigator in his laboratory.
His hands are free and he is not anchored.
As he moves about and observes, he photographs and comments.
Time is automatically recorded to tie two records together.
If he goes into the field, he may be connected by
radio to his recorder. As he ponders over his notes in the evening, he again talks his comments
into the record. His typed record, as well as his photographs, may both be in miniature so that he
projects them for examination. What he's imagining is a scientist going out in the field, a scientist
sitting at his desk reading, etc.
can have this little camera and instead of having to memorize everything that he saw, he can take a picture.
It can automatically be associated with things. It can be put into his notes for later recall.
And I mean, this is not unlike, you know, people sitting at computers taking screenshots of things or, you know,
take like it's very common for people to use screenshots as a way of
associating different items, right? You're at your computer, take a screenshot, put it over here,
you know, stow it away for later, whatever. So there's like something very, very interesting about this, but this ultimately goes to this whole machine that he imagines that supports this
workflow that's called the MemEx. Well before we get there, a point on the point you just made
is that I think what Bush has here in this article Bushy, it's an old enough text I'm bringing back
the joke, is that he is very good at imagining the sort of logical consequences
of present-day technology and imagining where that will go and maybe the exact mechanisms were wrong,
but that it seemed inevitable that if we continue improving present-day technology, what would we
get? So if we have cameras and we make better cameras, what will those better cameras let you do that present-day cameras won't?
And what he doesn't have any of at all, and this becomes to me even more starkly apparent when we get into the meme-axe,
is an imagination for how is this going to impact the way that people actually use the technology?
What are the cultural consequences? What are the effects on scientific practice? What are the ways in which, you know, these
technologies will meet the world, will solve existing challenges, will create
new challenges? Any of the, what I would describe as the real consequences of
this technology, I think are absent from this piece. And it's startling to see, you
know, the signs that we would get technology like
this were there already at this early stage, but that what that would make the world look
like or what that would allow the scientists to do is not apparent at all, at least in
my opinion.
I mean, I do think he doesn't think about like larger, broad social things as much. But I feel like that was kind of the focus here is this if we can, if we do have this
technology which he believes will happen and he kind of he goes at length on talking about
trying to prove to people that you could do this, you know, maybe not like full detail
but look okay I promise this stuff would happen like the dry photography but the whole point is like how it would change the practice of
the the physicist yeah, and I think he gets all that stuff wrong like all the stuff that he says there is is is
Mispredicted and and that we can jump ahead to the memex if we want that's the most famous
Thing from this paper. I think it's the one that I see the most commonly cited,
at least among scholars of the future of coding.
But there's some other stuff in like the vocoder
and the thinking machine that I think we should come back to.
Especially the thinking machine section
has some real choice quotes in it.
So can I just get like a score from both of of you like this first prediction of the Cyclops camera?
Oh a game! We'll play a game. Yes. All right Cyclops camera, right?
We could make it very binary like is this prediction correct? Yes or no?
And I think that would be quite boring because it's mostly no because it's not like a hundred percent correct
So I would say like, you know, what,
out of 10, how correct is this?
Like how true did it come?
How true did it come?
Yeah, how true did it come?
It came correct.
Yeah, the Cyclops camera.
So this idea that you could have a tiny camera
that you take with you everywhere you go,
that you could immediately see the result of your photo, came very steadily truer and
truer as time goes on.
If anything, it could have happened faster, but that's not any fault of Bush's.
I'm going to give it a nine out of 10.
Nine out of 10? Yeah, because I think he, I mean, even if he like kept it, he didn't, you know, foresee of bushes. but dry exposures. You can definitely do this with Polaroids.
All of that stuff does happen.
You can get pretty instant feedback.
And then if you can expand this out to,
like he says that he trips the shutter
and in it goes without even an audible click.
Digital obviously does not have an audible click
whenever you do the shutter.
So even if you try to give him some digital,
the thing that I think he got wrong,
which is why I'm giving 9 out of 10,
is just the form factor and use case that is going on here. I don't think you'll find that this,
maybe there was a time period where scientists walked around with a camera strapped on their
head. You could try to argue GoPro is some way is this, but I don't think it's being used in the
the use case he's imagining. Yeah, I think it's a 9 out of 10 for me too, right? Like the specifics are not correct,
but like the vibe is correct, like the general sentiment of we will be able to like snapshot
either through a screenshot or like a camera, anything that we can see and keep it for later
use. That is so true now, like that's such a big part of, I guess, having a phone.
It makes me wonder if it is actually all that wild
of a prediction or if we,
because there's, I'm very not familiar with fiction.
That's not my thing. I'm not a fiction kind of person.
I'm not familiar with fiction.
Yeah, I have only a passing familiarity with fiction.
And I especially fiction of this era, but
it seems obvious in hindsight at least that the thing that you see with your eye
is the thing that you're going to want to record with like you want to
preserve the memory of the thing that you saw and so this idea of put the camera ever closer to the eye seems obvious.
And we don't really have that right now.
Like we had Google Glass and we have Meta Rayman things, whatever those are called.
And it seems like that's inevitably what we're going to get is some kind of way of taking
what you see as you see it and capturing that.
And then like more, not just, oh, not just what you saw, you see it and capturing that and then like more not
just oh not just what you saw but a wider field of vision or greater zoom or
you know wider color spectra or whatever it is like we can we can extend the
breadth and depth of vision but that idea of putting it right by your eye is
I think a really important part of this that To me seems obvious in hindsight and I have no idea if it was obvious at the time. There is a journey into other worlds
by John Jacob Astor the fourth
Predicted speed cameras and spy cameras. So this like very miniaturized camera you asked GPT didn't you?
No, I looked at Wikipedia. Oh.
List of existing technologies predicted in science fiction.
Oh, wow.
Yeah.
That would be a good, uh, can we read a Wikipedia page for future episodes?
Yeah, we should do Wikipedia.
We should do an episode.
No, I'm 100% serious.
We should do a Wikipedia episode.
So that was Cyclops Camera.
Microfilm? Should we score microfilm also before we move on? Because we kind of talk about that. No, I'm 100% serious. We should do a Wikipedia episode. So that was Cyclops Camera.
Microfilm, should we score microfilm also before we move on?
Because we kind of talk about that.
I mean, microfilm is a real thing.
Microfiche does exactly what he describes.
Right, right.
I mean, microfilm, if you look at his little or like the little summary, The main thing it's talking about is like the ability to store tons and tons and
tons of information on something small, right? Just being able to have a hard drive or being
able to like, you know, you can download Wikipedia. It literally uses the example of storing
Encyclopedia Britannica on something as small as a matchbox? Is that what it means? Material
cost five cents. We should do the maths. What's that inflated? 90 cents.
I want to read a sentence just because I like the sentences in this.
If the human race has produced since the invention of movable type a total record in the form of magazines, newspapers, books, tracts, advertising, blurbs, correspondence, having a volume corresponding to a billion books, the whole affair assembled and compressed could be lugged off in a moving van.
Cool. The microfilm stuff, he is not quite predicting the technology.
The technology existed quite a while before that.
In fact, even in 1938, Harvard University was using it to preserve newspapers.
Which if you go to libraries today, you can go see a bunch of old newspapers on microfilm.
