Waiting..
Auto Scroll
Sync
Top
Bottom
Select text to annotate, Click play in YouTube to begin
00:00:04
one of the things that annotation is most related to you know annotation is um you know something that we do you know in the margins uh
00:00:15
or in the context of something else and uh one of its applications that's most you know kind of useful for all of us is you know the more personal aspect of taking notes um with as as a
00:00:30
part of an overall kind of personal knowledge management strategy or practice so this year for annotate partly as a as a kind of a reaction to the kind of absolute explosion of innovation
00:00:44
in the note-taking area that's that's happened over the last uh three four five years um is to do a uh a whole kind of a dedicated track or dedicated session on
00:00:56
note-taking and kind of the intersection of note-taking and annotation and where it's all headed i am super flattered to have a kind of an
00:01:09
extraordinary panel with us before i introduce them let me just kind of give give the the basic why here um which is that uh you know note taking it is an intensely
00:01:21
personal and sticky thing sticking in the sense that you know once you start using a certain system or a certain practice um you're likely to stick with it for for a very long time
00:01:32
and so ch choosing becomes really important um and and also because of the way that these systems are designed and the choices that they make um in terms of how they encode and store information
00:01:46
um the design of of these apps is super important because it really shapes how we think so thinking about thinking and thinking about note-taking applications becomes um something that
00:02:00
it might make sense to do so that's the purpose of the session here today and i want to um go ahead and introduce um our speakers so ward
00:02:14
uh many here will know uh ward cunningham who's the founder of wikipedia along with jimmy wales and who really innovated much of that system or i think wrote most of the original code
00:02:28
um and invented a lot of things but primarily among them for us today um are a couple things one is um the notion of kind of creating pages into existence simply by naming them uh
00:02:42
and and simultaneously linking them uh it's maybe something he'll he'll chat about if he has a moment which is really a phenomenally important insight that many of the apps here have
00:02:56
adopted uh as a strategy and it's kind of a feature of a lot of the way that these kind of modern wikilinking apps work um and also
00:03:07
a concept of automatic backlinking which is another feature that that is kind of omnipresent in in this space um daniel joyen uh is the founder of read wise
00:03:21
which has quickly become the kind of universal glue um that links all the apps at what i call the edge to the apps that folks use to map different topics
00:03:33
together uh and um you know it's probably one of the most used apps we see here with hypothesis and in terms of how people are are using our what we build with other
00:03:44
folks eduardo is a an engineer at google who's created the agora which is an open source federated note-taking platform
00:03:56
which really implements a lot of interesting and cool things that have to do with how you might collaborate on building your personal knowledge management system with others
00:04:09
bastion is the lead developer and maintainer of org mode which money here will recognize as as the mode inside of emacs that really jump-started
00:04:21
a note-taking revolution in the early 2000s of kind of extremely efficient um note-taking and organization um
00:04:33
that has become a model for a lot of house the other apps that have followed have kind of worked um oliver is a kind of the the renegade uh counterculture open source um
00:04:47
and privacy guy here uh whose world brains memex project is beloved uh by many for how its approach um that taken in terms of how
00:04:59
you can store and archive and annotate the web as you travel around it and uh connor um uh conor has perhaps uh
00:05:13
revolutionized this space more recently more than kind of any other single uh service with the creation of rome which is used by tons of folks i'm super excited to hear
00:05:25
what he has to say about where they're headed um and junior who is one of the key team members at the lochsec project which is although they're a relative newcomer are
00:05:38
implementing and kind of ticking some of the key boxes that people want to see in these kind of apps um maybe more effectively than almost any other kind of of the relative
00:05:50
newcomers so i'm super thrilled to have all of them here today and um we're going to go ahead and jump start folks in this order so i will
00:06:04
go ahead and what we're going to do here is we're going to do a series of whitening talks you know kind of five to seven minutes um on on their kind of vision for where things are headed uh we may have
00:06:17
room for time for you know kind of a question or two after each one um but we'll want to move relatively quickly through all of them and then get to a panel discussion which will follow this in which should be more of a kind of a
00:06:30
freewheeling uh discussion with plenty of time for audience participation so without further ado let me kick it over to ward so i uh uh you know i get going and i
00:06:44
have trouble stopping so i i thought i'd kind of follow the outline dan scent and uh there's plenty of room in that to uh say useful things uh you know he has
00:06:57
what are the trends that i see and of course trends on the internet are not very positive right now there's a lot that has gone wrong uh
00:07:07
you know in the 90s we all thought that it was the savior and it turns out to be the destroyer in many ways and i think that really is the advertising model advertising being just
00:07:20
a structured form of lying really and you know uh i'd say don't trust anything you care about to somebody else with a business model
00:07:31
because their business model isn't to make you successful i also sometimes call it clear cutting of the internet you know and that seems to get it emotionally i think i tried to replace some of my familiar jargon with
00:07:45
metaphors and that's one i like uh you know the the other thing i like is that some of us maybe it's an old-timer thing they're seeking a quiet place to uh
00:07:59
to listen to the wind in the trees that subtle thing that you know inspires might take some practice uh curiosity to to hear that i use that to
00:08:13
talk about some development practices that are could be called listening to the wind and the trees and and responding you know on the internet i think it's it's it's making a space that is more like a
00:08:26
park than a shopping mall and uh i'm not i i think my colleagues here probably you know are park builders too my particular vision you know is is uh
00:08:42
i wanted to make something that was federated but still recognizably a wiki and wiki has never been federated usually there's one big database and everybody shares it so i just said well you know if we want
00:08:54
to have small sites that don't get ravaged maybe everybody brings their own site and this was at a time where you could say
00:09:06
node install wiki and have a wiki and and you know the internet has gotten more difficult than that and we could talk about why it's gotten so difficult but but i'll say the the idea was that
00:09:21
i call it propagate copying in other words you're expected to copy you're not successful if you're not being copied and i think there's a lot of places in the world where you know impact in the world is
00:09:34
when your ideas spread and they spread by other people having your ideas and maybe maybe after a while they forget you were your ideas but you know you know if you can't charge money for them then who cares if you know the success is spreading those ideas
00:09:47
and that's a foundation of creativity really is is understanding other ideas and then taking them and making them your own sometimes we'll call that the the chorus
00:10:01
of voices is a little different than github where you want to have a bunch of people contributing to a piece of code where there's one correct piece of code in the end and and here here we're thinking you know
00:10:14
there is this sort of fork capability and we use the word fork because it means you're taking it but we i'll talk about a chorus of voices
00:10:25
where where nobody has to be saying the same thing or at the same time but that it somehow comes together and makes something uh
00:10:37
probably improvisational more than anything but that that is that's what we seek and it turns out that a lot of it looks like note taking you know people mentioned that you know
00:10:50
about the time you have a thousand pages you're ready to start a conversation because something will come up and they say wait a second i got a page on that and then you pull that page out maybe share it and talk about how it's different in the
00:11:02
current context but that's that's a body of work and it might not be all pages you wrote it might be pages you found other people and just put a little bit of your own spice on them
00:11:14
the other thing that i think is emerged to be really important is is we we talk about kind of shaping the white space around ideas that
00:11:27
that this is you know like improvisational musicians talk about free space or sometimes white space or poets talk about having a place to create a place in their mind to
00:11:41
create and that that's a little bit different than remembering things it's really it's truly trying to move things over side so that you can think of something new and doing that
00:11:53
collaboratively is pretty amazing the the uh the other thing that it's been at the heart of what we've done
00:12:04
in the last 10 years is is mixed content on a page you know that we have a markup but we never made a universal markup in fact the most common markup is just simply plain text
