Discuss this episode in the Muse community
Follow @MuseAppHQ on Twitter
Show notes
00:00:00 - Speaker 1: There just really doesn’t seem to be an effective
concrete practice for taking like day to day insights and accumulating
them, like rolling them up into a snowball of novel ideas.
00:00:16 - Speaker 2: Hello and welcome to Meta Muse. Muse is software
for your iPad that helps you with ideation and problem solving. This
podcast isn’t about Muse product, it’s about Muse the company and the
small team behind it. My name is Adam Wiggins. I’m here today with my
colleague Mark McGranaghan. Adam, and a guest today, Andy Matuschek.
Hello, thanks for joining us today, Andy. I think you’re about as close
as there is to Rockstar and the tools for thought space.
00:00:39 - Speaker 1: That’s a really distressing statement.
00:00:42 - Speaker 2: Yeah, we’ll, we’ll talk more about why this
space is so small a little later on, but for those that might not know
you that are listening, maybe you can briefly give us your background.
00:00:51 - Speaker 1: Sure, I’ve kind of a meandering background. It
begins in technology. When I was a kid, I was constantly developing
video game engines and kind of these tools for creative people. I, um,
with a couple of roommates, I worked on the, the first native Mac OS 10
graphics app and did that for a bunch of years and then made some open
source software for developing.
I was always really into tools for others.
Went off to Caltech and kind of got introduced to science, serious
science. And uh kind of got my, my very pragmatic engineer perspective
But unlike all of my peers who who went off to get a PhD, I, I went off
to Apple and got a different kind of, it kind of felt like a graduate
program of studying at the, the heels of all of these people with like
jeweler’s loops that they were using to to look at individual pixels of
devices and There, there my work became much less about just programming
and much more about kind of the intersection between technology and
design. I, I got myself involved in in all these projects that it kind
of the through line was that they, they were about what was central to
dynamic media, uh, as opposed to just pictures on screens. So things
like, you know, interactive gestures and like the 3D parallax effect
and, you know, crazy page curls and And all this stuff we’ve talked
00:02:07 - Speaker 2: Uh, the way that Apple’s environment maybe has
less of that distinction between design and engineering or there were a
lot of people that sat really on the intersection of those two things
and it was part of what allowed them to do and continues to allow them
to do really innovative things on interface and and maybe you’re a
person that sits in that place as well, right?
00:02:26 - Speaker 1: Sure, yeah, yeah, it’s it’s interesting because
like from an org chart perspective, there’s really heavy boundaries
between engineering and design, and like I was on the engineering side
of the house, like I sat with the engineers, but uh for several years,
I, I would like Spend much of my day sitting in the human interface lab,
like next to a designer, and we’re just kind of like tossing prototypes
back and forth all day. And so it became this kind of mind meld thing
where those people could tweak values in the prototypes I built and you
know, I would end up tweaking design elements as I was building
prototypes and it kind of just the titles fell away.
But over time, I kind of, I began to feel that these experiments we were
doing with the dynamic medium, I would love to see them applied to
things which had More, more meaning, more impact in the world. And so I,
I got really interested in, in education research. I started writing
about that. And uh the folks at Khan Academy reached out and asked
whether I’d like to do that kind of work with them.
Um, so I joined Khan Academy and and took along, uh, one of my Apple
colleagues, Mei Li Ku, who is a wonderful designer and, and together we
started this like R&D lab, uh, at Khan Academy where we explored all
kinds of uh novel educational environments from that perspective of like
trying to trying to look at what the dynamic medium alone can do.
Trying to make these active learning environments and I did that for
about 5 years and um I started getting a little disillusioned with
institutional education and um I started getting really interested in
the kind of knowledge work that people like you and me do every day,
where you’re reading information, writing information, creating new
things, pursuing uh novel ideas every day, and I’m wondering how we
could augment some of that.
Uh, so now I have this kind of independent research practice where I’m
pursuing oddball questions like what comes after the book? Can we make
something that does the job of a book but better? Uh, it’s just been
sort of a delightful experience.
00:04:12 - Speaker 2: And I think one of the uh pieces you’ve written
in all your writing is delightful, and I certainly recommend everyone uh
read it, but uh uh read as much of it as they care to. But when I’ll
link to because I think it particularly illustrates maybe the place
where you and our team kind of overlap and thinking is the
transformational tools for thought article, which both describes sort of
your current work around the the learning and the space repetition,
which you can tell us about, uh, but also the kind of the meta elements
of how do we develop these kinds of tools in the first place.
00:04:43 - Speaker 1: Yeah, that that was a project with uh my wonderful
colleague Michael Nielsen, who’s also been investigating the space
which we might label tools for thought. And people have defined this in
different ways that the term stretches back some decades, but uh I like
to think of it as tools or environments which expand what people can
think and do. And you know, a great example of this is writing. Another
great example is numerals. So there’s a tendency to, to think about,
you know, kind of computer implementations of these things and of course
there are instances which are very interesting. Um, I find it very
powerful to reach back to you know, these, these cultural.
00:05:16 - Speaker 2: Uh, ancestry tools for thought.
Absolutely. Another great example of that is, I think Brett Victor has a
piece about this, which is essentially the chart, is the charting
numbers, you know, on an X Y axis or, you know, line graph or that sort
of thing that we we take for granted nowadays where it’s easy to crank
that out in a spreadsheet or whatever, but that was an invention that
happened not even all that long ago. It’s, you know, a couple 100 years
back or something like that and the existence of this new. Um, tool, or
actually, I think as you argue in that piece, medium, you would even
call it a medium for thought, might even be more accurate, basically
allows you to have new ideas or see the world in a different way. So the
tools shape the kinds of thoughts you’re able to have and the kinds of
works that you’re able to create.
00:05:58 - Speaker 1: That’s right. If all you have is Roman numerals,
Roman numerals, uh, then it’s very difficult to multiply.
Suddenly, if you have Arabic numerals, it becomes quite easy by
comparison. So kind of in the what comes after the book space, one of
the things that my colleague Michael and I had been exploring is just
this observation that most people seem to forget almost everything that
they read, uh, and sometimes that’s, that’s fine.
The thing that really matters in a book is, is the way that it kind of
changes the way that you view the world for many books that really is
the impact that matters. Uh, but for other books, for instance, if
you’re trying to learn about quantum computation or some advanced
technical topic, uh, it really is kind of a problem, uh, that, that you
forget. Uh, most of what you read because these topics build on each
other as the book continues. And so you end up starting reading a book
in English, say, and then halfway through the chapter, uh, you start to
see there’s like a word of Spanish and, and then by the end of the
chapter, there’s like whole sentences of Spanish and then then like the
whole second chapter is in Spanish and say that you don’t know Spanish
as a language, you read this book and you’re like, well, I thought I
was reading an English book. It’s like, no, it’s actually written in
this other language that you have to. Learn, just as you would have to,
you know, learn vocabulary, if you were trying to speak a foreign
language, you need to like learn the vocabulary, both conceptual and
declarative of this domain that you’re seeking to enter. Uh and so, and
so the experiment is kind of been, well, can we make that easier? A
project that that paper describes is this textbook called Quantum
Country, which tries to make it effortless for readers to remember what
they read. Um sounds like kind of a crazy thing, but It takes advantage
of really a fairly well understood idea from cognitive science, about
how it is that that we form memories. It’s reasonably well understood.
There’s sort of a closed set of things that you need to do in order to
form a memory reliably. Uh, it’s just that like logistically, it’s
kind of onerous to do those things, and it requires a lot of
coordination and management. And so most people don’t do it or it’s
kind of difficult to do it. Uh, but it’s pretty easy to have a
computerized system assist these things. And so, basically, as you’re
reading this book, every 10 minutes or so of reading, there’s this
really quick interaction where, you know, say you just read about the
definition of a qubit, after a few minutes of reading, there would be
this little prompt interface where it’s like, hey, so how many
dimensions? Does a qubit have? And you try to remember like, uh, how,
OK, it’s two dimensional. So you think yourself 2 and then you reveal
the answer and it’s like, oh yes, it was 2, and so you say, cool, like
I remembered that. And then we say like, OK, so a qubit is really a two
dimensional what space? Like, how do we think about representing this?