Isn't that Microfiche?
Isn't that what that is?
Is that the same thing or a different thing?
It's microform.
I've never heard of microfiche.
Microfiche. A flat piece of film containing microphotographs of the pages of a newspaper catalog or other document.
The journal is available as a microfiche.
Yeah, this is definitely, like, he's not predicting that it would be, like, that it would exist,
because he even says, today with microfilm reductions by a linear factor of 20 can be employed and still produce full clarity when the material
is re-enlarged for examination.
So this is more him talking about, you know, pointing out this is one of the things we
could use as a, you know, everyday kind of thing instead of just for large archival things
or whatever.
So I mean, it's not quite a prediction, except
for its use. And I think it's use is wrong. It didn't, it didn't get used the way he talks
about it, which is like for the everyday scientist having their own library of microfilm. Maybe
there was a period in time where there was they did this. I don't know.
Yeah, this, this might actually be a case where the critique that I raised earlier about kind
of missing the cultural implications of these advancements is this is one where he actually
got it, where it's like the idea is that a person doing serious work, for example, would
want to have a large corpus of information available to them and that some kind of like
compressed form of that information would be really valuable. So having ready access to large volumes of
information and both information that you produce personally, like the work that you're doing,
that you're documenting, you want to be able to produce a lot of that and hang on to it,
but then also to share that work with other people in a very efficient way. And he talks later about sort of, you know, daily deliveries of new batches of scientific
publications and works of colleagues and that sort of thing getting shipped around.
But then also, you know, a large corpus of public background information.
I think that part definitely came true.
It's sort of like the personal computer vision, right?
That it's like, oh, computation is this amazing thing,
but if it has to be this giant mainframe in a building
that is, you have to go there to do your work,
that creates the desire for that computation to be with you
and to be available at all times.
And so I see that, that part of this prediction
as being quite on the mark.
So I'm gonna give it an 11 out of 10. Whoa
I'm giving it a 10 out of 10 10 out of 10
like look we're reading this PDF right now because we're able to
Like live the envisioned reality that especially gave us. All right, then we get the author
Need not write he could talk his thoughts to a machine.
Vocoder, a machine which could type when talked to.
But you might have to talk a special phonetic language to this mechanical super secretary.
Like, okay, firstly, super secretary.
For what it's worth, this is not Bush writing,
this is life writing.
Right, yeah, this is them summarizing what he said.
Yeah, special phonetic language.
I mean, okay, to be honest,
when you go like, hey Siri, you know,
that's kind of using a special
language isn't it? Hey Siri, okay Google. Prompt engineering. Prompt engineering?
Maybe? Yeah, like knowing that you have to talk to the computer in a particular
way using natural language right quote-unquote natural language but you
have to use it in a particular way if you want the thing to actually do what you want it to do.
Yeah.
And that there's these tricks you employ to get a better result.
I guess there's like different parts of this.
What, like one part is just text to speech and right, that tick that's done.
Right.
Another part is like, yeah, talking to your computer as if your computer is an assistant and I guess you could
say we do kind of have that in multiple forms too right through like the hey series style of thing
but also nowadays LLMs as well whether that's a good thing or not is is up for grabs but we do have that. Vocoder I would say 10 out of 10. 10 out of 10 again.
Yeah just to be clear like the vocoder for Bell Labs you literally did have to speak
differently in order for it to understand you. I can't remember the you know the exact
way in which you had to but you had to just have different vowel
Sounds and different things like that. It couldn't just like understand us speaking normally. It's like, you know earlier
Speech to text people had to adapt a North American accent in order for them to be understood. It's gotten better now
But you know you can watch all sorts of videos of Scottish people
trying to talk to Siri and not having any idea what they're
saying.
I keep saying, hey, Siri, put boy milk on the grocery list,
and it gets it so wrong every time.
Boy milk?
Jimmy, yeah, boy milk.
I added boy milk to the grocery list.
What the hell is boy milk? I's, I get it all the time.
Boy milk?
I have no idea what this is.
Jimmy, did you score microfilm?
I can't remember if you scored microfilm.
I'm gonna be contrarian here and say it's a zero.
Because the part that he was predicting, in my opinion,
was that it would be used by the everyday
Scientist for storing all this stuff and it was not hey Siri. What's what's boy milk?
And if you say if you say add cinnamon cereal to the grocery list it it says okay
I've added cinnamon and cereal to the grocery list which is bad because then you can't have cinnamon cereal and boy milk
Do you mean soy milk?
No, no, I don't like soy milk.
Jimmy, zero out of ten really even though people actually do use data compression in
the world?
Yeah, he's not predicting data compression, one.
Two, he was talking about things that already existed when he was talking about microfilm
and using it for various things.
So it's a bad prediction, because it's, yeah, okay.
Well, the only thing he's predicting in the future,
in my opinion, is the use of it by the scientist.
And maybe I'm wrong, right?
Like, I will be happily, like maybe in the 1960s,
scientists were sitting around at their desk
with lots of microfilm
of pictures they had taken.
I would actually believe that that is true.
So mine is either a 10 or a zero, depending on if that happened.
And I just don't know.
And that's not something easily Google-able.
Nice.
And what's your score for the vocoder?
The vocoder, I would say a 10 out of 10.
Nice. I'm going 12 out of 10, just cause like,
music in the 1970s would not have been the same
without the vocoder.
It's such a, now that of course is a different use
of the vocoder than the one that he foresaw,
but it's still a vocoder and it's extremely valid
and very good, so 12 out of 10.
Okay, we're now over halfway through the predictions.
I feel like this might be a good point to do the disclaimer
that all the way through this entire thing,
Bushie is imagining that the only people doing any
of this work are men, and he's very clear
that he uses he, man, he, he, he he he. I've highlighted all of them.
Just to remind myself that like, you know, like sometimes I'm like, no, I don't even need to
mention it this time. This is the same with all these things. But holy moly, like it's, it's,
it's so prevalent and no, he doesn't just use he because it's a like the grammatical thing and you can
you can tell this because there are some times where he does refer to women and I will quote
one of the times he refers to women. One of them, so he's talking about a machine,
one machine will take instructions and data from a whole room full of
girls armed with simple keyboard punches.
I just find it interesting.
It's like, why is that, why is that a woman's job?
And why are the others all clearly men's jobs?
Right.
And I guess that's just because, you know, that that was the norm at the time.
But I think it's really worth stating that this is I know this this article was in the past this this piece of writing was in the past but it
clearly had influence on the things that we're currently using in some form you know and hopefully
this gets shaked out more and more over time but I think that's still pretty prevalent in the tech world that we're currently in.
So sorry to be a broken record, everyone, but it didn't need saying again.
Matthew 14 I have this whole passage highlighted
as the, for me, the low point of the essay.
Matthew 15 Well, the thinking machine bit.
Matthew 14 Yeah, this thinking machine bit. So let's, let's back up a second and introduce what this one is. So the, the short two sentence summary is that it's a mathematical calculator,
give it premises and it would pass out conclusions all in accordance with logic.