00:12:18
and hyperlinks so the double square brackets is about what you get there but you know when you're just talking that's pretty nice but we can do uh lots of other kinds of formats and we kind of lean to the more complicated
00:12:31
things like map markers and and knowledge graphs and uh data data representational data in various forms and what we like to say is
00:12:45
if you really have something you want to get done instead of doing it somewhere else and talking about it in wiki maybe you can do it in a wiki and share your setups you know if you've got an electron
00:12:58
microscope hooked up you know me they're pretty ornery and if you take some great pictures with it maybe you set it up with a wiki page and share that wiki page with the colleagues across the world could be interesting how am i doing time
00:13:11
wise did i run out uh wrap it another minute another minute okay well i've got a minute and a half worth of stuff here uh one thing that that's come out of that idea is
00:13:23
first of all when you own your own server you can pretty much hook whatever you want to it uh we've started writing multiple servers different you know implementations of editors and these get
00:13:35
called the outpost the model is that you could have a place in antarctica and it's not very pleasant but you can do things there you can't do anywhere else and you don't cut off every connection with the real world
00:13:46
when you're down in antarctica you know one of the things that that that's that's done for us has really freed us to think uh uh of the future in a very positive way one thing i should mention about annotations
00:14:01
is that our pages are a little problematic in that they don't have a specific home they float around the internet they have a history but they don't have a home what we've done with annotations
00:14:13
recently is add descriptions of intention you know this is this this is a little line that says when you're here you can push this thing
00:14:25
and this will happen and that intention is independently evaluated by what i might have called the testing system a few days ago now i call it the
00:14:37
robot critic and that robot critic is an outpost that might be reading what you write maybe there's several of them and it's commenting back in the federation about whether what you say happens happens or not
00:14:51
and this allows us to grow the system you know in another world you'd call them unit tests or function integrated functional tests is what i call them before but now i think they're annotated in tension and the
00:15:04
conversation around these annotations is just a conversation with you and another robot that might be a little more like the check engine light in your car than anything
00:15:17
uh artificially intelligent but still very reliable and that's our path forward so thanks for the time and i hope i conveyed the feeling of what we're doing as much as how we do
00:15:30
it lord thank you so much um daniel you're up nice to meet everyone my name is daniel duan and i'm one of the two founders of a small tool called
00:15:42
read wise it's still pretty small and most people haven't heard of it so i'll just give a quick overview of what it does just for context but we started working on this back in 2017 and we were really focused on the
00:15:55
mission of using software to improve the practice of reading which is a pretty grand mission and to make that a little bit more immediately actionable we noticed that
00:16:09
all these readers like us who used kindle or who used real later services like insta paper or pocket most of these people had accumulated this very high quality data
00:16:22
in the form of highlights and notes and annotations while they were reading but very few people like maybe less than five percent of the people who were reading in this way actually had a way to go back
00:16:35
and revisit those highlights and make use of them so that's really the origin of read-wise is we developed a somewhat novel workflow to go and get your highlight data which
00:16:47
isn't so easy to get with most of these services and then once we had it we would just make make it easy for you to consistently review uh those highlights and and we do that many different ways from
00:16:58
a daily email which is kind of the uh easiest the most consistent kind of like lowest engagement all the way through a web app and mobile app and people immediately enjoyed this because like i said they had already
00:17:12
curated this really high signal high quality data for themselves while reading and going back and and revisiting that content is is pleasurable and it
00:17:24
may create the opportunity to take action or it may spark a connection or you may use it as a writing prompt poole immediately resonated with that so from there in order to expand that value prop we
00:17:37
started with amazon kindle kindle obviously has a huge monopoly share of the ebook space you know something over 90 um so most uh book highlights were
00:17:49
coming out of kindle but we started to support uh the so-called long tail of reading apps so we'd add apple books which makes it really hard to get them out google play books
00:18:00
insta paper pocket pdfs medium and a cadre of other sources so then our value prop kind of evolved into being a place that makes it easy to get
00:18:14
all your reading data and annotations into a single place and once we did that then probably most interesting for the folks here because uh most people who you know for example know what rome is
00:18:26
um are like the the very elite power users you know they're they're not really interested in our um kind of simple easy daily habit but would rather get those highlights that we've liberated and bring them into a place like rome
00:18:39
where they can do even more powerful stuff so we've also built a series of export integrations to complement the import uh integrations and we've become kind of like a
00:18:51
very very tiny zapier of like reading data um so that's that's kind of the background on read-wise and what it is um you know what what we're focused on long-term is
00:19:06
you know we have this belief that when the personal computer originally came out it immediately was better for writing than anything that existed before analog it was better than typewriters it was
00:19:18
better than writing by hand so everyone's heard the phrase software is in the world and software immediately ate the practice of writing you know which obviously includes note taking what software didn't immediately
00:19:31
eat was reading because reading on a desktop computer you know on one of those you know small low resolution crt monitors like that was not a better experience than just reading an ordinary
00:19:42
book or reading a printed out research paper or something okay but technology is finally caught up and we now have the devices for example e-ink tablets you know normal tablets
00:19:55
smartphones where you can now consume content much better on a computing device so it's just a matter of time until a new category of software is created which is
00:20:08
essentially the analog of the word processor but for reading and that's very much what we're focused on you know helping to innovate there's obviously going to be a whole ecosystem
00:20:20
around that space but but that's the place that we're interested in playing um you know helping to create software that doesn't cater to the mass market you know people who are reading fiction
00:20:31
or romance novels but instead people who are reading with a purpose creating that that productivity elite software thanks daniel oh thank you very much
00:20:46
dan and uh it's a pleasure being here and i don't know in this uh this company uh so i'll try to be short which is not one of my strengths so uh please excuse any any discomfort
00:20:58
and i'm starting a timer uh yes so an hour this is the project that uh brings me here and i will try to um like uh just like this and hopefully um
00:21:12
like uh see if he can useful dollars which is our intention so first of all what is an aorta so we're going to be using different definitions throughout
00:21:25
this short deck or at different layers of abstraction or levels of abstraction i guess but um in its uh first definition i will say that analyze a crowdsourced distributed knowledge graph in particular
00:21:39
as it is currently as you will see it is a knowledge graph that is assembled or you will say integrated out of volunteer data entity mapped is all we request which means
00:21:52
each resource is explicitly about something right um and in particular of course we have started with uh as as many of you may know uh with a corpus of notes just uh nodes that people are taking in
00:22:06
diverse tools and uh we are trying to extend that to uh annotations um and it's also a social network in a sense uh perhaps yeah because um the knowledge
00:22:20
graph that we're building uh or that we can build uh both um it's pretty i mean contains knowledge producing a social context like all knowledge and in particular uh
00:22:32
very often this knowledge is about people so uh you could say that the knowledge chart contains a social graph and that's probably true of many tools in this space right and uh more generally perhaps an hour is
00:22:44
a public space this is our inspiration right um and here i would just say for now that it is not a market but a commons right so a bit of all the other vision here
00:22:57
which is uh when i say our uh i believe i say uh you know uh belonging to us all um and so uh you know perhaps this will be a bit um
00:23:09
obvious to someone else for a few goals um the vision is that the social knowledge base the space we're in needs an integration layer or an interlay uh that can be managed as a commons by
00:23:22
communities um so we call it the aura for short the name isn't important i actually believe our implementation very humble uh may not be important either but i believe the concept
00:23:35