And say you don’t remember that, it’s this linear algebra concept. OK,
it’s a vector space. That’s fine. Like you reveal it back, you didn’t
remember that. See market is like, I like, I didn’t remember that
detail. And um this is already doing something for you because it’s
kind of signaling like, hey, maybe you weren’t quite reading closely
enough or just seeing that answer that you missed, like as you read the
next section, if that topic comes up. Maybe you’re more likely to
remember because you were just uh corrected and you saw that correct
answer. But somewhat more importantly, 10 or 15 minutes later when
you’re looking at this, this next set of prompts, and you, you see kind
of the new things from this section, that prompts about the two
dimensional vector spaces that you failed to remember, that one will
appear there. And so you’ll, you’ll kind of get another chance. And
then once you remember it there, the idea is a few days later, we will
send you an email and you’ll say like, hey, uh, let’s let’s remember
these things about quantum computing that you were working on, let’s
work towards long-term memory, and you’ll you’ll open up that review
session and linked in the email, and you you’ll kind of do this
interaction again, just, just a couple seconds per question. It takes
about 10 minutes to go through the material. And that 5 days later will
kind of reinforce your memory of that material about as well as the 10
minutes later prompts did, not, not exactly, but, but just roughly you
get the idea. And then if you remember things after 5 days, then, you
know, maybe you will next practice them after 2 weeks and after a month,
after 2 months, after 4 months, and so it initially seems like this kind
of onerous thing, like, oh, I’m gonna like be working on these like
memory flashcards for this thing I’m learning, but Because the way
human memory works is that it’s stabilized in this kind of exponential
fashion where you can have successive exposures that are further and
further apart. Uh, it only takes a few exposures before a particular
idea can be remembered durably for many, many months at a time.
00:10:11 - Speaker 2: And this is a space repetition systems you’re
talking about, um, which I had some exposure to through Onki, which is
this little kind of I don’t know, uh, it’s definitely a tool for
thought, but it is, uh, very nichey, I would say more than a little
You have to be really motivated to do it. And so you can use a tool like
this to increase your retention or understanding of something you’re
reading a science paper, a book. Something you you do want to get a deep
grasp of, but you got to really work hard at it, right? The tools are
very taping it all together yourself in a way that requires pretty big
commitment and investment.
And one of the things I think is really interesting about the work
you’re doing is whether you can take that and build it in a way that’s
fun, relatively low effort by comparison, maybe even you know, sleekly
designed and just more, more enjoyable overall.
00:11:06 - Speaker 1: Yeah, one thing that characterizes, I think a lot
of opportunity in this space is that there are many exciting ideas which
have been explored by technologists or by academics, which are promising
at some foundational level.
The underlying mechanic of Aki is fundamentally the same as the
underlying mechanic of quantum country if you look at it from a certain
angle, but there’s this core design piece missing, that’s kind of
keeping that idea from really having the transformative impact it could
By that, I don’t mean the fact that Aki is like hideous. I mean, it is,
and, and it will kind of like turn off basically everybody who looks at
it for that reason. But there are deeper issues to your point, it’s
really hard to write good prompts. Uh, both in the sense that people
start by being bad at it, and so they’ll write prompts that don’t work
very well and that are boring and onerous to review, and they mostly
won’t realize that that’s what’s happening. They’ll just think like
that’s what this is. And then also in the sense that even if you do
know how to write prompts well, it’s quite taxing. It takes a lot of
effort. It’s a context switch from the experience of reading and it’s
valuable insofar as kind of reflecting on material that you’re studying
and synthesizing it, distilling it and turning it into a question
actually does. go quite a long way to enforcing your your understanding
of the material, but maybe you’re only going to do that for like the
most important things in your life. And it’s pretty interesting to
wonder like, OK, maybe you do that for the top 10% of the stuff that you
ever read, but what if it was like really pretty easy and low effort for
you to remember the top 70% of the things that you could read. You could
save that special effort for the stuff that really, really matters. Um,
that’s kind of what quantum Country is pursuing. One of the main things
it’s wondering is, can we make this something. That it basically
everybody who’s reading it and is serious about the topic can take
advantage of and really see the benefit of.
00:12:49 - Speaker 3: I think this thread also reflects one of the
challenges in developing new tools for thought, which is you actually
need a lot of different skill sets. It’s not just a matter of
engineering or computer programming, you need engineering, products,
design, writing, marketing, community, often you need at least all of
those things. And I see a lot of people approach the domain as basically
pure engineers and they they. Tend to kind of bounce off or the products
don’t stick because they’re missing a lot of those aspects.
00:13:15 - Speaker 1: That’s right. And I’ll add one more actually,
that that’s kind of Michael’s in my hobby horse here, which is that
you probably also need some kind of domain expertise.
So many of the, the projects in this domain, even if they do actually
have the design skills and the technical skills involved as well as some
of the other peripheral skills, they’ll be doing things like trying to
make a tool to do math better or something like that, but no one on the
team is a serious mathematician. And so they’ll make something that
seems really cool and it makes for a really good like product
presentation, but no mathematicians really going to use it to do serious
Maybe it works in an educational perspective, but it’s fundamentally
limited. It’s it’s like a toy in some fundamental fashion. And so to
that list, I would add, you need some kind of deep domain expertise too
for a product like Muse, maybe that is somewhat diffuse. So anybody
working on a product, the domain expertise that’s relevant there might
be like, you know, the visual design of a product or like doing this
kind of conception stages of a product.
00:14:11 - Speaker 2: Well, our domain is thinking. So luckily we have a
domain expert on that, and that’s Mark, right?
00:14:15 - Speaker 3: Yeah, I feel like a sort of secret that we have,
we had with the lab and now we have with use this understanding of the
creative process and thinking and a lot of it actually comes. From the
study of how this stuff happened historically. And you mentioned
reaching back in history and learning from that something we’ve done a
00:14:29 - Speaker 1: Yeah, it’s fantastic. I think it’s just really
attractive to build tools.
It is built into my DNA like I grew up that way, and it’s actually a
My tendency when I see an opportunity or I see a problem space, is like,
oh, wow, like I’m going to make a tool to like help with that.
And that’s like a useful tendency, it’s a cool tendency.
But often, I’m not like really solving a burning problem, or I’m
solving an abstract problem that isn’t connected to something that is
like concrete and intrinsically meaningful and that like actually is
about doing the work. So like the analog and muse would be if maybe
I’ve done like one serious creative process that was about like a
concrete thing, and then I. Like, wow, like I’m really interested in
the creative process. Like I’m going to devote, you know, the rest of
my days to working on building tools for the creative process, which I
like, I’m never really using to do any subsequent serious creative
process. Like I’m I’m doing it in order to make the tool because I’m
fascinated by tools. That’s a tendency that I have that I have to
00:15:24 - Speaker 2: The other thing that comes with it, if you come
into building a tool with the domain knowledge.
Is that over time you get focused on building the tool and maybe you
actually know the domain less well.
So there’s there’s quite a parallel for me personally between uh
Hiroku and Muse in that both are some kind of creative process.
Hiokku’s web development, um, which is one kind of one kind of
creativity, one kind of creation, act of creation with Muse’s, it’s
thinking and reading and making decisions.
In both cases, there is a process where a thoughtful professional sits
down and they start in one place and they end with a solution or a
And studying and understanding that process both it’s fun for me to
introspect for myself, but then the the ethnographic research aspect of
going out talking to in the lab and in the build up to Muse, we talked to
hundreds of creative professionals about their process, which was always
an interesting thing because of course it’s this very private and
intimate thing and also I would say 98% of the time people are vaguely
embarrassed because they feel like it should be better.
It’s like, oh, my notes are really messy, or yeah. Yeah, you know,
don’t look at my office. It’s, you know, things are, I, I should have
some, I don’t know, some they have some idealized version of what it
would what it would look like the reality I think is the creative
process is messy and that was something we we fed into Muse was sort of
embracing that a little bit.
00:16:46 - Speaker 1: I think it’s critical that you all not only
experience that ethnographically but also personally that you have this
deep personal experience of that process. Otherwise I fear it’s too
The insight from the last year that I’m most excited about is is kind
of this nugget in the middle of the the paper you you referenced, Adam.
I call it like that the parable of the Hindu Arabic numerals. I hope you
don’t mind if if I kind of recap it here because it just seems to
It’s this observation that if you are the Roman royal accountant and
you’re just struggling through these tables of numbers and you find it
very onerous and it’s kind of taxing and it’s error prone, imagine if
There was like Roman IDEO and you could go to them and say like, hey,
please help me like with my accounting process, please redesign this.
You know, IDEO’s process is pretty amazing in a lot of ways. They’ve
helped make a lot of really powerful products, and they have this
process that is really interesting where they go and they they embed,
they will like sit with the accounting departments and like interview
extensively as you talked about interviewing people about their creative
process and like really try to internalize it, they’ll do all this like
synthesis and diagramming. And they’ll come up with words to describe
what people are doing, and it’s all great, but I think there’s just no
way that Hindu-Arabic numerals would be the result of of that process
if, if what you’re starting with is Roman numerals, because the
transition requires the deep insights of a mathematician and also deep
insights of a designer. So just for instance, place value, this notion
that like if I have a 6 and it appears in the right moment. Spot, then
it’s like a one digit, but if it appears in the second or rightmost
spot, that 6 is still 60 in certain fundamental ways, and you can still
perform the same fundamental operations on it, like with addition and so
on. It still works the same, but it has this alternate interpretation of
being like 60, it’s in the tens place. That is a profound mathematical
insight that depends on deep intuition of like commutivity, the laws of
distributivity. Uh, it’s not something that somebody just like doing
some ethnographic research in the field is going to come up with, yet
simultaneously, it’s also not something that most mathematicians are
going to come up with. And so it’s a great example of how you like, you
really have to have the same, the people on the same team.