So this is the bit where he's sort of predicting that calculating machines
will get a little bit more sophisticated and that you can feed them
premises and they can kind of chase them down to their logical consequences
automatically. It's a far cry from the abacus with its beads strung on parallel
wires to the modern keyboard accounting machine. It will be an equal step to the
arithmetical machine of the future. But even this new machine will not take
the scientist where he needs to go. Relief must be secured from laborious detailed manipulation
of higher mathematics as well, if the users of it are to free their brains for something more
than repetitive detailed transformations in accordance with established rules.
A mathematician is not a man who can
readily manipulate figures. Often he cannot. He is not even a man who can readily perform
the transformations of equations by the use of calculus. He is primarily an individual
who is skilled in the use of symbolic logic on a high plane, and especially he is a man
of intuitive judgment in the choice of the manipulative
processes he employs."
Jimmy, does that check out for you?
Does that make sense?
Like this idea that it's like mathematical thinking is about specifically about being
skilled in the use of symbolic logic on a high plane?
Because to me that, I can't even tell if like that makes
sense in the historical context, but for me that feels like it just fundamentally misses
the point of what mathematical thinking is, but maybe I'm just bringing too many of my
own preconceived notions to it.
So let's, let's set aside gender issues here, because obviously, he definitely, I think
it's no doubt he's, he has these kind of gender norms assumed it here the past seriously the
past sucks and the present man when I say I'm a guy I'm gonna have some
agreement here it's not on the gendered things obviously anybody can do this so
I think you know taking the word symbolic, I don't think he means in this passage,
while he might mean it later, like trying to be charitable here, I don't think he means
literally symbolic logic. You got to think about him being a physicist here. And if you listen to
like Sean Carroll, for example, who's a contemporary, you know, theoretical physicist,
and you know, more akin to the kind of work that's being done here.
He says he's not very good at mathematical manipulations.
If you had him sit down and do a bunch of calculus, which he even mentions here,
he wouldn't be anywhere near as good as it is as a mathematician.
He's a little rusty, and, you know, he depends on computers to just do the math.
And at this time, you know, something as simple as taking the sign of a number or whatever would have
been a big complicated process that they have to sit and do a bunch of calculations.
And they have tables, like using look at tables for that.
Yeah, they have tables.
Yeah, right.
If you don't have a table of that particular calculation, you got to sit there and do it. So what I think he's trying to say here is just that,
I think he's completely right that like,
it's not enough for there to be arithmetic going on,
just like these accounting machines.
It's that you have to be able to,
I think of something like Mathematica or whatever.
You have to be able to work at this much higher level of abstraction and just let the machine do all of this
Mathematics and simulation and etc for you and this is what you're you're thinking about how these things all fit together
I think if that's what he's saying, which this is how I take them to me
Then I think he's totally right and you know really did predict what how the physicist is going to use computers in the future.
There's a real sliding scale right on the very limited viewpoint you know he even talks about
the abacus there's really simple calculations being carried out by computers obviously I think
that was happening back then it's still happening now you You know, slightly more advanced from that. There's perhaps
just doing it many times, you know, like I think of people, you know, scientists using Python to
write scripts, to do a lot of number crunching, that it's like one level of abstraction higher
than, you know, typing in each individual calculation to do.
It's getting a computer to tell the computer to do the calculations.
And being a scientist who works with data means using notebooks
and programming a lot of this stuff yourself.
I think this is talking about largely the same thing. I guess the thing we don't have is like a magical computer box that figures out any logical question you ask it. But things like Mathematica, I guess they aspire to be something like that. You know, and, and I'm,
I don't know specifics, but I'm sure that there is research out there that tries to build
programs machines that do that sort of thing. So it's a nine out of 10 for me.
There's a there's another passage I want to quote here that talks a little bit about that.
Is this general purpose? Is this sort of hinting at the computer, the general computer?
The kinds of thought that are this sort of creative problem solving that mathematicians need to do.
Or maybe that's what I'm bringing to it. I think creative problem solving.
But he specifically says symbolic logic on a high plane, which is vague enough that it could mean the same thing as
creative problem solving, which is also vague enough that it could mean whatever I want
it to mean. But he says all else, all other kinds of thought, should, all else he should
be able to turn over to his mechanism just as confidently as he turns over the propelling
of his car to the intricate mechanism under the hood. Only then will mathematics be
practically effective in bringing the growing knowledge of atomistics to the
useful solution of the advanced problems of chemistry, metallurgy, and biology. For
this reason, there will come more machines to handle advanced mathematics
for the scientist. Some of them will be sufficiently bizarre to suit the most fastidious connoisseur of the present artifacts of civilization."
Is that suggesting that we're going to get specialized computing hardware that'll be able
to solve different kinds of problems? Because that's sort of an interesting, sort of missing
the general purpose nature of the computer if it's thinking that
we're going to be like specialized calculating machines all the way down.
Yeah, so Bush, Bush built a differential analyzer.
What's he?
Bush constructed a differential analyzer, a mechanical analog computer with some digital
components that could solve differential equations with as many as 18 independent variables
So so this is in 1927. This is what he was working on. Whoa. He was a big
Analog computer fan. Yeah, and I mean as you know
He didn't give the funding for the digital computer that became ENIAC right?
Like he was a he clearly has this bias towards analog
computing and analog computers, if you take this that as the idea here, it
makes perfect sense where he says, you know, we'll come with more machines to
handle advanced mathematics for the scientists, some of them will be
sufficiently bizarre. He's thinking that for any of these kinds of problems, you're gonna have to build a new analog computer to make them.
Yeah, I wonder if he knew the...
Like, I wonder if he was aware of Turing's work and Church's work contemporaneously?
Is it around the same time?
Contemporaneously? is that it?
I mean, he was asked to help fund a digital computer.
So he was familiar with the general concept.
But I don't think digital inherently means general purpose,
and I don't think analog inherently
means non-general purpose.
That's a distinction that I want to draw,
is like the special purpose computation
versus general purpose computation is something that I'm
wondering if he has that in mind and if he just doesn't feel like it's scalable or that
it's like, you know, if you want a machine to do a computation, build a machine directly
to do that computation for the sake of efficiency and effectiveness or like what are the tradeoffs
that he sees here? I don't know.
He does have a later like 50 years later
I think essay going back on this but it's like not it doesn't add a ton
It's like rewriting as we may think and talking about digital computers
I didn't get to read it as close as I want to I wanted to but also like you have to think he
Definitely is familiar with this stuff. Like one of one of his grad students is Claude Shannon. Yeah
Like he's he's definitely very plugged into this
I think the you know the fact that he didn't want to fund it because he didn't think he could be done before the war
Ended he's he's definitely being practical here. I think he's imagining that digital computers
Yeah, he obviously saw their use,
but he didn't think that they would be able
to do all of the stuff that the scientist needs to do.
And just another passage here to trigger Jimmy.
The scientist, however, is not the only person
who manipulates data and examines the world about him
by the use of logical processes.
Although he sometimes preserves this appearance
by adopting into the fold anyone who becomes logical, much in the manner in which a British
labor leader is elevated to knighthood.
Whenever logical processes of thought are employed, that is whenever thought for a time runs along an accepted groove,
there's an opportunity for the machine.
Formal logic used to be a keen instrument in the hands of the teacher in trying of his students' souls.