is useful so here goes um i think we are all here in a in a nexus to some extent around the continuity and we face a unique opportunity
00:23:47
as communities because uh we are part of the community sometimes and uh it is the right time to step up as a group and invest in interop resist vendor locking
00:23:59
and uh try to agree on solutions that uh essentially keep us away from centralized systems right that we know have limitations
00:24:12
and to minimize this uh i we believe the ones working on the hour for now that we want to keep the cost of inter low as low as we can so from these all uh a few design principles for the
00:24:26
hour uh stem uh and that will be at the hour uh should be free and open source and it is and they actually require a little of iterators and try to give back plenty
00:24:37
and the aggressor should try to make use of existing conventions which is an extent it can be derived from the from the second point okay so this is um i guess relatively nice let me actually uh
00:24:49
okay i hope you can see this um so this is really really nice but going to towards my concrete how does actually work so we have our reference aora and it is built around a few just a few common building
00:25:03
blocks so we have uh nodes repositories as the initial copper we have the we killing as the one iteration primitive we are depending on and we have a set of
00:25:15
optional conventions that we can add to this and using this we assemble uh essentially these nodes annotations in alpha and also social media activity like which is being tested
00:25:28
into this social wiki like construct so you can see an example here on the right um the idea here is that all resources that map to the same weak link so the same entity because in the hour
00:25:41
the weakening describes an entity they are all uh attached to the same node in this knowledge graph uh and the algorithm the overlay that we have that is
00:25:53
shown to the right presents them sequentially essentially as a sort of like sequential wiki and then try to integrate them in the context so essentially show connectivity like at the bottom and also in the aorta as a
00:26:06
group um so we are also like trying to develop some uh some features that uh perhaps um can remind you of like precisely a social network we're very inspired by the
00:26:18
uh internet around the year 2000 i'd like to say and wikis are a big part of this also a journaling movement and everything to blogging etc and of course also like the contemporary
00:26:30
projects that we have heard about here and um we are trying essentially to build uh our commons let me just keep skip ahead because i'm a bit short on time
00:26:42
just for the basics which is like how did you how to join into the hour okay um the um it is a three-step process right the dldr is you can take notes with any tool your
00:26:54
preferred preferred tool and then optionally submit your notes to the hour the algorithm has to be completely detached from any of these implementation specific details for this we currently have like a
00:27:06
default setup which means um essentially markdown nodes with wiki links as an extension the only extension we require actually do not require but make use make use of
00:27:18
and uh git as the essentially uh underlay in person under the under underlay terms as well as the your platform to host your uh the corpuses right the corpora uh after
00:27:32
you take notes and you can use you know as i say any tool um and after publishing them to git you can let another know and the aura will just keep pulling your notes uh while they're available so you keep
00:27:43
control your data which is definitely like one of our design principles uh and here uh okay so just i guess shortly late i have this is the end of the hanging talk there's much we can cover
00:27:56
but uh of course first of all thank you for listening uh there are pointers here you can jump to these slides by uh it's engaging go our slides in diagonal or of course explore
00:28:07
at your own pace um we have uh features that we believe could be used for of use to the community which is sort of the intended behind the hour to you know uh require a little and try to give back at no cost so uh
00:28:22
yes uh please also let us know if you have any questions uh you can reach out to me or any other members of the franchisee collective and thank you very much eduardo we have one question
00:28:32
specifically for you which is how do you from robert hayes field how do you approach the problem of people using different words to describe the same concept
00:28:45
yes this is a very interesting question which uh of course yeah so the short of it is that we try to not be opinionated about exactly how you do this but rather
00:28:58
instead experiment with um what we call well our reactions in this case uh is uh have to do with this which is experiment with
00:29:09
atomic or basic concepts that could be used to eventually reach consensus on you know canonical entities equivalence
00:29:24
and you know other interesting semantic relationships on a social level so precise so basically we try to the idea here is to build this uh sort of meaning without impulsing like uh you know a
00:29:37
course when it comes to like applying a schema or a particular um a particular way of doing this uh so essentially or surely we don't we are we actually don't do this now but
00:29:50
rather give users tools to optionally do this according to different standards great um thank you thank you uh thanks a lot for having me and for inviting me it's really exciting
00:30:07
to meet these people that have been following your work for a long time i'll try to be sure to be a able to enjoy the conversation so i'm bastian gary and
00:30:18
um i'm the maintainer for imax awkward as i said i would like to go back to classics and 2005
00:30:33
centuries ago we had socrates criticism of written culture saying that notes and writings come with forgetfulness [Music]
00:30:46
words written words are just dead passive expressions of thoughts and the people who express these thoughts cannot reply they are not here
00:30:58
to claim authorship authorship for what i said so thoughts written thoughts are like orphans so he was a very strong criticism
00:31:12
and the paradox on is that his student plateau has been writing he has been writing dialogues famously and these dialogues are about real
00:31:24
persons discussing about something they do interact live thinking out loud and the output of these conversations is not new information actually it's all about
00:31:38
unlearning what uh the persons believed they knew i'm sure you're all familiar with this but it's always interesting to think about these two philosophers one who
00:31:51
said there is a danger uh in notes and written words and written culture that we are completely surrounded with today and the other one the students will use
00:32:04
writing as a way to convey uh not just information but a way of thinking and a way of of interacting with oneself so just starting from this
00:32:19
i define notes with a purpose so words written words are like a pharmacon that are both a remedy and a poison and i i get from there from that this
00:32:35
definition of good note taking tools they help us fighting forgetfulness by stimulating thoughts with contextualized information about the woo the what
00:32:47
and why so this is very general but i think it's uh still helpful to guide us into thinking what uh what we should do about the tools that we are shaping and
00:33:00
writing and good digital note-taking tools they easily blend into online conversation so i'm really glad i heard this word of conversation uh
00:33:13
by uh uh in previous discussion by ward cunningham because i think that's really key here into designing the the thing we are working on so we have models and nt model
00:33:28
the anti-model is the word document floating around with no versioning no tracking of authors and no easy sharing and most of the time with no context and i have three models one is mediawiki with pages and
00:33:41
discussions and history and authorship and all what you need to get creative for what you write and get the context about what you write the the second model is email driven git
00:33:53
workflow uh with versioning and authorship and also this same plain text format for having conversations about what you do in the same place plain text format for patches
00:34:06
that you contribute uh when you write code and the other model is blogs and errors feeds i think all the three models are really conversations
00:34:17
and that i i think good uh digital note-taking tools are really uh going towards i will just take one minute uh to present old mode i think this is not really the purpose of this
00:34:31
conversation and i invite people to discover it so this is an argument file plain text with some formatting the main so org
00:34:45
motto is about your life in plain text it's a personal information manager authored by cast and dominique like sometimes like 18 years ago already plain text format
00:34:58
similar in spirit to markdown a set of tools to manipulate these oak files mostly accessible from within new emacs but not only i think there are some modules in vs code for example it's a free
00:35:11
software so it's published under the gpl3 or later and it's standing off on the shoulders of e-max and this is really the the engine for every augment feature
00:35:25
also because if because it's free software most of the features i've been designing have been designed uh not implemented by but proposed and designed by the users and i
00:35:37
think that was key by on on starting a useful tool is to have the feedback of users and to be guided about what they really need and the single powerful idea that casting started with
00:35:49
was usually you had tools for to do items and task management and two is for not taking and he completely uh refused this distinction and said we need one
00:36:03