00:18:57 - Speaker 2: That is a great example of the domain knowledge,
and I wonder if that connects to something.
I feel like I see the trend of people with design as a skill set. I feel
like are more often drawn to what I would call consumer or sort of end
user things. So they’re more interested in working on social media, you
know, let me get a job at Instagram or Facebook or something like that.
And I wonder if that’s because then they only need to be an expert in
the design domain, and if they’re working on something that’s more um
for an end user that’s not really a specific domain, you don’t need
that knowledge or the things that you need. To understand the problem
space of Instagram is not deep specialized professional knowledge. It’s
just being a person with a smartphone that likes to take photos and post
00:19:40 - Speaker 1: They can certainly be a lot more successful in
People are sometimes surprised that Apple doesn’t really engage in
anything that looks like design research, and here I use that word to to
kind of mean that the ethnography that you’re describing user
interviews, the walls full of sticky notes where you’re trying to like
And summarizes your quotes. The Apple designers don’t really do that.
But they’re primarily designing products that solve problems in their
lives. Like I use email, like, let me make this email a little nicer,
and so like they can do that.
But I think as soon as you leave that domain, things start getting hard,
like Apple iBooks, there aren’t a lot of like really serious readers on
the design team. I think that’s part of why Apple iBooks is not good.
The various attempts at social music platforms, that’s something that
requires like a set of ideas that have been pursued by various products.
It requires like, you know, kind of a landscape review, understanding
people’s social interactions really deeply, that’s also not part of
the process. The Instagram designers, I think they are doing something
that the Apple designers aren’t, they’re talking to users a lot about
how they feel when they’re interacting socially, and that’s a piece
that has always been missing from Apple’s process, but to your point,
they’re not this like goal of of taking and sharing photos. That’s
something they already like.
00:20:52 - Speaker 2: Well, we’re already pretty far into it here, but
I feel like I should um stick to our format, which is introducing the
topic. Maybe I’ll do that here and Andy, you, you suggested this one,
which is uh environments for idea development, particularly idea of
development over time. I thought it might be interesting to compare what
that phrase brings to mind for each of us.
00:21:12 - Speaker 1: Sure. So one of the hobby horses I’ve been
thinking about recently is, I’ve been reading this literature on
deliberate practice. Eriksson is maybe the prominent individual there
and there’s this, this extensive research on the practices of dancers,
musicians, athletes who have these very formal and intense. Hence
preparation and practice structures that stretch from youth into
eminence. So touring international pianist is still working on these
like fundamental skills and activities. And I think it’s fascinating
that by contrast, knowledge workers really don’t seem to take their
fundamental skills all that seriously insofar as kind of like improving
them in a deliberate daily ongoing way.
00:21:51 - Speaker 2: Yeah, I’d be curious to even just enumerate what
we think are some of the foundational or some of the core skills for a
00:21:59 - Speaker 1: I was about to try to do that because I think it
actually connects to this to this phrase. I’m sure that y’all could
add some more, but I think reading effectively is is one of them,
writing, communicating effectively is one of them.
But taking an inkling and developing it over time effectively seems like
another just really important idea of creative work.
And so that that’s what made me suggest the topic that if I speak to
people and ask them like, hey, so you know, this kind of interesting
notion comes out of a conversation, and you think like it might be worth
pursuing, then what? People’s answers are uh. They’re not good, you
know, and like people do come up with things, they managed to develop
ideas in spite of this, but it’s clear that this is very haphazard, and
it doesn’t always feel like haphazard in a good way.
People will say things like, well, you know, maybe I write it down in my
notebook. It’s like, well, and then what? Well, uh, maybe later I’ll
like flip back through and see it, like, no, no you won’t, uh, or, you
know, you can like you can schedule time, you can like put aside time to
like think about that idea, and maybe if it’s like a really important
idea you’ll do that. But you won’t for like, you know, something cool
that comes out of a conversation that seems like it might connect to
something later. There just really doesn’t seem to be an effective
concrete practice for taking like day to day insights and accumulating
them, like rolling them up into a snowball of novel ideas over time,
except insofar as, you know, they kind of happen to accumulate in your
00:23:17 - Speaker 2: Yeah, that makes sense and obviously connects very
well to the To the Muse story for me, it’s become because of this
product that I now obviously have been using in the process of our team
Because it for me represents the place I go to do my deepest thinking.
There’s almost not quite a ritual, but let’s say when I, when I go to
make a muse board for something that I feel like is something I need to
do a deep dive on, I know I’m really getting into it. That signals it
Almost to the point that sometimes I’m, it’s an idea I’m excited to
explore exactly what you described, like the team is having a
conversation, something serendipitously comes up. I think I should
I think there’s something there. I put it in my notes to do that. So
A fun, exciting opening a new door, opening a fun Pandora’s box kind of
But it can actually also be the other way around, which is I know it’s
maybe more of um something important to insult to research or understand
deeply that maybe has is a problem in in my personal life or like a
government paperwork thing or some other something like that.
And I just know, OK, I’m going to really get into it.
This is not shrugging it off. This is not quickly jotting down a couple
of quick notes in my notebook and moving on by creating this board. I’m
kind of mental. Making myself a commitment to follow this rabbit hole as
deep as it goes until I feel like I have my head around the problem or
or I’ve solved it, which is sort of an interesting effect, mental
effect that the product seems to have on me.
00:24:36 - Speaker 1: It’s really interesting. Can I ask the and then
what? Like something comes up in a team meeting and so like you add it
to the muse board. What’s the and then what? How does that idea grow?
00:24:45 - Speaker 2: Yeah. Well, importantly, I wouldn’t add it
straight to muse from the meeting. I would put it more into my kind of
like inbox GTD style. Like just stop it’s the same it’s the same list
where I put down, um, you know, we’re out of we’re out of milk, you
know, get more, it’s just like little notes here.
Another way I’ll think of it sometimes uh in team meetings is realizing
we need kind of an internal memo to pull together diverse thoughts on
the topic and like really articulating what the problem is, um, and
really trying to lay it all out so that not just for my own thinking but
so we can all sort of be on the same literal page about.
Something, particularly maybe something that’s a long time ongoing
problem and there’s people that weren’t on the team before and they
don’t have some of the past contexts you want to put it all together.
Yes, so then what for me is deciding I want to devote a chunk of time to
this, you know, maybe it’s 20 minutes, maybe it’s an hour, maybe it’s
more to really dig in, to really just face whatever this is head on and
And you know, maybe it’s something like an idea for a new product
feature, for example, which again tends to be more on the fun. Uh, the
fun side of things. And so then, then there’s this whole process
around, you know, let me assemble prior art and get together some ideas
and sketch some things and all this kind of stuff.
The output varies, but sometimes there’s just a clear insight of like,
oh we should do X, it’s a decision basically, and then I will go and
take action on that, but other times it’s realizing, wow, this is a
really much deeper hole than I thought and You know, it needs more
thought or it needs more whatever.
And then maybe I want to, for example, it’s a team activity, maybe I
want to bring it back to the team and say, we thought we could, I
thought I could think about this briefly, have a solution, and then do
it. But actually it’s a lot deeper than that. What do we want to do? So
I think it’s, I think it’s just like understanding or not quite
enlightenment, but getting to this new place of understanding about
whatever the thing is, and then that in turn implies a next action.
00:26:38 - Speaker 1: One of the questions I’ve been exploring in this
space is what to do when it’s not really possible to make a lot of
So talking with people about their practices, one common approach that I
hear relates to what I just heard you articulate, and that’s that
something kind of reaches a threshold of interestingness or apparent
importance. And at that point, you’re going to like carve out some time
and sit down and really think about the thing.
That’s cool. And sometimes that is enough. I noticed that for a lot of
the most interesting ideas that I explore, one session doesn’t often
really doesn’t yield all that much. In fact, often it doesn’t
necessarily feel like that session really produced a significant
increment at all. Uh. You’re just kind of like manipulating the terms
of the equation, so to speak, getting a better handle on it. And so one
element that I noticed often really seems to be lacking from people’s
processes, because it’s kind of it’s hard to orchestrate is
marination, where it seems like sometimes what ideas need is just kind
of consistently returning to them over time and asking like what do I
have that’s new to say about this difficult question? OK, I can say a
few sentences about it that seem kind of new, like it’s interesting,
but it’s still not. Something. So I’m going to leave this for 2 weeks
and I’m going to come back and like, what do I have that’s new to say
about this? And maybe if you do that, you know, 6 times, something
starts to emerge. That seems really difficult to orchestrate.