It is readily possible to construct a machine which will manipulate premises in accordance with formal logic,
simply by the clever use of relay circuits.
Put a set of premises into such a device and turn the crank,
and it will readily pass out conclusion after conclusion, all in accordance with logical law.
We may someday click off arguments with the same assurance that we now enter sales on a cash register
zero out of ten okay you know one way that I think about what he says is that
like he's talking about building machines for different kinds of problems
and I think when we read this, we think,
no, we don't do that. We don't build new hardware machines, but we kind of build
new software machines every time we have a new kind of problem. So in some ways,
I think, you know, this is, this is correct, but not in the way it expected to be
correct. And that's why it's a nine out of 10 and not a 10 out of 10
We can in fact do like formal logic and you know
Check that the structure is valid and all of that stuff, but that's not what he's talking about
That is what he's talking about literally what he's talking about He's talking about like like basically like first-order logic like you could build a machine that could do like well if this thing is true
And that thing is true. Then the whole statement is true, which
is totally not how computers work. He's not saying that's how computers work.
He's saying you can build a machine that can do that. Which you definitely can. Now?
Yeah, I could write, I can show you the program that does it. I've written that
program like 1,200 times. That Where you could say like, you know,
some predicate and if that predicate is satisfied,
then there's something that happens.
Otherwise something else happens.
Okay, now you're just being over here being facetious.
13 out of 10, this is exactly what computers are.
It's just described in a way that completely misses
the way that we use them culturally. because nobody like Jason Morris, I think in the future of coding Slack is like,
hey, we need to create software that supports legal processes and debate and argumentation
where you'd put all your premises in and arguments and, and then as you evaluate them, you
can sort of keep track of who's right and who's wrong and who committed a
contract violation or whatever, what have you, you've built stuff like that too.
Jimmy, you've told me about you making that kind of stuff.
Actually, it's the, um, it's the fever of clicking slack.
I think you're referring to that.
It's a support group. Yeah. Slack, I think you're referring to there.
It's a support group.
Yeah, for British labour leaders who are elevated to knighthood.
So far Bushi is getting pretty high scores, right?
Yeah.
The first four predictions, pretty high scores, you know, some blips.
But I think it all comes crashing down in the next one. Are you ready to go
to the next one? Do you have more thoughts about? I'm done trolling Jimmy. We can go
to the next one. I was confused. So it's like, are you being just narrow here? No, you were
being broad. It's fine. But yeah, you can literally and they do, you know, there are
just things for formal logic, right? Or you plug in an argument and see if it's valid or not.
They even made machines.
In fact, Billy Mays was the creator of one of the such machines in the 1940s.
Are you serious?
That actually just took logical arguments and pumped out.
Like it was a analog machine that did that.
That was why he was asked to comment on
Rial on by Rial on the paper, huh?
Was it actually wolf maze was that the name did we finally figure that out? Yeah. Yeah
Yeah wolf maze w maze wolf. But yeah, he was actually so last episode talking about Turing
He was actually somebody who had built one of these machines and that's why they thought he might be
actually is somebody who had built one of these machines and that's why they thought he might be on Turing's side and that's why Ryle wanted a commenter that
would be on Turing's side. I think that shows that even Ryle thought the paper
was bad. Yeah. Looking for a second paper from a legit philosopher to, you know,
shore it up. At the time too it it wasn't like, yeah, it was going to be post, it was going to be published
together.
Yep.
So cool.
Sick.
All right.
The Mimics is the next section, right?
This is the next prediction that you, Lou, are saying zero out of 10.
Zero out of 10.
Interesting.
Wait a sec.
Should we say what it is?
No, let's not say what it is first.
Everybody knows what it is. Why is it zero out of 10? We do not have a Memex. No, no, no one. Okay,
actually, it was interesting to reread this, like, well, like, reread about a Memex, because I think
a lot of people are familiar with what a Memex is. but this article places emphasis on the structure of it, and I know
I know some people are like, yeah, obviously, that's that's the whole point of a Memex.
I think some people say that the internet is a Memex, right, and Wikipedia is a Memex.
But to be honest, I think the internet and Wikipedia are both more similar to the thing
that this says that a Memex is not, right?
In fact, I think I picked out a bit.
Here it is.
The human brain files by association.
The Memex could do this mechanically. When data of any sort are placed in storage, they are filed alphabetically or numerically
and information is found, when it is, by tracing it down from subclass to subclass.
It can be in only one place unless duplicates are used.
One has to have rules as to which path will locate it and the rules are cumbersome.
Having found one item, moreover, one has to emerge from the system and re-enter on a new path.
So that's basically the status quo, right?
That's how files and data and information is currently stored at this time, right?
Like it's talking about...
In a library.
Yeah.
In a library in your cabinet you have folders and in those folders you have little
pieces of paper or maybe some clip together and in that and you flick
through and has so it's sort of this hierarchical you start from like this
zoomed out point where you have like categories and in those categories are
categories and those categories are categories and there's this path you
take to go down them the article article says, and I quote,
The human mind does not work that way. It operates by association. With one item in its grasp,
it snaps instantly to the next that is suggested by the association of thoughts,
in accordance with some intricate web of trails carried by the cells of the brain.
It has other characteristics, of course.
Trails that are not frequently followed are prone to fade, items are not fully permanent,
memory is transitory. Yet the speed of action, the intricacy of trails, the detail of mental
pictures is awe-inspiring beyond all else in nature. Man cannot hope fully to duplicate this mental process artificially, but he certainly
ought to be able to learn from it. In minor ways, he may even improve for his records
have relative permanency. The first idea, however, to be drawn from the analogy concerns
selection. Selection by association rather than by indexing may yet be mechanized.
So it's saying the way that we organize information currently through like structures and categories
and boxes and sections is no good because that's not how we think.
And I think that we still do that bad way of organising things in pretty much
every website, in every single piece of information that we consume. I still think it's this top
down, structurally thing. I don't think we have this Memex-y idea of more like a interconnected web.
Sure, I think like on Wikipedia you have like pages that link to other pages and stuff like that,
but it's like that kind of just feels like slapped on the top of this structure that we've stuck with, right?
We're not- we don't have a memex yet. We don't- we haven't fully realized
this full vision of what a memex can be.
Jimmy, what's a memex?
It consists of a desk, and while it can be presumably operated from a distance, it is
primarily the piece of furniture at which he works.
On top are slanted translucent screens on which material can be projected
for convenient reading. There is a keyboard and a set of buttons and levers. Otherwise
it looks like an ordinary desk. In one end is the stored material. The matter of bulk
is well taken care of by improved microfilm. Only a small part of the interior of the Megamax
is devoted to storage. the rest is mechanism.
Yet, if the user inserted 5,000 pages of material a day, it would take him 100 years to fill
the repository, so he can be profligate and intermaterial freely.
But listen, it's so much more than a desk, man.
It's so much more than a desk.
So that just, for the listener who has never heard of this thing before, it's a
desk. There's a scanner, like a document scanner, a little glass square on your
left that you can put a document down on scans it in.