tool one from the same two for note-taking and to do items so the features that we have is this minimal syntax that i've been talking about folding is at the core of augment since
00:36:14
the last 20 years 18 years but folding was already present in imax ogmod made folding for the people restructuring restructuring changing the
00:36:26
structure of the document should be as pleasant as editing text and i think that's still key in every note-taking document and we are very passionate about this perspective
00:36:39
notes are like a database you can customize the way you view and navigate the to-do items and you should be very flexible about this context you can capture notes from your emails
00:36:52
from a file from a url we have modules on firefox and chromium to be able to keep notes within all files documents are live documents literate
00:37:04
programming is implemented because you can evaluate code from your documents and sharing you can export your documents in multiple formats so when you combine org files with git
00:37:16
for example sharing these files with someone else then you have the flexibility of this uh very minimalistic format and the powerness the the the power of
00:37:28
emacs to edit it very fluently and an easy way to share and to have a conversation with others based on your structured file and that's it for me
00:37:40
thank you thanks so much dan for organizing all of this and inviting me to the talk um today i i thought um to riff on um your intro in um your
00:37:55
invite invitation email where you mentioned the sentence how our future note-taking infrastructure can serve us wherever we are as constant companions and
00:38:06
here it became clear that the reality of right now is that our workflows change over time um so our apps and services change over time too and so it's really also that our companions change and
00:38:19
in the whole like problem of interoperability um what we see is that the current landscape makes it very difficult to migrate between services or integrate between all of them and this goes also to the
00:38:32
heart of our shared concerns about interoperability and data portability between the services um that we use every day and also between the services that are like where the founders and team members are right now um
00:38:46
present in this call and we also know all how tough interoperability is so standards need to be developed adopted but also also most importantly they need to be updated over time and
00:38:58
that last part is actually also one of the more tricky ones and it represents a massive coordination problem between the actors of an ecosystem it's one of the reasons why http is still on version 2 after
00:39:12
30 years and but ultimately interoperability is ux problem so the question is how can users integrate their current tools with other tools they use and easily migrate um and the goal here
00:39:25
is how do you stay in the flow without interruptions and these interruptions can come in the form of copy pasting stuff around data migrations etc so that's the reason sorry so that's the
00:39:38
reason why this talk is called in the flow sorry and um how do we solve this problem with um bespoke interoperability and what i
00:39:49
mean with this is if we look at the intro ability as a ux challenge with the goal of increased flow states there may be steps that tool builders can take that
00:40:01
do not require large-scale coordination and that already brings large benefits to users and in the long run create more interoperability in the network overall and i found it very helpful to think
00:40:14
about the context of the tool you're building so what's the workflows that span beyond your tool which other tools do users use where an integration has outsized effect
00:40:27
and myself i'm the founder of an open source software called memex and it's a tool to collaboratively curate search annotate and discuss web content such as websites pdfs youtube videos and soon also images
00:40:41
and in the context of them of memex we have identified that the larger workflow is what we call the creator workflow and it's a loop of content discovery you see here on the top right uh capture synthesis
00:40:55
and then sharing and those happen all in very very different tools and they're all interconnected in those workflows so when we thought about this problem it was the question how can memex strategically provide
00:41:07
integrations to key tools and workflows in this loop so for example in the capturing step we enable importing of html files and your browsing history but we also have a copy paster that
00:41:20
allows people um to search everything they saved and then define a custom template on how to copy paste content from namex into other applications say for example a list of all of the
00:41:31
websites that you have curated for a specific topic you can do that with two clicks and you have it in rome or notion or whatever application you use to further synthesize your thoughts
00:41:43
mimics also integrates with read-wise so that people can collect their annotations into one single place for multiple sources uh it's what ban also described this that becomes a very powerful use case right now
00:41:55
specifically for power users that use tools like roam notion or evernote um then we also have a read api for all of the shared collections so that curators can integrate their curations
00:42:07
into other applications and then people can subscribe to those as rss feeds later uh that's not there yet though and so here's an example for example on how culture hack it's a they built an
00:42:20
editor to to write blog posts about culture change um has integrated the memex collections into their editor to quickly reference annotations and pages they collaboratively saved and curated and that's that worked
00:42:33
via our read api that we developed for and in beginning now for like trusted partners that want to integrate mmx data and trial it with us
00:42:47
so yeah these all are making memex more interoperable without the need to adhere to many standards or coordinate with too many actors some of those integrations are more bespoke than others though the key is that it's small steps
00:43:01
that can be taken to make data more accessible over time and so that more youth standards can be tacked on or new ones can emerge for example
00:43:12
um for example the work we did for the read-wise integration we can use for a direct roam integration later or our read-only api can be used to make the rss feed possible
00:43:24
a great example for a an immersion in tropicality is the double brackets that rome obsidian notion and many others are starting to use and we're adding this to our editor later too and with this uh we don't even try to adhere to
00:43:36
defined standards that many um we don't try to adhere to define standards that many tools not even adhere to but that the most important tools in our network do and so it's very likely that this
00:43:51
becomes a very like default syntax uh in many other tools that currently may not even think about uh what bi-directional linking means or what concept tagging etc and to get
00:44:04
there we didn't need any central coordination so no standard was needed to be defined just a bunch of tools that saw the benefit of that syntax admittedly for annotations it's a bit more tricky
00:44:16
to harmonize them since they're not just concepts or pure data they also belong to an author which has currently their identity tied to a specific service so the easiest interoperability may be enabling an import export in an open annotation
00:44:30
data model whereas the tricky part is making annotations accessible and zoomed across different services and for that i don't really have a good solution yet so historically interoperability was seen as a potential decremental
00:44:43
decrement or potentially decremental for a service because you may be able to lose users to competitors more easily i think this is currently changing tools and compute consumers realize that there's an enormous advantage of
00:44:56
building interoperable tools and this is because there's an infinite amount of custom workflows that people have no single tools will serve that workflows evolve too so users really want adaptability of their workflows they want custom
00:45:09
integrations to their favorite tools they want to migrate easily to tools that serve their leads better so i feel for for tool builders that are serious about building useful software interoperability can actually be a huge
00:45:22
spoon uh to their success um it'll make easier to acquire users it will make it easier to help users embed your own tool in their workflows um you may be also losing some users but
00:45:34
that's okay i guess because it'll create sustainable pressure to you for you to focus on an audience well and build really really useful services for them and if you do that they also won't leave as easily
00:45:48
so i think the future of notetaking will be a highly fragmented yet interwoven network of tools services and paradigms that cover this infinite amount of combinatory possibilities
00:46:00
and so bespoke interoperability may help us to overcome this coordination paralysis that we often experience and yeah we can take the first step in meshing together and then see where more coordinated
00:46:13
interoperability is necessary without over-engineering a top-down model from the start oliver thank you thanks and i really want to pick up where oliver left off here
00:46:29
my thoughts um are actually very similar um in some ways um and uh and mine are really center around
00:46:44
us a small feature request um and my kind of view of this ecosystem um is moving from the edge the edge being
00:46:56
kind of where you are in the margin of a book where you might be annotating to the map um where you might be um doing your kind of thinking and reflecting
00:47:09
planning organization writing list making to kind of higher order functions calendar and coordinating task management and so forth and the different apps
00:47:20