00:27:58 - Speaker 2: It makes me think of a great article called
Solitude and Leadership, which basically is describing how you need to
You basically need to disconnect from the opinions and influence of
others in order to have original thoughts.
One way that the author talks about it is in that first session, like
you described, at the end, everything that you’ve come up with a
written. Down is really in a way just the thoughts of others that
you’re echoing back. And that’s fine. That’s a starting place, but to
truly get to something original or new or potentially breakthrough, you
Yes, he claims that he can sense when he’s sort of like sort of cross
from the more mundane thinking and into the more excuse visionary for
lack of a better, better word or just original, uh, when the thoughts
start to not just be an echo of what he’s read or seen or heard
And that always requires multiple sessions.
00:28:49 - Speaker 3: I think this also points to the idea that you
can’t always expect to sit down in a series of sessions and then kind
of one step after another, produce an idea all kind of in the forefront
When we think about thinking and ideas and tools for thought, we have
this very conscious perception of it.
It’s like I’m sitting down, I’m going to come up with something
that’s better than Roman numerals. At the end of the session, I’ll
have, you know, Arabic numerals. I think that’s just not how it works.
Usually, sometimes you can get away with that, but often it’s more of
your, like you said, marinating on stuff. That’s becoming this fodder
for your mind and then in the background, you’re having an unconscious
process of ideas, connection forming, inspiration, and then when you
come into a later session, you might be better prepared to have a new
idea. So I think it’s like you said, it’s really important to find
ways for the tool to support that marination, chewing, ruminating, going
over, rearranging without the expectation that you’re going to be
explicitly building up your new idea.
00:29:39 - Speaker 1: It’s really easy for tools to accidentally build
One of my favorite novel reading tools is this. liquid text, totally
fascinating set of interactions for manipulating PDFs, excerpts, things
like that. One very interesting design decision is that by default
documents are kind of a workspace and so you extract excerpts into like
this canvas and you can manipulate them, but documents are kind of
separate from each other in that sense.
So you can have a set of insights about a document, but if you’re going
to have inter-document insights, that’ll depend on your memory.
Now there’s a fix for that, which is that you can create multi-document
You can say like, well, this is like my thinking about the this.
Problem, you can kind of like bring several PDFs into it and kind of
like make your notes and make your excerpts and whatever. And that’s
cool because then you can have insights between them, but it still
requires this intentionality of saying like, cool, I’m gonna like bring
that PDF into this workspace and then like the notes and excerpts and
whatever like they live there. But if you’re working on several
interesting questions and ideas at once, it’s not at all clear that
you’re going to have interactions between those workspaces that are
00:30:40 - Speaker 2: Yeah, liquid liquid text is great, but I think as
a coming back to the environments for idea development. That creating
room for serendipity without just total chaos is maybe a subtle and
00:30:53 - Speaker 3: I’ve thought about ways, by the way, to do this
not subtly. One notion I have for an experiment is the idea collider. So
you have something like your, your notes or your wiki pages, and every
morning it just gives you two random pages and it’s like write a third
page, which is the synthesis of these two things. Oh cool. I’d love for
someone to do that experiment. Have you tried it? No, no, it’s kind of
a it’s open. Request for research. So if anyone listening wants to
develop it, let us know. That’s great.
00:31:14 - Speaker 1: It connects to a set of ideas that I’ve been
exploring for the last year or so.
I’ll share it, maybe that’ll generate some more.
I’ve been doing this kind of strange note taking practice that really
came out of trying to solve this problem of like, how, how can I make
marination effective? How can I, how can I make a process where I can
like do something every morning and cause there to be increments on my
understanding. of some ideas or some problems I’m trying to solve.
And so I have something that’s kind of like a personal wiki basically.
The technology is not really important. It’s more about the practice
that’s important and the practice is that I try to write these notes
that are densely linked to each other where each note is about a
Sometimes the note is a question like what are the most important design
considerations when writing prompts for the mnemonic medium like one
country and sometimes for Since the children of that note are
declarative statements like space repetition memory prompts should focus
on one idea, and then that note will kind of accumulate not just in one
session, but over many sessions, all of the things that I have to say
And sometimes I’ll learn that the title was wrong. It’s like, oh,
actually they shouldn’t always focus on one idea because sometimes
it’s really good for, you know, these memory prompts to like synthesize
multiple ideas and these things kind of evolve over time, a term that
some have used is is gardening.
Uh, I call these like evergreen notes because they’re trying not to be
fleeting notes, like notes from a meeting that you’ll never really
return to, but rather uh notes that you water and which grow over time.
And just to get back to your idea, Mark, one of the practices that I
found most rewarding here is this notion of a writing inbox, where when
something seems interesting or juicy, I have a place for it to go, and I
start my writing most mornings by looking at that writing inbox and and
training. those as a set of provocations or prompts and asking like,
which of these things do I feel like writing about this morning.
In this way, ideas which seem promising, even if there’s already a lot
written about them, I can kind of throw them back in the inbox and then
it’ll like it’ll appear for consideration on upcoming mornings. But I
think that inbox gets even more powerful if you start to introduce
fancier orchestration methodologies into it. So one possible
orchestration methodology is like the one that you just mentioned where
like maybe the inbox this morning. contains these like pairs of notes.
Uh, so it’s going to kind of combinatorically like walk my tree here.
But another thing that seems pretty interesting and that I’ve been
playing with is this idea that I had this interesting idea in a
conversation with someone. I don’t really know what to do with it yet.
It still feels promising, like, I don’t want to lose it, but I also
don’t really have anything more to say about it right now. So I can
like kind of snooze it for a while. It’s like, OK, I can go out of my,
my writing inbox for a while, and it’s familiar from Gmail. And then
It’ll like come back in a while, but in a modification on the snoozing
functionality that I’ve been finding very interesting is the
parameterless snooze. Normally you have to say like come back in a week.
I think that kind of overhead is unhelpful and is often
counterproductive and it’s better to just say like, no, not today. And
to say like, well, if I’ve said that 10 times, then like, probably this
should just go away a long time.
00:33:57 - Speaker 3: Yeah, does it like exponentially back off and
reminding you. I think by the way, that snoozing or moving things out of
you is really important. It’s actually a big difference in just having
a big pile of to dos because there’s a limit to how many things you can
have in your head at one time. And often we have new ideas that we want
to bring in, but there’s no space. And the only way to do that is to
actually kick stuff out from your working memory, and something like a
snooze can help with that.
00:34:21 - Speaker 1: Muse is really Interesting in this regard because
the the constraint of the screen as a surface, it encourages users to
keep stuff to the quantity which they can see at a reasonable zoom scale
on a screen at a particular time. I like part of the design?
00:34:36 - Speaker 2: Yeah, well, certainly constraints are potentially
great for creativity. Post-it notes.
One that I reliably come back to both in my own work, but also just as
just this kind of very workhorse tool for thought analog world thing and
part of it is you just can only fit so much you can also use index cards
for this as well, yeah, maybe with an index card and a Sharpie and that
sort of limited amount that can be on each card.
Of course you can have any number of cards.
So yeah, obviously with Muse, you’ve got the, you’ve got the expanding
boards and you’ve got the sort of the 3D nesting, but certainly
there’s I feel a desire to make what’s on the screen at the time kind
of fit together as a collection of things that feed each other and when
I start to have a section. Of the board that starts to feel like a
rabbit trail, then I want to make a subboard that and so it feels like
you’re going deeper down the rabbit hole or something like that.
00:35:30 - Speaker 1: One of the things I wanted to ask you about is
kind of muse relates to this note writing practice I’ve been doing is
the practices of refactoring or revision, polishing, gardening.
Uh, something that’s been very useful in my practice is kind of having
ways to think about writing at different levels of fidelity.
So I’ll kind of have a place where daily notes go that are quite
fleeting and kind of scraps will start there. And when something is
titalable, there’s some, some atomic unit that I can point to and say
like, OK, that’s that’s the thing. Now it can get its own notes and it
can be linked to from places. But almost, it’s almost like the goal
over time is for these things to adhere and Crete into larger elements.
So a a note that’s a single claim is like not that useful. It’s kind
of this ross, but eventually some number of notes that make a claim will
become like a, like a theory or like a noun phrase, a coinage or
something. And that larger note that, you know, contains references to
all these constituents, it feels like an increment that’s meaningful.
And so the pressure in the system to like over time refine, refract. To
create ever higher order abstractions is very helpful in my writing
practice and I’m curious how you think about that.
00:36:38 - Speaker 3: I would say that Muse supports that, but doesn’t
require it. So you can certainly use Muse as a persistent corpus that
you’re accumulating over time and building up to these pristine and
complete notes that are basically publishable.
But you can also use it in complement with other tools. So maybe you’re
doing it in your head, maybe you’re writing stuff out in notion, maybe
you’re using an authoring tool like Final Cut Pro, it’s more flexible
It’s very important. It was a very explicit design decision that boards
and cards in general do not require titles. I think that one of the kind
of original sins of of file systems is in order something to exist, it
has to have a name, but a lot of things just aren’t named yet.