There's two screens in front of you side by side, portrait orientation for looking
at documents. And on your right hand side, there's a bunch of buttons and like
keyboard kind of stuff basically. And so it's,
it's like a little database you scan documents into or you load documents in if you already get
a delivery of microfilm or whatever. And then you can recall the documents and on the screens in
front of you, when you have a document open, there's these little kind of code letters at the
bottom or little symbols or that sort of thing that let you jump to
related items and you can build up these relations by saying hey
I'm gonna pull two documents up one on the left one on the right and I'm gonna you know match up the codes
Somehow it's kind of vague on the details
But basically set these two documents as having an associated relationship so that whenever I'm looking at one
The other one is one of the suggested related documents.
That's basically what it is.
It's like information storage and retrieval and the ability to manually build up associations
between documents.
And some of the little controls on the keyboard let you like scrub through pages of a, you
know, if you're reading a book or something like that, you can scrub through pages of a, you know, if you're reading a book or something like that, you can scrub through pages kind of slowly or quickly, and then you can also
scrub through the chain of related things. So you can pick a trail, as he
calls it, of relations and kind of scroll through that trail. So it sounds in some
ways a little bit like a hypertext, but it is, at least in the description
that he has given, you are manually building up the
relations between these things. Or maybe the relations come when you get a load of microfilm
or whatever it is. But it is not something that I think a lot of people mistake. And I'm going to
quote from an article in the Atlantic from, does it have a date? May 9th,
2025.
Bush, who oversaw America's research efforts during World War II, imagined a system that
would allow researchers to see all relevant annotations others had made on a document.
Okay, so far so good. Oh yeah, because you could like, you might have a stylus that you
could use to write margin notes on documents. That's another thing Bush describes.
His memex wouldn't provide clean singular answers. Instead, it would contextualize information within a rich tapestry of related knowledge, showing connections, contradictions, and the messy
complexity of human understanding. It would expand our thinking and understanding by connecting us to
relevant knowledge and context in the moment in ways a card catalog or publication index
could never do. It would let the information we need find us.
I think that's another like common misunderstanding of this work is that it's not doing the work
of like a recommendation engine or a semantic system that's going to like
look at the information in an article and make suggestions of associated
things. The associations are things that like people create for each other. It's
like manually entered metadata which is very limiting and I think makes this
MemEx a little bit of a solution in search of a problem or at the very least
the problem statement that he sets out at the beginning of the article where he
says there's this volume of information that we are having trouble keeping up
with. His proposed solution I don't think gets to the heart of that issue because
it's still assuming that somebody is going to be reading through all of these
large volumes of information and building up this this interconnectedness this this web of links between documents.
Well, well hang on I think I mean the way I think about it is that if you put information onto a
computer you like you probably want to structure it in some way, manually, right?
Because if you didn't, you just have a big list, right?
You just have a big list.
Maybe that's maybe that's best for the kind of information you're inputting, but usually
not I'd say.
Like even if it's just like, hey, I have photos on my computer, right?
I'm going to put them in folders, probably, right? You know, like I have albums
on Google Photos. It's the same with something like Notion. If you're using Notion to make
notes, you're going to, you are going to manually structure them. And it's the same with something
like Wikipedia as well. In fact, Wikipedia is more like a big list, really.
Well, there's a critical
difference between Wikipedia and this, which is that most of the time you start Wikipedia with a
search. You say, get me a document that matches this term that I'm putting in. Right. Now I'm
talking about writing Wikipedia. You know, like, say you think there should be a new page, right?
I think, you know, like whatever structure you have,
you're gonna be able to search for it.
Not in the memex, that's the thing.
Like search, that interface dimension
never enters the picture here.
Right, I mean, I wanna set aside search.
I don't think this is what it's about, right?
You know, so yeah, like maybe it does, maybe it doesn't.
I think like full stop, you're gonna need to,
or you're gonna probably want to structure things
in some way, right?
And Wikipedia, you know, when you make a new page,
it's, you know, you probably, you just make a new page
or these things don't go in folders,
but often you get like categories of pages.
You know, you get like list of pages about blah,
right? And so, you know, we always have these choices that we, we, we have to make a choice,
you know, like, no choice is a choice in how we structure information. But in things like
notion and in like our file system and in the paths of URLs, there is only one choice, really, which is that we have a tree, right?
We have a tree structure.
It's hierarchical.
Yeah.
And I think this is getting across that the tree structure
is not how our brains work, right?
That's not how we process this information
or structure this information or structure
this information internally. But you know, like if you're only stuck with a filing
cabinet, you can only do the tree as well, right? Like in the real world the
tree, you can't have things, I mean maybe you, I guess one of the times you don't
have a tree is like, you know, when you have like a conspiracy theorist board on your wall with all the red
string, right? That's less of a tree, that's more of a Memex. Like, for me, it's like, the traditional
way is a tree. And the Memex way is more like a mind map, it's more of a network, right? Where
there's like no cost on adding a new arrow, right? You don't have to think
about the space or if things overlap. It's like this multi-dimensional like neural network
kind of. That's how I think of it.
Yeah, so just on the search part, I think this is how the whole MIMEX starts is with
the question about search. It says here, the personnel officer of a factory drops a stack
of a few thousand employee cards into a selecting machine, sets a code in accordance with established
conventions and produces in a short time a list of all employees who live in Trenton
and know Spanish. Even such devices are much too slow when it comes, for example, to matching
a set of fingerprints with one of five million on file. Selection devices of this sort will soon be
speeded up from their present rate of reviewing data
in a few hundred a minute.
By the use of photocells and microphone,
they will survey items at the rate of thousands a second
and will print out duplicated of those selected.
And he continues on saying like,
this is not even good enough because we're hindered
by this hierarchical reasoning and we need
to have instead, he doesn't imagine like full text search, he never mentions that here,
but he imagines that there's a bunch of metadata that people are adding to these things and
so you'll be able to search by metadata and he even says, you know, just talking about
Wikipedia, he does specifically say, holy new forms of
encyclopedias will appear ready-made with a mesh of associated trails running through
them ready to be dropped into the mimics and they're amplified.
This is why I was a little surprised by your talking about lack of social implications
and his connecting back to the original problem, he said, I think he thinks the only way we will solve that problem is by this kind of collective effort of people going and, you know, having the ability to share what they've discovered in this vast record with other people.
people. There is a new profession of trailblazers, those who find delight in the task of establishing useful trails through the enormous mass of the common record. So when he says trailblazers here,
he actually means literally people who make trails to publish for other people, right? I agree that
he doesn't see like a automatically mechanized indexes that search for relevance based on keyword.
What he's imagining is a human scale endeavor where people actually full time jobs are to
take in this new associated information associated with things before publish it, let other people
do it, which honestly, I think science sounds more interesting
than what we have. To me, it sounds like artificial, artificial intelligence,
like mechanical Turk versions of some of the ideas here
that have actually come true.
Like there are cases like people on Wikipedia
whose main contribution is going through
and like adding links between articles and filling out things like see also putting things in categories.
There's a lot of bots that do this also that like go through and find articles that don't
have enough connective tissue and tag them.
But like here now in 2025, we know that the real transformative technology here is LLMs. We know that building
up a transformed version of the full content of the source material in a way that makes it
recombinable and searchable in a way that goes beyond just like full text search, right? Like
it's what LLMs are doing is not get me all
of the articles that contain the word mechanism
or get me all of the articles that like have mechanism
in the title or something like that.