that are out there kind of fall in this spectrum of focusing more to a greater or lesser degree in part of that spectrum and of course it's
00:47:32
highly overlapping um some uh the features of functionality for instance in word mode have a lot to do with basic note taking but also include those higher order functions as well
00:47:44
for the purposes of my feature request though i want to look at just the edge and the map and the problem that kind of exists between these two um
00:47:55
because if you're out there annotating um and um you know the awareness of your your notes your topic map or the other aspects of your kind of
00:48:08
your pkm your personal acknowledgement system is really limited to the home app that you tend to keep that stuff in um you can import from say hypothesis to another app but it's a pretty crude
00:48:21
one-way trip um that only moves um you know upstream from left to right so if you're in the margin of for instance a document and taking a note here on a wikipedia article about personal knowledge
00:48:34
management you can take that note and save it and in this example maybe use read wise
00:48:46
bless your heart uh daniel to take those notes from hypothesis and move them into rome but what you can't do um is in your in the annotation editor is
00:48:59
double start double bracketing the word note taking and have it reference um the your all the other notes in your kind of topic map that you've already created
00:49:12
um and i think you know this is a lot of folks work around that by just double bracketing the word um creating essentially what is a markdown link
00:49:24
and just importing it into the knowledge management system um where they'll use it there and then in that system once imported it will naturally refer to the rest of the topic maps but of
00:49:37
course you get some you lose some key features um in in that trip um so if you were for instance to start the process of taking um that
00:49:50
uh to making that link you would lose um the process of uh you know the the auto suggest um knowing that you you know what those um links are that are in your knowledge
00:50:03
base um catching any misspellings that you might be making and the and and just the easy process of being able to start and tab to this the the
00:50:15
response that you want um you may also want to know which one of your notes if you were using something like the agora which one of your notes is a personal
00:50:28
note that's only available to you versus which one of your notes is a public note that you use as a way to create some topics that other that you're kind
00:50:40
of sharing for the community potentially even for their collaboration and their input wiki style and you might even kind of extend the notion further to where you could be subscribed to
00:50:54
multiple pkms at the same time not only your own but also somebody else's or even public um kind of wiki based systems like wikipedia and um you know so
00:51:08
you could choose whichever you know whichever one you wanted and only subscribe to the ones that you want to rapidly autocomplete um and then this way you could almost imagine this as a as a form of kind of controlled
00:51:20
vocabulary um tagging um that um that the user would be completely in control of um so i think um what might be needed for so that
00:51:32
would kind of blend the edges between systems you have your knowledge map your knowledge topic with it with you wherever you go no matter what app you're using um and but it needs a few things
00:51:45
needs a kind of a commitment within the ecosystem to a protocol in terms of how to request things some pretty fast apis if you're doing type
00:51:58
type ahead and auto suggest some some perhaps conventions for referencing uh different pkms if we're in a kind of a multi-knowledge-based world um and then of course services and clients
00:52:11
uh to implement all that so that's my uh that's my bit um and let me go ahead and kick it over to connor yep great awesome um so i'm just gonna
00:52:24
actually be presenting out of my room um or the company one that we're using um and yeah it was interesting seeing the prompt of uh sort of like what's the future of rome and thinking about the future of
00:52:36
note-taking um and so i uh figured i would title the talk um rome is not about note-taking um and yeah start from the perspective
00:52:48
that we've never really been about notes which is kind of strange because uh if you look at our landing page it says rom is a note-taking tool for network thought now this is a landing page that we also haven't
00:52:59
updated in uh you know since october 12 2019 um but uh what i'm going to talk about is what rome's sort of long-term vision always has been and
00:53:11
why note-taking has sort of been the um the entry point and we use we use language that people are familiar with um rather than introduce two new ideas we we connected to something that was familiar
00:53:24
with something that was new um and it it seems to be um it's gotten us somewhere nice um so rome is always the goal of of rum i've been working around uh i've
00:53:36
been working on the problem since uh 2008. um and uh well um slightly bigger
00:53:49
um but rome is always but i've been working on rome as a company um full mish since um 2013. um and our goal has always been to build a platform for collective intelligence um this
00:54:02
slide right here um is the one i saw in like yeah 2000 2008-ish which is this is this is moore's law and it's sort of a stand-in for just um the exponential increasing leverage
00:54:15
that technology has given us over like our physical environment um and my concern at that time was that like all of our sense making institutions this is too small an image but um you know
00:54:28
whether it's our our print uh our you know journal system our systems of government they were mostly designed in an era of uh you know like print um print media
00:54:42
and uh or or the you know or broadcast media um and had a they they have been getting worse and worse and it sort of looked like an inevitability for me that they would not be able to keep up
00:54:55
with rapidly changing environment um this is and i think this year this has particularly um become obvious to folks if you look at sort of the uh dramatic reversals and the
00:55:09
difficulty that we seem to have in terms of building a truth-seeking or a sense-making society and the ways that that's been impacting our culture was sort of like the politization of um of everything
00:55:24
including including science right now um but i had some uh i had a lot of hope um i still have a lot of hope um you know stemming back from uh from
00:55:35
2008 or so um this is the great pyramid uh at giza um and i put a little um interactive model in here um so you could you know say that maybe all the workers
00:55:48
um who worked on kufu works 10 hours a day maybe they worked you know 12 hours a day it's kind of back breaking labor but maybe they were able to to do it um and um
00:55:59
yeah what i uh um but if you took the there's about uh youtube was was bragging about how there's about a billion hours of ads watched on youtube every day
00:56:19
and so my initial interest in getting involved in the internet at all was figuring out how you could um like tap into that cognitive surplus because you know every day and a half
00:56:33
um like we are we're putting as much energy as the total man hours that went into to building that pyramid um and i had been particularly motivated by clay shirky and yokai
00:56:45
benkler and the idea of common space peer production and that there were many potential wikipedias left to be created i do want to like shout out ward while he's here um and i can i can go into this in the
00:56:57
future but federated wikis have been in some ways even more inspiring um uh than than wikipedia itself um okay so yeah uh i'll fly through the history but basically from 2000
00:57:10
and uh until 2011 i worked on trying to build a collective intelligence and collective action platform that was based on local governments um it was cool for like getting invited to the white house and you know like honored by obama but like the
00:57:24
reality was that it was a complete tire fire um because we were trying to uh one i was 19 and um we were trying to both figure out how to create a user interface for
00:57:36
um like really crowdsourcing uh and and extracting tacit knowledge and organizing knowledge from a ton of different people in a community um and directly channeling that into a um into the political arena
00:57:50
and we had focused on local government because we had hoped that that area would be less um prone to like the hyper partisanship or people intentionally misrepresenting each other's positions or just
00:58:01
um sort of uh willful misunderstanding or sort of uh narrative pre-selection of stuff but it's it was i found it to be just as bad as as a national um discourse
00:58:14
ironically we ended up getting acquired by aol um which only confirmed for me that like um a lot of our media institutions are not necessarily interested in finding like the nuance
00:58:27
uh or like novel position especially if there's a way in which the like incentive landscape allows them to score a ton of points against a perceived enemy so um this is a
00:58:38
um an image i love um around you know one of the things that made it really hard running that first company was that there were so many hypotheses where we could be wrong and there were so many uh places where
00:58:51
you know in order to solve one problem we created a ton more and um and i realized that you know i'd not been approaching it sort of scientifically um so uh from you know 20
00:59:04
you know after after that acquisition i was interested in how do you simplify the problem and basically take something from you know if your if your goal is something massive like try to figure out how to uh build a better sense making system and a better
00:59:18