00:37:11 - Speaker 2: That was one of our design goals with Hiroku was
that you’d be able to put an application online without giving it a
00:37:17 - Speaker 1: Oh, that’s great. I didn’t know that.
00:37:19 - Speaker 2: That’s wonderful. The original implementation was
Apps by default were untitled some long.
00:37:25 - Speaker 1: They have cute names.
00:37:26 - Speaker 2: I recall. This was, I think one of the, one of the
really lovely pieces of work my partner there, James Lindenbob did,
which is what we now call haiku names, which I think have been fairly
widely adopted, which is sort of taking an adjective and a noun that
were carefully selected so that they go together and they convey a
certain vibe that kind of connected to our brand or whatever, plus we
eventually had to add some numbers on the end just because there was
enough of them. Um, but the idea is something that looks nice. It
doesn’t look unfinished, it doesn’t look like untitled, but it
doesn’t also require you to figure out, wait, do I want to call this my
wiki or is it the team wiki or is it Team Wiki 2 or is it the, cause
it’s like an idea I wanna pursue an unfinished thing and I don’t quite
know what it’s gonna be yet. I have this hunch that I’m exploring and
then yeah, you get all hung up on the name, um, and yeah, for for sure I
see the file system. Uh, world of things having kind of that same
problem where you use his names are important when we know that we sense
that and so if you have to give it a name to even get started on
whatever it is you’re creating, that can be a bit of a, a bit of a hold
up. Now it’s nice, it might be nice to title something or label it
later. Muse has labels for that reason, obviously rename your Hiroku
app. There’s lots of other examples of that, but being able to just
start with, it doesn’t have a name and eventually actually the act of
naming it is you’re sort of upgrading it from Random tidbit of of
random idea, random tidbit of knowledge may not amount to anything to,
00:38:47 - Speaker 1: Yeah, I really like this word upgrade. It accesses
a design direction or a design space that I’m curious about with this
taxonomy of notes, taxonomy of creative work. Taxonomy is too too rigid
a word. It’s obviously much more fluid than that. Almost the ceremony
of giving something a name, giving some. A coinage, and that that feels
that the object feels more complete when it has a name, almost like it
wants to like it wants to have a name. It’s OK with not having a name,
but it’s in a happier state when it has a name.
00:39:19 - Speaker 3: This is a feeling that resonates very strongly
with me. When I’m doing a project, a huge milestone is when I come up
with a good name. And I don’t know why it is just, it feels so much
more. Real when that happens.
00:39:29 - Speaker 1: In designing tools for thought in general, I think
this is a powerful practice to avoid the tyranny of formality by saying
like, OK, there are 6 types of notes. There’s the fleeting note,
there’s the claim note, like, that’s terrible, screw that.
But you can still have an opinion about process.
People ask me like, what software do you use for your note taking? and
it’s like totally the wrong question.
What matters is kind of the methodology, but having the methodology and
Mind, I can’t readily like communicate it or install it into others'
minds except by having them read like thousands of words of notes.
And one of the things that Tools for Thought can do is to encourage a
particular methodology, not by imposing formal structure, but by
implying certain kinds of structure, by making, for instance, objects on
a canvas feel somehow more complete when they have a title. You’re not
imposing the necessity of a title, but you, you’re suggesting that
one’s work should perhaps culminate in a title.
00:40:24 - Speaker 2: My creative process is always heavily oriented
around finding patterns, which is why it’s important for me to have a
lot of I guess raw material and input.
Uh, you can call it data, but it might be something like user interviews
or it might be something like looking at some other products in the
space that I want to compete with or improve upon or something like
Um, it might be a series of bug reports, and I’m trying to get to the
root of what this is in some kind of complex system in order to do that.
I want to, you know, it’s been very difficult to track down, but if I
could somehow kind of look at all of it together and extract out what’s
the, what’s the pattern here? That’s, that’s the place where insights
come from me. I, I glean that’s not necessarily the case for everyone,
but for me it is this process and if I can somehow get everything
together, I can get all the relevant stuff in one place, that’s half
00:41:12 - Speaker 3: Uh, one last idea and tools for thought before we
transition into the meta, and the, the mummonic medium can be thought of
as a way to optimally position you to remember things.
There’s this point where if you’re at as a learner, you’re, you’re
best position to recall vocabulary phrases. It’s like just as you’re
about to forget, basically, you get prompted again and as that happens
more and more, those times become longer and with a system like space
repetition, you get this software-based support to help you remember
I’m curious if you think that technique can be applied to Skills. Uh,
this is an idea that I’m really intrigued by because yes there’s a lot
of interesting things that are like facts and figures, but there’s also
a lot of things that are our skills and abilities, and I wonder if we
could apply the same technique to learning how to play chess or how to
use a video editing program or something like that.
00:41:57 - Speaker 1: I do think that’s possible. I’ve spent a few
years experimenting with it now, and so is my colleague Michael, and it
begins with this observation that it’s possible to use spaced
repetition memory systems for more than just recall. So the the typical
way to use them is like, OK, what’s this term? What’s this definition?
And that’s cool. I mean, that’s useful. But you can also use them for,
for instance, applying an idea. And in fact, in quantum country in the
final chapter, we have these questions that look a little bit more like
lightweight exercises from a textbook or something like that, that share
the property of the recall prompts that you can kind of, you can do them
in your head, they’re quite rapid. They’re semi fungible, they’re
lightweight, but they’re things like what would the output of this
circuit be? And these are different from the recall prompts and that
they’re not the same. Every time you see them. So you’re actively not
trying to remember the answer, but you’re trying to like go through the
work of producing the answer.
You can also write conceptual prompts, concepts distinguish themselves
from declarative knowledge by focusing on how things relate to each
other and kind of systems and structures.
You can ask questions like for instance, when I was studying the history
of philosophy, contrast positivism and existentialism.
Now we’re making a connection, but in terms of developing a skill, like
maybe you want to like learn to think in a Danological fashion or
something. So you can also write a prompt that says, take a decision
that you made this morning and it could be as simple as like deciding
not to exercise when you normally would have and justify it or condemn
it from a dentological perspective. And so this is like a task.
So zooming out, I think space repetition becomes most powerful when we
think about the items, not as flashcards, but as micro tasks and what
the system is doing is batching. The transaction costs, which would
normally be associated with orchestrating all of these tiny micro tasks
that you could use to practice a skill or develop a worldview or
self-author in some way, and putting them together so you can say like
I’m going to do 10 minutes of like my self betterment session very
broadly construed, and that’s going to involve remembering certain
chess moves and also practicing this line of force motion in chess and
also reflecting on logical positivism in a certain way. Uh, and so on
00:44:06 - Speaker 3: Yeah, that’s really interesting, and I’m, I’m
wondering if you can extend it even further. So I think one element of
space repetition is it’s kind of helping you with the mechanics of, OK,
you commit to spending 10 minutes a day on this problem and we’re going
to use the software system to make that really productive.
You’re gonna see a lot of cards, for example.
But I think another element is basically identifying what you need to
get better at. In the case of memory, it’s pretty straightforward.
It’s like the, the question. that you answered incorrectly last time or
something like that, are the ones that you need to see now.
But in the case of chess, for example, it might be that your endgame is
weak, or you don’t know how to handle attacking knights or something,
and that is potentially much harder to identify programmatically.
But it seems like it’s also within reach. And so I’m curious about
systems that both um help you mechanically, but also in kind of the same
system, identify your weaknesses and where you can improve.
00:44:48 - Speaker 1: There’s a lengthy history. of people trying to
solve that particular problem, going back, I think now almost 5
For me, the most promising kind of subfield or sub approach is called
intelligent tutoring systems.
There are a few systems in the wild that have been commercially
The most notable is called Alex ALEKS. It’s an algebra tutor which has
some fairly clever mechanics for identifying your weak points and then
focusing practice time on on those.
I would say that none of these systems has been wildly successful and
the field as a whole has not been wildly successful.
I don’t fully understand why.
I’d like to spend some time studying that because it seems like a
somewhat obvious progression once you kind of get into the space
repetition space of trying to schedule stuff more efficiently, choose
construct cards more effectively, perhaps dynamically. I have read some
papers about people in the fields theories about why it hasn’t worked
very well. They center on things like the non-regularity of topics. So
an intelligent tutoring system on algebra will often share very little
in common in its implementation with an intelligent tutoring system on
geometry. They can share, you know, some kind of fundamental like
modeling, the learner primitive type stuff, but the representation of
the ontology is first off very difficult to construct and second off
very. difficult to like systematize and encode in a consistent way
across fields. My like personal hunch, and again, I haven’t read deeply
into this, but my hunch is that part of why these systems have not been
more effective in my practice is that they’re universally incredibly
dreary. They, they have this intense feeling of being in a skinner box,
like you’re a rat in a wheel, you are being fed. These like morsels of
problem, and you like swallow, and then, OK, true, like, here’s another
morsel, like, do this one next, and I think it may be possible to like,
to recuperate the underlying conceptual ideas without the the
interaction framework that they all employ.