They do something fundamentally very different.
And that is the point at which I think we have a technology
that starts to actually address the challenge put forward
in the opening paragraphs of this piece,
which is that you...
I'm gonna completely disagree.
Okay, cool.
Oh!
And I can't tell if you're intentionally trying to do that
or you're being sincere.
I'm being serious here.
I'm not troll, I'm being sincere here.
I think that the pre-LLM technologies
do not do what he wanted to have happen in the opening paragraphs in pre LLM technologies do not do what he wanted to
Have happen in the opening paragraphs and that LLM's are the first time we've had a technology that do that and they don't match
What he described. Yeah
Okay, just to try to clarify
Do you think LLMs have the potential in the future to do that or are currently doing that?
I think they're currently doing it but they're doing a bad job of it and they're only gonna get better.
Yeah, okay.
So, you know, it's harder to argue that like, you know,
they could never do it, right?
So I, but. Sure, yeah.
So let's, I'll take a weaker stance
and say that they're currently doing it.
That LLMs currently do the thing that he wanted
in the opening paragraph and that,
that that is doing it in a fundamentally different
way than what he predicted and that the things he predicted technology wise like the MemEx
are similar to what we had before but that none of those things really solved the problem
that he was setting out.
Yeah.
So let me give a concrete example so that and we could try to say that this is just
bad but I think it's actually just fundamentally why it's not going to work.
So I was trying to do a literature review of the exact sort he talks about here, where
I wanted to know what is the current state of the art on building a virtual machine on
the JVM to perform well.
Okay, which there are things written about this.
They do exist in the literature.
And what I was specifically interested in was and ones that don't require me to change
the JVM.
Okay.
And I did deep research to try to ask this question.
I did a bunch of chats of various models to ask this question and get advice. And all of it was completely,
utterly, not just wrong. It just did not understand at all what I was asking. And you could say that
this is just bad. But well, this is this is an example of something that it fails to do.
But that doesn't mean that there aren't many,
many other examples of things where it succeeds
at doing the thing that I'm suggesting.
Yeah.
Like you found a counter example.
What we're trying to do with Bush's thing
is actually get back to that research
that humans performed.
And what I think we get from LLMs
is trying to summarize that research.
I think that is not at all what Bush is trying to accomplish.
He's trying to make it so that you become aware of the research that's been done.
Yeah.
Okay.
And so like he gives an example that Mendel was not recognized because it didn't get in
the hands.
Mendel's concept of the laws of genetics was lost to the world for a generation because its publication did not reach the few who are capable of grasping and extending it.
So what I think LLMs don't do is they don't give us a way of finding the most recent relevant
documents that were published and understanding the finding those documents so I can read
them to understand the state of research.
What they do instead is try to summarize what has come before.
I think the Mendel example is a great one because for the memex and for Wikipedia and
for you know, the web and Google and those sorts of non LLM approaches to this problem
to solve the Mendel problem. It all depends on
somebody being familiar with Mendel's work and linking to it, adding it to some
corpus, adding it to some, you know, prominent page in Wikipedia, putting it
in a place where people are gonna find it, doing the right SEO keywords for
Google to rank it highly when you search for relevant
things, whatever it is, whatever the process is, it depends on somebody finding it and
making it known, which is the pre-existing problem. It's just technologically amplified.
So let's say that I have Google Scholar and you know, or archive or whatever, right? Like
we put, we have some, what it relies on is a social
agreement that like if you're going to publish things you put it out there so that it can be
indexed in one of these places yeah which i'm not willing to fully grant but i'll let you have that
that social agreement is for now and this seems to be the kind of thing he's imagining is that
there's a collaboration of scientists putting things out there.
And that's what LLMs no longer require. That's a critical difference with LLMs is that they
break the social contract. That's not true at all. How does the LLM get all of its data?
The LLM is out of date. It can't tell me that data that was published yesterday. But that
is a short term technical problem. Like if you don't think we're going to have up to
date LLMs on a daily basis within a decade. I think you can. Like, you know, these things, you know, with all the tool use they
do nowadays, like it's not trained on that, but it can access that information live. But it does
that the fact that it's doing it live mean that it doesn't count for you? But it's using the old
technologies that he say that Ivan is not allowing.
Yeah, which I'm not, I don't count that.
Yeah, I'm not interested in that.
I think like if we're not training LLMs on a daily basis, like updating their world knowledge.
Let me give my example here though of why I think the old technology can solve this
problem right?
You let's just say you publish into some, you know, it doesn't have to be one central
place but you publish into some central places, right? You let's just say you publish into some, you know, it doesn't have to be one central place, but you publish into some central places, right? And now you have, let's say it's Mendel,
you have some keyword search on genetics or something else that you get alerted on Google
Scholar when something is published with these words in it. It doesn't require it to be high
page rank. It doesn't require it to have been linked properly or someone to put in Wikipedia.
That publication happens.
I have a search criteria for it.
I get notified and I can look at it.
LLMs don't have like even if you ask LLMs are weighted in the same way that page rank is where the chances of it saying
What a certain word is based on the statistical probability that that word came after the next word the former word and the yeah
I know the mechanism. Yeah corpus etc. Right? Yeah
so like it's the same problem there that you're worried about with the old technology and
If something new comes out,
the chances of you finding it in an LLM
would be much lower because it is going to be biased
towards the old things because they occur more often
in their data sets.
I don't really want to explore this too much further
because I think it's kind of digging into narrow,
increasingly. Why not? This is fun.
It's fun, but it's getting increasingly narrow. And my argument is going to be, I think the state
of LLMs are going to change in this regard specifically, and that they're going to,
that this is one of the areas in which we're going to see a significant transformation in
the way that work is done, that knowledge work is done.
Not that, you know, oh, we're gonna use LLMs
to replace artists all over the world
or that kind of thing, like I'm less certain about that.
See, that I agree with.
But like, what I'm imagining LLMs being the function of
is not the index.
What I'm imagining LLMs being the function of
is allowing us to very easily build these kinds of indexes.
Yeah, maybe.
That could be too.
I don't disagree with that.
How do you take something that can OCR, that can associate, that can do all of the functions
where it can build up the documents that he's imagining in this-
Yeah, this knowledge graph or whatever.
These trails.
Yeah, this knowledge graph.
I think LLMs would be super useful in helping us build up this knowledge graph or whatever. These trails. Yeah, this knowledge graph. I think LLMs would be super useful
in helping us build up this knowledge graph
that's human led.
Can we zoom out a bit?
Like, so the, you know, there's a couple of,
there's a couple of things mentioned in this Memex section.
You know, one of them is to do with selection,
which is like how to select, I guess, the
thing you're looking for, the things you're looking for.
And I guess like it's, I guess like you could hone in on that task of selection, needing
to select things quickly and efficiently.
Like I guess we can do that without memexes, you know? We can search and select
things very quickly and very easily. I don't think you need a memex for that, but I think
in this article it's associating that task with memexes. Maybe you could say that he's wrong because
you don't need a Memex for that.
Memex wasn't the selection solver, right?
I don't know what was.