way for people to learn from one another and you know update on complex ideas um what's the easiest way to start and i was interested in you know uh flywheels buy wheels and
00:59:29
feedback loops you know focusing on something that was small enough that we could get off the ground and then um uh you know be able to progressively
00:59:41
um extend our reach and so um one way of thinking about it and this is the approach that rome ended up taking we get a lot of criticism right now um for the fact that we you know charge 15 bucks a month
00:59:54
but uh rome was essentially unfundable for years um people didn't think there was any uh any possibility that you could build a business in the knowledge management space especially not with something as weird
01:00:07
as a like graph-based outliner so we couldn't really raise venture funding or investment our initial go to market was focusing on and i i do want to challenge ward's supposition that like
01:00:19
a business model means that the incentives are not aligned between people um our way of getting funding off the ground was uh we were able to piece together from a bunch of different sources
01:00:30
um uh like um the idea we were able to convince some funders in the like effective altruism research space that there was some plausible chance that rome could make
01:00:44
their researchers you know two to ten percent more effective and if it's a space where you know there's yeah in that case there weren't a ton of um uh math phds who were you know um
01:00:58
interested in like a certain kind of um applied philosophy approach to um uh it's ai alignment and you know um ai safety um we we bootstrapped the company for a few
01:01:12
years um just you know keeping it open for um like a dozen to two dozen people um and so this is this is currently the stage where i would say we're at um
01:01:23
we we iterated around that and took a big leap about a year and a half ago to open it up to the public um and you know made a decision to um to charge a fair amount even though
01:01:35
like currently about uh about a third of our users are either on a free or discounted plan um but rome made a conscious decision to um you know we we expect that we are the
01:01:49
very early stages of the problem of trying to figure out how to enable many-to-many communication and building a real hyper media tool for um for representing thought i'm going to get into that
01:02:02
but but yeah currently for the last year or so the main thing we've been focused on is sort of how do you um how do we internally use the tool to think better together so
01:02:14
that as we add more people to the team we are not overrun with coordination costs currently we are about a dozen people on six continents and um and that can be pretty tricky
01:02:25
um because you don't have uh a lot of the the advantages of in-person coordination and so you know the goal for rome is to figure out how can we construct better than plaintext
01:02:38
um like better than than pros essentially um for allowing uh modul like um allowing a um
01:02:50
extremely remixable extremely reusable and like robust collaborative knowledge work so you can get into a situation of um you know uh
01:03:02
serendipitous collaboration uh one might say or like um yeah um how can you like how can you get some of the dynamics we have in open source but in all sorts of other domains
01:03:14
um and particularly the thing that we're really interested in is you know can we um can we figure out how to efficiently allocate expert attention so that
01:03:28
you know the person who's who has the key insight can have hooks into um into the work that other people are doing and you know focus on just the part that is relevant for them but still be able
01:03:40
to bring in all the relevant context that's necessary um and uh and through that um you know like be able to have a highly leveraged contribution and eventually be able to bring in
01:03:52
a larger um more cognitively diverse different skilled group of group of people mine will be a short one because this whole collaborative annotation idea
01:04:06
is pretty new to me and to the log seg team although we have we do share some similarities in our visions so uh my talk will mostly be a
01:04:19
brief introduction to what we do and mostly i will listen to the ideas from this panel so my name is junior and i'm from the log seg team in case
01:04:33
you don't know logitech is a privacy first open source platform for knowledge management and collaboration
01:04:45
it starts as a personal um project from our founder attention and it draws draws many inspiration from other tools for example room research so
01:05:00
we thanks for all the great tools that we learn from and we think in this age of information explosion and we try to build a tool that help
01:05:16
users manage their knowledge more easily in the form of connected graphs instead of just a standalone content what's more to make the user the so
01:05:28
owner of their data and we put privacy and the data security at the most important position by utilizing encryption embracing
01:05:43
an open source method open standards and enforcing data security in all aspects we chose the open source way because we believe not only
01:05:55
it improves transparency and encourages participation but also it makes a tool for knowledge
01:06:06
more to everyone as it should be we have a long way to fulfill this vision but so far the story looks pretty good although it begins as a humble personal project
01:06:22
from our founder logsec already have a pretty vibrant user base and a welcoming community in the app the web app and the desktop app have
01:06:36
recently went through a query factory resulting in a much better architecture and user experience and we also just launched the early access for our plugin api
01:06:50
and the users are starting to build amazing plugins with it finally we think logsic's vision aligns with the collaborative and
01:07:06
you know collaborative knowledge management ecosystem in large we all understand the importance of open standards and fully embrace them and we aim to
01:07:20
build not only better tools to manage knowledge but also a better platform where people around the world can share and collaborate and build communities
01:07:34
around it and by listening to the previous talks from this grid panels i in my previous experience i think this
01:07:49
is a problem mainly unsolved for example the interruptive interoperability between different you know note-taking apps but after after listening to your talks
01:08:02
i start to think that that this is a problem we do can solve you know by embracing open standards and build upon it yeah
01:08:17
so that's all for my talk junior thanks so i'll kick it off with a with one question um which i call the kind of the 25-year question um 25 years being a um
01:08:39
a long time not a crazy long time but maybe a sufficiently long time that um technologies that are kind of well conceived and durable
01:08:52
um tend to last um the web passes the 25-year test i think ward's wiki is got a couple years to go but it's getting close um and so if we think about
01:09:05
note-taking kind of the future of note-taking in the concept of the 25-year test um what are what what's most important to think about um and either
01:09:17
in terms of features or capabilities or properties um so um i'll toss it out there um for anybody that wants to break gonna raise their hand and jump in
01:09:28
oliver do you want to kick it off yeah thanks for the question um i'm i'm not sure on a on a like technical level what will be needed um but i assume that
01:09:43
if like a 25 year mark wants to be hit by any product it requires a significant sustained effort over that period of time to work on this and i know for example
01:09:57
connor has a very strong like is very vocal about his uh like 25 year plans um similar to mine is like to fig to figure out what's the core purpose that drives
01:10:09
you to build a product that can then solve a problem that will only be solved solvable in 25 years like for example we both describe
01:10:22
this need of our society to solve the sense making and decision-making crisis that we have because if we don't solve this in the next 25 years the chances that we're surviving the century
01:10:35
or thriving beyond this century are are limited and um i don't want to be in a mad max scenario i want to be in a star trek scenario and um so one part is definitely a strong
01:10:49
vision on where things should go and then an ability of the infrastructure and the product to iterate towards this um and be adaptable to the changing environment that
01:11:01
10 years from now will provide or 20 years from now or 30 years from now um yeah uh connor so the premise is like i'm curious what's motivating the question like um
01:11:21
or sort of like there seems to be an unspoken assumption in there that like um that uh so the web and wiki technology are 25 years old and that can you can you just say a
01:11:34
little more dan around i mean what what will define the note-taking apps that we use in 25 years i guess maybe is a another way i mean what what is most important another way to
01:11:46
put it is what's most important about what we're about to build um what do we need to focus on in order to provide really durable lasting value for people to organize their thinking and collaborate with
01:11:58
others so it sounds like for you there's a uh like a um like durability is an interesting um it's an interesting idea because the thing
01:12:12
that like this this problem is uh you know we're all we're all sort of following off of van of our bush's memex in a sense right and like you know from him englebart's uh you know
01:12:24
like um uh yeah like the mother of all demos and like there's a you know the original idea for how to how to deal with the crisis of um information overwhelm and the fact that
01:12:38
you know this the knowledge needed to organize our or to run our civilization was something that no one individual could possess and the like you know expertise was becoming so siloed it was hard to