00:46:39 - Speaker 3: Yeah, very interesting. I check out that
00:46:40 - Speaker 2: So if we come to the meta side Of how tools for
We all have some familiarity with the human-computer interaction
academic field and dabbled in that in various ways, even if none of us
Then Andy, you ran a corporate R&D lab, which is sort of a one
commercial approach to tackling innovation.
We, uh Mark and I were part of An independent research lab, which was an
experiment in that, uh, and then all of us in various ways have been
part of either classic Silicon Valley startups or bigger innovative
And despite all of these, I feel like we still don’t have the level of
attention, funding, and just people who are passionate about.
Yeah, computers and more broadly information tools that can help us be
smarter, more thoughtful, make better decisions, be self-actualized, all
of that bicycle for the mind stuff. I’m still trying to figure it out
why that is. What’s the, what’s the gap there?
00:47:41 - Speaker 1: This is an ongoing mystery and a topic for
discovery and discussion because in my mind, the wind condition for my
work is not creating a particular tool for thought that that’s really
powerful, but causing this to be a field. I view it as not a field right
now. It’s kind of like this proto fields like some people doing stuff.
We don’t have the Maxwell’s equations. We don’t have a powerful
practice, but it kind of wants to be a field. I would really like it to
00:48:04 - Speaker 2: And in order to get there, no one graduates from
design school and says I’m going to go into Tools for Though.
00:48:08 - Speaker 1: Well, I mean, some people have that intention, but
they mostly don’t, and they mostly can’t.
00:48:11 - Speaker 2: Yeah, can’t is a really good point. I we got a
lot of emails that can switch with people saying, hey, I’m about to
graduate from this design school or I’m working in a startup over here.
How can I get into To this field, I kind of said, well, what field? I
didn’t have, I didn’t have anything like an answer for them.
00:48:27 - Speaker 1: I don’t think there is a good answer.
Almost everybody who’s been successful, it’s difficult actually to say
that anybody’s been terribly successful recently in this space, but
anybody who’s had even moderate success has something weird going on.
They’re like independently wealthy or they have some cash cow that
they’re like milking in order to let them do this essentially
economically unproductive activity, or they have like a whole bunch of
connections that they’re using.
I have been helped in my thinking on this recently by reading uh Nadia
Eggbal’s new book Making in Public, which analyzes the economics of
open source production, and there are some connections between the the
challenges of trying to provision tools for thought, work and also the
challenges of trying to provision work on. Open source. They both seem
from an outside view to be kind of economically unproductive
Nadia’s insight that really helped me and that seems to have some
analogs and tools for thought is that it makes sense to separate the way
that we think about the economic model of consumption of open source
from the economic model of production of open source. So when one
consumes open source software, that is a non-excludable resource, so the
code is just, you know, it’s available online, you can’t readily
charge tolls for it. Uh, it’s also non-rival risk. So you downloading
the code doesn’t really like make it more costly for me downloading the
code. There’s very near zero marginal costs.
The analog and tools for thought is once I like publish that paper. On
the great idea I had in Muse. This is a non-excludable resource out
there, and it’s also mostly non-rivals, you know, the 100th person
consuming that paper and consuming those ideas. It doesn’t really cost
any different from the 100th person.
But the production looks pretty different. It’s a it’s a small country
of people. It’s perhaps excludable, and there are some rivals elements
in open source, for instance, Nadia characterizes it as being about
attention. The scarce resource for the open source maintainer is their
attention, they’re being bombarded by these like requests and like
well-meaning people trying to contribute code and so on and so forth and
it’s very draining and this actually makes the resource rival risk
because the 1,000th contributor to the repository doesn’t cost 0
additionally relative to the 100th contributor. And so one way to think
about this that she suggests for open source that I think applies a bit
for tools for thought and relates sort of the strategy that I’m
pursuing now is we should think about funding production. Than funding
Normally with media goods, we think about funding consumption. Like you
go to the store and you buy the shrink wrapped package of software, and
see like you’re buying a good, you’re buying an artifact. And when we
think about commercializing or monetizing software, likewise, we think
about the good or the artifact, or perhaps the services associated with
it in the modern world, like I’m going to sell support services if I’m
red hat or something, modern models might sell cloud services, but a
different way to think about all this is to think about kind of verb
instead of noun, funding the process of production rather than funding
the The output of the production. This is more common in the arts,
somewhat more familiar in the arts. Like if there’s a musician you
really like, your contribution to buying their albums or whatever, like
it’s probably not earning them very much money, but increasingly it’s
a popular thing to like be part of their their fan club or sponsor them
or something like this. And when you do that, when you sponsor the
musician, it’s not really that you’re like buying a particular song or
like buying an output, whatever. It’s more like, I like what you’re
doing and I I I want you to keep doing it. I recognize that you need
resources to keep doing what you’re doing. And I want you to have those
resources. So like here I am funding your process.
And that’s roughly a model that I’m exploring for tools for Though
presently, wherein I’m soliciting funders to cover the production of
what are typically public goods. So I’m going to sit here and like do
this work and think about space repetition systems and the most
prominent, the most useful long term results of that is going to be an
essay, or even if it’s instantiated in software, and even if that
software is proprietary, it’s going to be a set of ideas, interface
ideas, which are instantly stealable. And so those are public goods, and
it’s probably a lost cause to try to monetize either the essay or the
like interface ideas in the software, yeah, file a patent on it, but
like that’s not gonna work. And so instead, maybe we can think about
supporting this stuff in terms of uh recently I’ve been phrasing it as
like funding a grant, like an ongoing grant akin to the way that you
would for an academic research lab, which also produces public goods.
00:52:33 - Speaker 2: And it it sounds to me like you’re describing
somewhat of a patronage model and you talked about this on a past
podcast. in what’s happened with indie games, Steam Early Access and
Kickstarter being the two channels there, um, and that that’s maybe a
good example in a lot of ways, even though games are so different.
It’s the upfront production is where the cost is. You do have to do it
upfront. There’s several years of development by these, by, you know,
whatever size team there is. And when people invest in that, yeah,
they’re getting some things like access to a community and ability to
influence the game and ability to play an early buggy one that probably
isn’t very fun. And maybe that feels good for the person or it’s fun,
but ultimately it’s more about wanting to support something they want
to see exist in the world.
And I see a similar thing happening with the boom and subscription
newsletters. We’ll see, you know, whether that’s a bubble that will
pop or something sustainable. I I hope closer to the latter, but I think
it’s a similar thing, which is that people think this is someone and
and probably that personal connection is part of it. When you get a
subscription to I don’t know what the New York Times, there’s a maybe
a similar thing there you’re saying, I want to fund good journalism.
There’s something more powerful, I think about that individual creator,
whether it’s the musician, whether it’s the indie game creator, uh,
whether it’s the newsletter author where you you feel like you sort of
know them as a person and what their work is and you’re really funding
them because you believe in them and their worldview resonates and
you’re sort of saying, I want, I want more of this in the world.
00:54:00 - Speaker 1: This leads I think, to a significant challenge.
It’s comparatively difficult or it seems comparatively difficult to
fund teams with this model.
Like a lot of the advantage does seem to be from this personal
connection, you know, if you like go to my Patreon page, it’s it’s
like it’s personal in a lot of ways. Like I’m writing like, here’s
what I’m thinking about. Here’s what I’m anxious about. And you’re
also perhaps there because of my presence in other places like you heard
me on a podcast or you, you saw me on Twitter or whatever.
If now this is like the team for the something game, it’s more diffuse.
And then there’s also simply A matter of funding amounts.
So it seems at this point pretty likely that Patreon is going to be able
to raise an amount of money that can basically support me, which is
exciting and kind of surprising to me, but very nice.
Assuming that persists, I can continue producing public goods of this
kind, but it seems unlikely that it could support a team.
I really don’t see that happening. So I, I don’t quite know what that
model is. And one of the things I’ve been thinking about is that if the
main useful long term output of this kind of tools for thought research
is not The specific software that is created, like we don’t use Ivan
Sutherland’s sketchpad anymore, but rather the insights, then maybe
it’s actually OK for some or all of software components of these
elements to actually be proprietary.
If you’re my patron, maybe what that means is that you’re funding my
work, you’re funding my research. So that’s going to include essays,
which are, you know, freely available and perhaps software which is used
to produce the insights, those core insights that are captured in those
essays. And if you’re a patron like that software is also freely
But otherwise, The software is perhaps proprietary and perhaps generates
revenue, which can then support a team.