Maybe you could make an argument that Google's big index was a kind of Memex, if you think
about it more loosely, whatever.
I don't know.
But the other purpose of the Memex is about these trails more generally, right?
Like, you know, selection can mean a few different things in my understanding.
Like selection might mean I know exactly what I need, give it to me, you know, and find
it in this big stack of files.
Like it's kind of like a very mundane job.
But selection can also mean like, I think, learning
from people who have gone through this data in the past, right?
They see, oh, you looked at this and then you went to that page?
I wonder why that is.
I wonder what the link is, you know?
And it's not necessarily something that you want a computer to do for you.
It's almost like part of the learning is following this trail and, you know,
when you're browsing through Wikipedia, like clicking on the C or SOS. And I think this idea
of using a memex for that, like, sorry, the idea of like making these associations,
for me, it feels very similar to embeddings, in in the AI world where you're like rating how similar things are to each other in all sorts of different ways in the ways that we can't even think about because we're
And the thing is, I put this as a zero out of 10 prediction because LLMs doing these, you know, and models making these embeddings, that's not helpful, you know? Like, I feel like the point is
that we do it and like we join the dots and we are able to inspect those embeddings. Not for it all to be in this big
black box that we can't do ourselves. I can see the link there, but I don't think it's
something we're able to do.
I completely agree with you, first off, on the LLMs making the associations isn't the useful part, it's
us.
But I do think that the web, I think, okay, I think there's a lot of things about this
that I actually really like of the Mimic's original vision that I do think we're lacking
in some stuff.
I think it'd be interesting to talk about the physical desk.
I think that's an interesting aspect here.
But I do want to just like try to, you know,
defend the web as it.
One example of a trail that I think exists,
a practice of publishing trails
is awesome blank GitHub repos.
Yeah.
Right, awesome whatever topic you want, GitHub Repos. Yeah. Awesome whatever topic you want.
GitHub repos.
Awesome front end frameworks.
Awesome React hooks.
Awesome.
Awesome LLM.
Awesome everything.
Markdown editors.
Yeah.
Awesome terminal color schemes.
All that.
And so you would have, this is very explicitly just a trail.
You're getting, here is how these things are associated.
Here is a link to all of them and I can very easily pull all of them up.
I think implicitly you get trails as well in blog posts where it's like I'm writing about a blog post and as I'm writing,
I'm linking to things other blog posts you might be interested in these articles.
You might be interested in this is kind of I am I'm
giving you the overall picture of something or my take on something and I'm giving you ways to go
further and then of course we end up with somebody else's trail as well so like in some ways what the
web did that's different from this is you know Bush is imagining you physically have to get a trail
and then you install it into your memex whereas now all the trails are all completely interlinked together. You
know, you go from one trail and now you're on somebody else's trail, assuming it's a
web page.
Right. Like Gruber at Daring Fireball frequently links to Kotkey, who also runs a link aggregating
blog, right? Like these, these trails link to each other and,
or like Boing Boing and The Verge
and those kinds of websites, right?
They present themselves as new sites,
but they are actually just like,
aggregating and sharing links in a sense.
Or even something like the Wirecutter, right?
Where it's like ostensibly here's a bunch of reviews.
I'd go to Wirecutter not to actually read their reviews because I think their opinions are trash,
but I go to see like what are the top eight brands
that make a product in a certain area, something like that.
Or podcast show notes.
Yeah, sure. What's wrong with you?
What? Podcast show notes are great.
Looking at Oxide and Friends show notes,
they have so many interesting things that I would have never seen.
Yeah, it's like Memex-y.
Okay, I mean those examples.
So for me, like, I think they are indicators that the Memex style is something that people
naturally do and want to do. You know, I'm often writing a blog post and putting all these
like outward trails to other places that people might want to go.
It's like, hey screw it, you know, you might see one of these links
and you decide you want to go to it and not read the second half of my blog post.
Sure, you know, I'm happy to do that.
And, but the thing is like,
when I think about like actually making my blog, all of those pages I put on it, like the, I'm forced to using like folders and files, right?
I can't just upload it to one of these directories, or, you know, really forget the directory altogether and say, these are the things that it, that it links to, that it relates to, you figure it out, you know, like, like I want to do the memexy kind of data structure, but I'm forced to use this tree like thing, you know, like when I first started it, I tried to keep things in categories, like the 100 rabbits wiki is this very tree like thing.
is this very tree-like thing, and I tried doing that for a while,
but actually, you know what, I hate that style.
Like, I hate that kind of Wiki nowadays.
Like, that's not how it works.
It's like, you click, oh my God,
like, you click on one of the categories
of those Wiki pages,
and you're like, you've like, closed,
you've like, you've closed off your,
I don't know, your routes to've closed off your I don't know your
roots to these other places I mean I don't think that's really true but
that's how it feels you know that's how I am isn't that what it's all about you
know the feelings of computer do you know what I mean I'm feeling confused
because I don't know what you mean oh I think this is interesting because, you know, you're right.
Computers kind of fell into this old style way of, despite, I think, you know, people
like Engelbart wanting to move away from it, we ended up with files and folders and all
of our technology kind of assumes these paths.
And I know there's been some reports, some of them being all like, oh, look at how dumb
people are getting, but some of them being all like oh look at how dumb people are getting but some of them
Actually just reflecting on this fact that apparently zoomers are not as into file structures
When they when they do their use their computers for things
They are much more flat and search based and tagging and etc right which I think makes perfect sense
Yahoo the sort of maybe the last major search engine to emerge before Google changed the
nature of what we think of as search.
Yahoo is an acronym that stands for yet another hierarchical officious oracle.
Google really was foundational in changing the way that we approach the computer away
from hierarchy and towards, it doesn't matter where you put it,
we'll find it and we'll serve it to you
through a search interface.
It doesn't matter where it is.
That place goes away.
And that's Google's main advancement.
Yeah, no, I think there's so many things
that have tried to pursue this idea.
I think of Alex Openhour's work on thinking about operating
systems and stuff like that.
I think this is definitely a direction that I
think is really interesting.
I even think of UNISEN getting rid of the file system
for storing code and storing it in a database that's
looking at names or metadata, et cetera.
I think this is a direction that we wanna go.
And I think at the same time,
the web is layered on top of this hierarchy,
and it does provide this way of non-hierarchically browsing.
And if you think about the MIMX,
it's probably internally stored still in a hierarchy
because that's gonna be an efficient way to store it.
And that's the kind of mixing of boundaries
that we have going on with computers,
where sometimes that hierarchy leaks into what we have to do,
because it was a nice, efficient way of storing it.
And we haven't fully abstracted away
from that underlying mechanism.
So did the desk help with this?
With not having to store things in hierarchy?
Yeah, like, I don't know, like, I, I can't imagine, because I can't imagine a
solution to these things I have griped about, right? I can't imagine a non
hierarchical way of like, structuring the data on my computer. You know, so I'm
like, tell me, tell me, Jimmy, tell me'm like tell me tell me Jimmy. Tell me that the desk
Figured it all out. Tell me that we need to bring the desk back. Yeah, so I like this desk. I I think that
Actually making even if you do it backed by digital stuff
Having the physical desk with a physical layout for doing this work seems really interesting to me.