01:12:50
um have a good understanding of more complex systems i mean the original pitch was microfilm like that would be pretty durable um if we still had material for reading microphone but like how durable are vhs tapes either right now right
01:13:02
so um like the i think i i sort of want to say something like that um uh i'm reminded of of um of a essay from brett um
01:13:16
uh the web of alexandria and it was just pointing out like the most durable knowledge storage thing that that we know of um is dna which is constantly replicating itself
01:13:29
and propagating in like um you know and is is you know the the information itself is undergoing change um so you know our first feature when we opened up to the general public was
01:13:42
uh export to plain text as markdown export as json so it could be machine readable and like manipulable by other people um and i uh i hope that is one of the reasons that um you know people were able to get a head
01:13:55
start on uh um you know like sort of open source alternatives and other ways like the nice thing now is that there's many ways originally people were concerned about starting to use rom because they were concerned about data lock-in but
01:14:07
now there's a half dozen different things that can read that realm data but we started with the perspective of like if we figure out how to solve the problem well and represent the ideas better than you can represent them
01:14:18
in text um then like it will like uh other people will want to have their own way of reading that data so yeah i mean
01:14:31
uh memory diamonds i don't know if like the internet shuts down we're so like um like that's uh um like there's a lot of things that could that could make things really a lot worse so
01:14:42
like durability isn't it like i i can't get a notification if a paper that i read five years ago is falsified that's the thing that makes me feel like the whole infrastructure we have of of knowledge is completely not durable
01:14:56
like our our journal system is not durable like uh the library of congress i think 30 of links within five years are dead so like i think the distributed web gives us a bunch of hope i think that there's a lot of um like
01:15:09
wards federated wiki the idea that like when you when you copy something um you uh you make your local copy of your own um i think there's there's a lot of approaches that are better than just a centralized server system we happen to
01:15:22
have started with the centralized server system because the hardest part for us to figure out is the ux of how the um the collaboration actually works and you know we haven't yet figured out how to visualize version control but like um
01:15:35
but really like the end is going to be it has to be decentralized so that's my that's my answer right i'd like to add to that that uh you know certainly decentralization and
01:15:49
you know json is nice because they claim it isn't going to be versioned yeah the the thing that we really tried to tackle is this idea that you
01:16:00
would actually do work in the wiki instead of just talk about the work so we wanted to have some way to to bundle uh data and computation in a way that could persist longer than
01:16:12
the implementation technology that we choose and and so we don't call it a domain specific language i think that dsl has been abused but
01:16:24
but we call it markup we just say different types of things have different markups and then we check and see how simple we can make that markup and still get work done because we know
01:16:36
we're going to end up re-implementing it over and over and over again so to to keep that one step of isolation from the underlying technology it will be digital i think we can be pretty sure and you
01:16:49
can probably represent integers but it might not be ieee floating point wait i do i do this give me a point that i kind of want to raise i'm reminded of um
01:17:01
ted nelson's eulogy for duggan of art where he said um i honored doug's legacy by keeping the links outside the file as doug did and there was i was it was interesting who wasn't included in this um like there's a project called codex
01:17:14
editor um that i think is doing really interesting things with standoff notation like we we started with uh like an embedded markup um but it's so not the right way like it's
01:17:25
that was for for pragmatic reasons of um you know what we could well we could ship um i do think the future is going to be in standoff notation i get that that's maybe a bold claim but like um i
01:17:38
think that keeping links outside the file is pretty essential if you're getting into collaborative stuff let's um jump to junior yeah um i think um predicting the future is always hard
01:17:55
but we can learn at least to learn from history and for me and we can for example for example we can learn from the web you know
01:18:06
after all these years the web is still around although it also has some challenges of its own so why is web still around and uh reasons there are many
01:18:19
reasons for example it's it's open it's and it's accessible to everyone and also as a platform it may change for example we may have
01:18:32
better browsers um different network protocols but the content itself is always there so that's
01:18:43
one of my take and also i just saw in the chats someone says that the note taking applications will always change you know it will but the
01:19:00
data itself the note itself will be around for a long time and but it takes it takes effort to make it your durable yeah so that's all i can think about
01:19:15
right now thanks uh bastian yeah just one idea i think git has been quite a change for all the developers
01:19:30
uh it's going to be 25 years in 10 years and less than that in four years now no like nine years and um i think versioning
01:19:43
will be pervasive into or not taking tools in ten years and perhaps gift is the right infrastructure for that i mean we will use it without noticing
01:19:56
uh by using online note-taking and i think this will bring a lot of interoperable interoperability into the things we do the same way it has
01:20:13
changed where developers can share code so i'm not sure yet all the consequences for this uh change but i see it coming and i see it as a
01:20:25
structural change for collaboration and nurturing thanks bastian uh eduardo
01:20:36
thank you um i mean i want to add um to the amazing comments so far perhaps another perspective i think that um a very interesting sub problem that is considering for this question
01:20:49
that is a great question by the way uh is that of the the seating problem the bootstrapping problem and essentially what uh what i mean by this is like the
01:21:02
responsibility we have as those potentially towards the beginning of a 25 year long process right so so here i think about uh
01:21:15
things like inclusivity and diversity right uh so making this making sure that the system reveals the systems we are building are as inclusive and uh and diverse and available as possible
01:21:28
and this is where the words uh of um or several speakers really resonates with me so that's like you knew so you know having something that can be
01:21:41
used by all the people and that we know that we will be able to be used by all the people in 25 years right so essentially what legacy we are living with the architectural decisions we are taking and also uh with the actual
01:21:54
seating of the knowledge and the ceiling of the nose and the sitting by example we are doing by being the first you know the relatively first few users of these systems so this i think is very important um i i
01:22:07
personally like the thing that um it whatever takes the place of this system you know i as you know i call it whatever i don't want to impose the name but you know like this these comments we're building this
01:22:19
knowledge comments uh whatever takes the place of that i hope that in 25 years time i will say a 13 year old or 15 year old with plenty of time and interest should be
01:22:31
able not only to use clearly for free what we are saying but also implement that would be my my wish someone that you know
01:22:43
can just uh that knows a bit for programming has an interest should be able to get and and and tap on to these like past knowledge commons that i hope we are building together thanks roberto um dana are you back
01:23:00
uh maybe can you hear me yep awesome um yeah so i'm probably the least qualified person to opine on um you know super long-term visions on where note-taking is headed given that
01:23:14
uh we're focused the least on that uh but what we always talk about is note-taking right now typically takes place within an app so you have to contact switch from whatever it is you're doing
01:23:28
to take your notes and so we imagine in the future note taking will be more integrated at either the the browser or the operating system level and in this way the notes will be much
01:23:41
more tied to um the actual context you're in that's definitely the the vision uh we see you know for our use case of reading we definitely want to help people take notes better
01:23:54
while reading as one example as opposed to having to have a separate note-taking app or a physical notebook open uh while reading uh thanks did i miss anybody oliver did you go already
01:24:10
yeah i did but i have actually one more thought about it and i think um i think it was eduardo who's who mentioned this um or parts of that um i think in order to
01:24:23
make these uh systems resilient and survivable over time uh the whole topic of interoperability which we talked about link at length in in the previous um uh
01:24:35
like presentations and also the introduction of this session was spiked with with this topic overall is if we if we build interoperable systems then it means that knowledge has a
01:24:47
continuation that that actually can be executed because of course knowledge that is in a note-taking application that has a lock-in is still there but it it does not it's it doesn't