One of the other problems I have here is, is that I can’t do all the
engineering work myself and also do great research. I kind of need long
00:55:44 - Speaker 3: Yeah, I do think the patronage model is really
promising a few comments there.
One is I tend to agree that it’s harder to find large teams with the
model. I do suspect that small teams are actually possible.
Another thought is, I think some of the most interesting work in this
area leans a lot into community.
So again, to draw a somewhat simple example from the gaming world, often
if you support these gaming creators at different levels, you get access
to like correspondingly elite Discord channels, which seems like it’s a
small thing, but it’s actually a huge human needs, like be a part of
community and and to believe in something that is important to you and
to participate with your peers. So there’s actually a lot of um kind of
community goods that one can provide as an independent creator or as a
Another interesting example there is Pladium Magazine, which is doing
really interesting work on political economy. And they have different
tiers for supporters and as you become a more substantial supporter, you
can participate in things like salons or even interact directly with the
Another idea to address the funding a team problem is, I think people
don’t like to put money into big mushy pots.
Yeah. Like you think about donating to some huge institution, you like,
what’s it going to go to? Is it like going to go to some, I don’t
know, like random building or like cutting the lawn? I don’t know,
it’s not very exciting. Whereas with an individual creator, like I’m
funding, you know, this work on the neonic medium, that’s awesome. And
I wonder if you can get a little bit of both by having an institution,
but also supporting more targeted funding. So it’s almost like you’re
having a two-side marketplace for funding as an institution where these
are the 5 projects that we want to potentially do research on and you
can back individual projects, and once it’s reached a critical
threshold, we’ll go ahead and do it. So if it feels like you have more
agency over what your money is supporting.
One other example there is you mentioned the work potentially being
proprietary. This is actually well precedented with keyboards of all
things, folks, you got to look up this, this. Crazy world of custom
00:57:27 - Speaker 2: Oh, it’s so you’re talking about the mechanical
00:57:29 - Speaker 3: Yeah, they’re usually mechanical and they do
things like, you know, someone says, I’m going to make a keyboard.
It’s going to cost you, I don’t know, $500. And if I get 200 orders
for them, that’s enough money for me to do a production run in China.
So I’ll do it. I mean, people pay 50 $500 for a keyboard, maybe
they’ll pay $50 for, you know, a better note taking app or something,
right? I think it’s it’s very possible.
00:57:45 - Speaker 1: And and do those people also have patronage or or
is is now just the product they’re selling?
00:57:50 - Speaker 3: Yeah, I think it’s kind of both. So you’re. Yes,
you’re covering the production costs, but also there’s this huge
creative and entrepreneurial element where you have to ask to pull
together the keyboard, like find the right key caps and get the right
producer in China and arrange it all right. And so you’re also paying
for that. It’s kind of an entrepreneurial activity in a way.
00:58:04 - Speaker 1: And do they like open source the like CAD files
and stuff? Is there like a public goods component at all in that world?
00:58:10 - Speaker 3: Um, that’s a good question. I’m not sure.
00:58:12 - Speaker 2: Yeah, I’ll link both the mechanical keyboard and
subreddit, which is just fun to scroll through for the great photos.
Uh, but also kind of relative to the what people are willing to spend
There’s an article by Kevin Lainoff, if, if I’m not mistaken, that is
basically an exercise in what could you price this out and, and they
actually end up with a price that’s over 1000 or something like that.
And again, it relates to people who are really, really into a very
specific hobby. They like the fact that it’s this one time run. It
feels very authentic. It’s just someone in their community, you know,
it’s it’s not a ongoing commercial entity. It’s just a person in the
community that has an idea for a unique thing to make that they want to
share with everyone else. To your point, they’re willing to put down a
lot of money to do that. And yeah, I think there’s a, it’s a very
different kind of calculation when you think I’m supporting something I
believe in. With the community I want to be a part of is a different,
very different kind of transaction than I’m purchasing a product, I’m
going to, you know, shop on whatever comparison sites, to get the lowest
price I can. I’ll never ever meet or even have any idea who was. Behind
making this product in the first place is very transactional,
mechanical, just give me the cheapest, simplest thing that will solve my
problems so I can move on with my life.
00:59:28 - Speaker 1: Another related problem seems to be the arrivals
to working on this kind of work have gotten more appealing.
And this is kind of a different angle on your point about Instagram.
When you look at PARC, it’s not so much that people there got paid a
huge amount, actually, that the total budget for the projects that
produce personal computing was not that large, but relative to the
rivals, uh, relative to the universities that essentially would have
been the employers for that staff, PARC was offering more than anybody.
And so they were able to assemble basically all the really great
computer scientists that existed uh in that period. It’s a bit of
Overstatement, but they got a huge portion of them. Whereas now this
work is competing with, you know, fairly lucrative jobs in the tech
industry and in more than one way. So like, yes, it’s true if you’re
young and you, you know, go work for Instagram, you’ll maybe make a
quarter million dollars a year or more, but also in this kind of
uncapped upside way. So if you are the kind of person who’s
entrepreneurial and agenttic enough to pursue this kind of original
technological work, you could probably be working on a startup and you
could be getting uncapped upside. Whereas it seems fairly difficult to
uh pursue a course of action that could yield uncapped upside in the
tools for thought space nominally speaking. Uh, because of the kind of
public goods elements, like the, the hardest thing that you do will be
to come up with the elements that is novel, unique, and immediately
stealable. And like, yes, you can start a startup around the, you know,
the, the kind of the software around that, but uh you feel like you’ve
shot yourself in the foot a little bit.
01:00:56 - Speaker 3: I certainly think there’s a lot of truth to that.
I just want to jump back to the absolute amounts and comment that I
think that the amount of money you need to find really interesting
projects in this space is in the scheme of things very.
Small. It’s gonna be a fair amount to any individual person or any
normal individual person, but just the absolute amount in terms of what
we spend on random funded startups or what our various levels of
government spend is just quite small. That, that makes me optimistic
that there’s a way to make this work.
Um, on the opportunity cost thing, I think that cuts both ways. Like,
yes, it’s the case that people can go to the Googles and the Facebooks
and earn a lot of money, but we’ve also seen with the lab that people
really value doing this rewarding, interesting, unique work and it’s
accessible to a broader set of people.
So like, is remote because we have a broader hiring funnel and so on. I
also think there’s a time and a dynamic element here where you don’t
need to spend your whole career doing research.
Actually, one of the ideas behind the lab, use the spin out and kind of
the whole group there is that we expect people to rotate over time. So
it’s not just you’re not just a career researcher or a queer
entrepreneur, you actually get a lot of dynamism from going back and
forth. You get different benefits in each world and then actually going
all the way back to our conversation about full-time toolmakers versus
practitioners. you help solve that problem. You spend maybe 4 or 8 years
in one domain and then you switch over and you get that hybridization.
01:02:10 - Speaker 2: I’ll fill in that I think the funding it through
commercial products. We said the paying for the for the result rather
01:02:19 - Speaker 1: Sure, yeah, yeah, kind of paying for the output of
the artifact rather than paying for the production of it. Yeah.
01:02:24 - Speaker 2: Obviously we’re pursuing the paying for the
artifact path with umm that we’re we’re selling this product
commercially if you want, kind of, but like it’s subsequent.
01:02:31 - Speaker 1: To the paying for the production element of I and
01:02:36 - Speaker 2: That’s right, yeah. Um, well, I almost wonder if
there’s a progression there a little bit, which is ink and Switch was
very much just a small amount of money, grant money that was people that
want, you know, some people that wanted to see this thing, see a certain
kind of research done in the world and very much public goods, we
published everything as we open sourced as much as we could. That was
And then the spin out commercial entity now we’re in a state where
probably we’re close. to that kind of patronage Kickstarter level,
which is, you know, I think a lot of the people that purchase the
product, now they’re thinking it’s less about does the exact feature
set that exists today, you know, how does is that exactly what I want or
is that that, you know, worth the price and more that they’re thinking,
I believe this team over the next period of time that my subscription
covers is going to make great things and it’s going to make this
product even better into something that fits into my workflow into my
life, enhances things for me, enhances things for others.
And you can imagine fast forwarding a few years when the product is much
more complete, uh, that at that point maybe it does come more
No one cares about funding the the team or the long term thing, it’s
just more about now it’s a good product, it’s very full featured and
has been developed over a long period. Time and so they’re going to
spend money that the price they pay is, is much more of a transaction to
just get this thing that does solve an immediate problem for them and
they’re not worried about the future or the team behind it or the
community element. So you can see that as sort of a three stage
progression. Uh, at least I hope or imagine that could happen here and I
could also imagine it happening with, with other things including
something like the pneumonic medium or other uh research work that I’ve
seen in process, but the making that transition step to step to step
that I think is another place where Mark and I talked about that in our
HCI episode that that I think is another weak point in the the field if
we can if we can call it that.