Mm-hmm. Why? Why? Tell me more. Because I wasn't too fussed about the desk. I was like, no, it's just like a...
it's not really the point. It's just like a computer. You interface.
So this paragraph I read before of the scientist being free to walk around and
then of course, you know, you can come back to the desk.
Really does remind me of the dynamic land kind of setup.
This idea of having a place in which you do certain things
that has these physical affordances.
Having that they imagined these two screens
so that you can always be looking at two documents at once,
which, yes, we have multiple monitors. I get it. But they're very dedicated to like screens that focus
on papers. They're projected and you're able to write on them. Right. I think that's really
interesting because you this this like affordance of comparison being a key is very valuable.
I find myself wanting to do comparison between two things often, but
also wanting to be able to write on it. So like you think about like an iPad and I often
do like reading on an iPad and writing on it, but I want to compare something and it
becomes a very difficult task to like compare between my iPad and something else. I think
that's really interesting having certain buttons that you know, you press or levers that you pull for these specific things.
I can imagine like a modern day version of this being some modular desk that you can
put a bunch of buttons.
You can have a bunch of screens in the way that you want it.
It's all you know, there's no mechanism underneath.
They show this great illustration of like pulleys and levers that are moving your microfilm
into place.
You don't need any of that.
I get it.
But I like this physical affordance.
And I think the system that he has in place, ignoring the desk, you can imagine, you know,
digitizing this.
And yes, you could have a completely flat system for storing all your documents, all
of your pictures, all of your movies.
You could do this today on your computer.
It just sucks. Just have one folder today on your computer, it just sucks.
Just have one folder and you put everything
in that one folder.
And then you could use Apple's tags system to tag things.
You could build your own tag system,
you could build your own indexing over it
to search whatever you wanna do.
There's no, the reason you would not do that
is because it's inefficient on the computer.
If you have too many files in one
Folder it gets mad at you
But conceptually you never have you could break those into you know multiple folders never have to think about it again
You know
Datalog is an example of a database system that doesn't have tables
Right, you just put things especially if you do like an entity attribute value,
right? You just put everything in one namespace. Yeah. And you hope that like in each entry,
there's like enough context and info so you know what it is. So it's not like,
oh, what's this record? It says true, you know, like,
yeah, for like entity attribute value, right?
You have this like, you have entities that are just some ID, and you have the name of
the attribute and the value.
So it would be like this entity, you know, is cool, true.
Enabled true.
It's like what's it?
What's enabled man?
What's cool?
Yeah.
But yeah, I guess I think that there's, there's upsides to hierarchy, which is why we haven't gotten rid of it.
Yeah, it I do find myself like, it's so many of the things I do, like either my own personal, just like daily life stuff, like managing photos or like, hey, I was writing some notes about my D&D character yesterday,
you know like I'm constantly presented with these hierarchical interfaces for these things and it
just like never fits you know so I just go for like the 90% solution and there'll always be some
time where I come back to it like you know know, a couple days later or a month later,
and, you know, I'm not able to like navigate around it in the way I want.
It's like, sure, I could add tags or I could manually like add these links,
but it's kind of like these, the UIs that I find myself using, that's not a first class thing.
You know, the first class thing is to do the hierarchy thing.
And then it's like a bit of work if you want to go
the extra mile and like link things around.
And maybe then you have like duplicate records and,
or maybe like one of them is like a primary record
and one of them is like the secondary record.
And if you delete the primary one,
then the secondary one dies too.
Like if you decide actually this is not the hierarchical way
to organize it. So I guess I have this hunger for stuff that is more like spindly,
you know, with like, with like small legs coming out and trying to grapple onto other
other legs. Yeah, I want I want legs.
You want spiders that could crawl over a web of information.
I want to be a hundred-legged spider.
This is where I think LLMs could be really interesting to help out build this web of
information, because it is more friction to tag things, to categorize them. But you could imagine, you know,
when you're taking your D&D notes,
as you're going through this,
a little LLM is looking at what you've put on the screen,
whether it's pictures, whether it's your notes, whatever,
and then on the top right corner,
there's your suggestions of tags.
And it's actually meaningful, useful,
and it takes one click to just add it or it could do it in just you know
Just in time really but it could do it
You know giving you those suggestions and letting you organize this in the way that you want without a whole lot of friction there
Someone make this someone make it so I don't need it out of them for this that's even better
You know bring the desk back and put it on my phone.
I can whip out of my pocket.
Like a mini desk, like 3D pop out.
I do.
I think I've lost the plot at this point.
I think I've lost the plot.
I think that, you know, this non-hierarchical is super interesting.
And I think, you know think a lot of people wanted
to explore it.
The reality is, hierarchy is valuable and so you're going to end up recreating it in
some places.
It's so much easier and cheaper to just make hierarchy that people haven't thought through
what we could get on the other side.
That's what I really think is the case.
Ivan, though, might disagree with me because he usually thinks I have these like,
oh, well, people of all conversion is because it's practical and useful.
Like, you know, text-based programming languages.
And Ivan's like, oh, no, it's just because they're not cool enough.
No, it's just because it's historical precedence, right?
Like, it's hard to break out of the grooves that we're in.
And I think hierarchy by default in computing,
the sort of the default hierarchical organizational scheme
is very likely as prominent as it is
because it has a strong physical analog
that like each thing goes in one place,
each thing goes in our place.
You know, hierarchy is powerful.
I think, yeah, hierarchy is powerful.
Hierarchy is very useful, but it is also,
it makes sense to me that it's the default
that we're trying to pull away from.
I could imagine the opposite being,
like, I'd be very surprised if the opposite were true,
if computing started non-hierarchic,
and we've had to, you know,
gradually introduce hierarchy back into it.
So Ivan, what's your rating for the Memex?
Oh, for the Memex, a 14 out of 10.
What?
Because it presages hypertext and the web
and basically the way that computer information is
organized.
It's also delicious bait for a whole bunch of venture capital bro types who want to bring
us the real future of coding.
It's the culture.
It's the meme, right?
It's literally spelled meme X because it's a meme.
That's why Bush, like Bush said, you have to just to pick a name arbitrarily,
consider a future device for individual use, which is a sort of mechanized private file
and library.
It needs a name and to coin one at random, meme X will do.
Yes, meme was definitely a word in 1945.
Yeah, Chomsky, you know, is a time traveler.
That's a, oh wait, no, it wasn't Chomsky.
It was Dawkins.
Dawkins.
It was Dawkins, right.
It was the attack scientist.
It was that sh**ty awful person went back in time,
could have killed Hitler, didn't,
gave Bushy this terrible name.
And that's how we got the future of coding.
I actually realized in the course
of recording this podcast that the future of coding isn't like us saying here's what
the, you know, the future is and like a sort of a, you know, we know the answer. We're
going to tell you, you know, we're, we're the coders and we're here to like dictate
reality. No, it's actually like kind of like an alternate history kind of thing where it's
like, what if we're living in the timeline where coding, this is the future
of coding, like coding happened and we're living out the consequences and we're here
to like warn you don't have the future of coding, like don't code. Yes. Get away from
that. This is an anti programming podcast. Do you really want a feature of coding?
Do you?
No.