01:04:25 - Speaker 1: Totally. I think that. Progression is, is likely
to happen in my work. And one of the things that seems to create the
weak points and is likely to create them in my work is that it’s not
always the same people who want to be working on these, these different
things. That’s both a weakness and the strength. Maybe I don’t feel
like doing the production maintenance of a commercial piece of software
like that that’s just not what gives me joy to today.
01:04:55 - Speaker 2: But there’s like a lot of people who really like
just like churning through task lists, and they love like the feeling of
like, check, check, check, check, and Those people will be like really
well suited to be on the engineering team for, you know, long term I’ll
note you even reveal your proclivities by saying turning through
taskless, because some someone that has more of the mindset of wanting
to keep a real existing thing running and serving people’s needs, they
would say, I don’t want to just think big floaty thoughts about
something that could exist in the future. I want to deliver real value
today by building production software and shipping it to customers,
01:05:17 - Speaker 1: For sure. And and so if you’re someone who values
the big floaty thoughts, and you want this, this big floaty thought that
you’ve kind of tethered to earth to actually live long term, you got to
find somebody who enjoys the other stuff to come and pick it up and take
control. That seems like a weak point in the process. Reminds me of tech
transfer in universities, tech transfer.
01:05:36 - Speaker 2: I don’t know if I know that concept.
01:05:38 - Speaker 1: It’s how many of the top tier research
universities actually get most of their funding these days. Uh, my alma
mater, the plurality of its funding comes from this.
I think it’s true of Stanford too, but essentially, uh, the model is
that professors are paid for mostly by public grants, and IH NSF, things
like that in the sciences anyway, and they produce mostly public goods.
They publish papers and so on, but also sometimes they They file patents
on those things when they are patentable things or they do spin off
startups or their advisors to startups or something like that and the
Uh, and so a great deal of Stanford’s wealth, for instance, comes from
the patents which underlies gene tech for recombinant DNA and uh Google
01:06:18 - Speaker 2: Another piece of the funding spectrum. Is
corporate R&D labs, which Xerox Park famously was Bell Labs and the one
I often use as as inspiration. Now that was quite unusual in that it was
a corporate R&D lab for the largest monopoly business, I think that is
certainly ever been in information technology. But Andy, you had your,
your at least brief run at doing uh on the corporate R&D side. How do
you think that fits into this?
01:06:43 - Speaker 1: It’s really challenging. I spent a lot of time
studying the players in this space and mostly came away with a pretty
My own work at Khan Academy was kind of weird because Khan Academy is a
nonprofit. So that the motivations are somewhat different there.
But even just looking at the for-profit space, it’s difficult for me to
get excited about corporate research labs as an institutional model,
speaking to some of the people involved in setting up and tearing down
Microsoft research, that really was not terribly successful for the
company and indeed was like successful in these other ways of creating a
sink that could keep talent from Going and starting startups in Seattle,
for instance, is like actually a useful and positive effect that made it
worth funding for, you know, Bell Labs had it was this kind of this
chaff to dodge antitrust litigation seems to be the prominent reason it
got so much funding. It did actually generate a ton of value for the
And Park, you know, I mean, there’s this fumbling the future phrase
that goes around. The fun thing is that Park actually was profitable for
Xerox, just barely because of laser printers, not because of personal
Yeah, Apple’s corporate research uh was really not successful. Um,
I’ve been having difficulty learning as as much as I would like to
about Dolby. Uh, which does actually seem to be pretty successful. But
another fairly successful example is Pixar. And one thing that I think
really distinguishes Pixar’s corporate research is that there’s
cutting edge graphics research that goes on there, but it it is very
much in service of these creative films. They are huge money generators.
01:08:08 - Speaker 2: Well, maybe that creates the connection and is
always one of the challenges is the disconnect between the uh the mad
scientists off thinking the big thoughts and the real world problems
that those can be applied to and Yeah, having the the graphics
researchers need to turn around and produce an algorithm or even work
code for the new movie that’s coming out on a particular deadline and
there’s a lot of money at stake for maybe that creates some realness or
or as a way to, as you said, tether the the thought balloon to the earth
01:08:38 - Speaker 1: One of the challenges that seems to exist for all
these labs that Pixar manages to avoid the mechanism you just described
and also existed at Khan Academy, which is a nonprofit so it had, you
Interestingly different funding issues is just this, um this challenge
So even the lab comes up with an idea, we came up with the laser
printer, we came up with a personal computer, we have the Alto, we have
the star, you know, whatever. And like, we want to get it out in the
world and have it be a major corporate strategic priority. This is often
the point where things fall over because if if the research really is
cutting edge, often it will mean at the highest level shifting the
company’s strategic objectives to really Capitalize on that technology.
It’s difficult to find organizations that have done that consistently.
Pixar makes use of their research, but I think in general, capitalizing
on really great like water rendering technology or whatever doesn’t
require shifting the highest level corporate strategy.
01:09:25 - Speaker 3: Yeah, Ben Reinhardt has done some really
interesting research on this topic in the context of DARPA. I highly
recommend checking out his work and one of the insights from the world
Or dual use technology transfer is that it’s extremely dependent on
thick social networks that are formed largely because of DARPA’s work,
and this I think points to another gap or opportunity, which is the kind
of institutional and community side, where if you have a place for
people to congregate and to gather and to form social connections, it
can really fertilize the creative work of the industry.
And we saw this a little bit with ink and Switch, you know, we had a
very modest community effort.
It was Slack channel. We had some articles published, we would tweet
some things, but even just that got all kinds of amazing people to come
out of the woodwork and say, you know, I’m working on this too, or this
idea or what do you think about this or how can I contribute? And I can
only imagine if someone invested a lot more in something like that,
you’d see correspondingly more results.
01:10:17 - Speaker 1: I’d love to make that happen. It was briefly a
kind of a high goal for the year until I realized that I couldn’t
really achieve the other things that seemed important if I pursued that.
So one thing I observed is that these kind of community efforts do seem
to be like the result of times of plenty. DARPA, uh, especially in its
heady days with just like excessive funding, is able to devote resources
to this in a way that seems difficult.
01:10:39 - Speaker 3: But I think DARPA is also an interesting example
of how, again, the absolute amount needed is not that big. Like the
number of people at the very core of DARPA is quite small.
01:10:47 - Speaker 1: Yeah, maybe I’m just thinking too small here.
01:10:49 - Speaker 2: The time of plenty point, I think seems right,
which is what we’re talking about is investing in the future.
Our particular niche and interest is this, this tools for thought. Um,
but in general, being willing to invest in the longer term, 5 years out,
There’s a few things that drive that.
There’s military is just a huge one because that’s just always an
existential question for a nation. It’s also things like and of course,
the space race with a lot of the stuff that led to the internet and a
lot of those technologies was at its core connected to a sort of a
military dominance or perception of that uh between the world
Or you have something like Bell Labs, which, as you said, you know, this
antitrust thing, this, this huge monopoly with so much money to spend
and so sort of in their interest and and government funding generally,
the larger pool to draw from potentially and a willingness for longer
And corporate R&D labs are always tough because they’re always when
times get a little tough and there’s always the up and down, you’re
going to look a little shorter term, of course, the first thing to go is
the dreamers that are that are looking further out. And that’s fair
enough. I don’t, I don’t think that’s a very pragmatic and reasonable
choice, but then that that comes back to well, is that a is that a way
to fund our future and at least the evidence seems to be despite some a
few cases, a few exceptional cases like PARC, uh, that that’s not
really very. sustainable or repeatable.
01:12:14 - Speaker 1: It’s interesting to look at HHMI’s funding
practices versus the NIH’s funding practices of targeting specific
researchers and trying to give them consistent funding over longer
periods of time. What I feel differently about spending a lot of time
right now on like community organizing, if I had something like an
endowment. Um, I probably would, and it’s weird because like I’m not,
I’m not really bleeding. I am in the red, but it’s, it’s not so much
that I don’t want to do it because it feels like ultra ultra urgent to
resolve that. It’s more that there’s a feeling of not being on steady
01:12:49 - Speaker 3: I think the future is still, I’ve written here,
we’ve talked a lot about structural issues like funding and things like
that, but a huge element is just the individual will and passion to see
something change in the space. And I think Andy is a great example of
that. There’s all kinds of currents that make that kind of work very
challenging, but he’s succeeded because of his will and talent and
persistence. I would invite more people to just try to take that on.
01:13:14 - Speaker 1: It’s very kind. Thank you, Mark.
01:13:15 - Speaker 2: B like Andy, I like that. Well, if any of our
listeners out there have feedback, feel free to reach out to us at
@museapphq on Twitter or hello at museapp.com by email. We always like
to hear your reactions to anything we’ve talked about and we’d like to
hear your ideas for future episodes. Andy, thanks so much for joining us
today and talking about these areas of mutual interest. Thanks y’all.
01:13:37 - Speaker 1: This was a really fun conversation. Thanks, Andy.