01:24:59
allow to take that knowledge into new systems that can gradually survive all of the extinction events that happen um for each of those applications as they become too old to serve the use cases of
01:25:12
of the users that they were intended to to serve in the first place um and if we see for example um the internet um as a interoperable system
01:25:24
has proved that to be right right even though it has its drawbacks it's hard to update but ultimately it survived until now over a 30-year period almost unchanged because it was built on
01:25:36
the foundations of interoperability and meaning uh you could move from one server to the other you could still talk to everyone else um so yeah over time if we as a collective
01:25:47
um community of note-taking or like just generally tool makers knowledge tool makers keep an intention to stay interoperable and make it easy for people to run move to new services and two if they
01:26:01
move to new services have at least as frictionless as possible although that's very that's one of the hardest problems that we can solve in like communicate with people that use other tools so
01:26:12
an example that i that i bring up often is why do we do we have to reinstall uh and and repopulate our entire social graph when we move from whatsapp to telegram
01:26:24
why do we why don't we why we're not able to talk to our whatsapp friends that are not that haven't changed at the service and so yeah of course this is gonna be a hugely um
01:26:37
complicated problem to solve but um uh yeah definitely i i would agree that's key that's definitely key and i will say that if in 25 years uh these people in the future some of
01:26:51
which we asked but another one of us uh cannot take their whole social networks together with all their data and just move platforms uh with the
01:27:02
uh snap of a finger we have failed we have failed them i think it's one of the biggest things 25 years ago there were 20 million users of the internet period there were no mobile phones like the level of
01:27:15
pessimism on this right i'm like are you guys not seriously trying to think about neural link and like you know room scale knowledge representations that are not verbal like what the hell are we talking about with note taking even is it
01:27:27
knowledge captures and all compression and i think on our might we will respect i think that what you point out is very interesting right i mean of course the media the form of the resources that people are going to share in the
01:27:39
future we i want to think we can't imagine if we could imagine them all we will be living a boring future in 25 years so i agree with you there that the forms and the media and the things being shared
01:27:50
uh will change but i believe that the basic problem which has to do with power dynamics to some extent i believe uh the data who really can make a claim on data and
01:28:04
and actually uh you know like interrupt it like oliver was saying that problem is sort of like orthogonal to the actual nature of the media yeah you know i want to claim uh
01:28:16
complain for a minute when we're talking about longevity and that when i ask people to put up a server and share their content and their content is unreadable because it's
01:28:28
in http and there's some theater there about privacy but i noticed that every raping and plundering site is running https you know that that basically means you're big and conglomerate and
01:28:41
dangerous and little people just putting up servers are denied access to this medium so who thought that was a good idea it used to be called breaking the internet
01:28:55
okay um we've got uh a couple a bunch of great questions here in the q a so let's move to the audience um but one short simple question that um
01:29:07
got a bunch of uploads right away was what if any role do you see ai taking in collaborative note-taking and i and i guess you can take it
01:29:19
that any way you mean it um you could in terms of ai in the formation and creation of notes or ai in terms of its the ability for ai's to reason over the collective collaborative knowledge that
01:29:32
we're all creating um anybody wanna jump on that one uh clearly an opportunity there in translation you know it's proven to be effective
01:29:50
even if humorous on occasion uh i'd be very suspicious of depending too heavily on any individual's artificial intelligence
01:30:01
uh for fear of you know uh you know the the the biases that are intrinsic and anything
01:30:11
done statistically anybody else on that well maybe some speculative idea i think ai will be able to get to infer
01:30:30
questions from the notes uh most of the notes i take for example are just information answers to question i ask myself and
01:30:42
i think in in as a jour party game as the ibm watson game i think it's very useful to go back to the questions for which we
01:30:57
have the answers because we will have too many answers and too few questions so i hope ai can help um tackle this issue thanks sebastian i will perhaps have
01:31:12
sure just two things like one is to uh how to to word on translation i'm looking forward to translation not only of natural languages but also of like mental models right
01:31:24
when i say x and you say why do we mean the thing it's the same thing right how to recognize those uh those occasions or how to actually translate to be more intelligible and and the other thing is like um i'm
01:31:36
also looking forward on the application of uh essentially the latest techniques uh genetic techniques that i can make creative creativity like transformer based models and so on to uh know the corpuses that are
01:31:49
actually social actually i'm a believer in the potential of uh perhaps training ais together as groups right to perform functions and so on
01:32:02
but all right let's go to uh one of the earlier questions um from chris aldrich who says i find it curious that we didn't hear much if anything about the rise of the bi-directional link either
01:32:17
within the wiki or ideas like web mention that allow one knowledge base to communicate with another is it important as i think it is um so um
01:32:31
this is you know perhaps particularly relevant too in the in the notion of the coming future of federated open collaborative kind of personal knowledge management
01:32:43
systems and how bi-directionality kind of plays in that future started off with ward since you invented
01:32:57
the bi-directional link what do you think about all that you know i i uh i don't think it's a hard problem as long as there's a will to
01:33:09
make it happen uh people who are creating things for a profit or advertising driven they want to be sticky and so the last thing they want to do is think anywhere else
01:33:20
so overcoming that is the challenge it doesn't have to be any more complicated than putting a recognizable code on every item in your
01:33:35
notes and then scraping that together and building indices from that from the uh area the the realm of interest you know one problem is that
01:33:48
you know with a bi-directional link you say what link's here and it turns out it's ten thousand you know and that doesn't help you wanna say of the people i care about in the moment for the notes what ten things i care and that you know
01:34:00
needs to be expressed somehow too the context in which you're interested you got a hand clop from that i saw it floating up the screen two handclaps well i think those are for you
01:34:18
uh anybody else yeah i'm i um eduardo and dan we had a conversation um i think a couple weeks ago uh that i found really interesting and and uh you also rift on it in your short
01:34:34
presentation uh essentially how how do we make it possible for for um in the ux or for example uh creating double brackets that you
01:34:46
when you hook into um like when you create a double brackets that it does not just allow you to search inside what the current tool is but all of the tools that are attached to it so it means also you create those
01:34:59
bi-directional links to multiple end points in the end like it can be rom it can be notion it can be mx it can be uh hypothesis it can be anagora like
01:35:10
and so um that that was that is for me one of the most uh straightforward and powerful ways of in making those tools more interoperable and making those tools interoperable
01:35:23
around the user's need again to be interoperable like when you when you for example pull up the different connectors um then it's obviously yours like if you if you
01:35:35
if you haven't hooked your own wrong graph uh into it then you're not gonna get someone else is wrong then you're not gonna get your wrong graph and you won't get anyone else's wrong graph as a suggestion there if you
01:35:47
haven't in some form connected to it and again this this again plays into this uh notion of how do we create a more user-centric um interoperability that allows them to adapt their workflows more efficiently
01:36:01
without making this a mandatory experience for for for everyone yeah if i may i think backlinks uh
01:36:17
will be the default uh for every note-taking system and we don't have backlinks in augmod right now but there is a tool called ogrome which implements backlinks and
01:36:29
that people enjoy very much what i think resonates with what daniel said previously that for now we have apps and those apps
01:36:43
are things that we need to open and we need to take notes within the apps and it's not very well integrated within the system so the whole problem of backlink and
01:36:55
link consistency in general is about having the note-taking process with closer to the to the system and being able to make references to the world system in a more generic
01:37:09
fashion uh the secret weapon of augment is that people tend to live in emacs so it's very well integrated with all the people use everyday the file
01:37:20
and even browsing the web on emacs are so this kind of integration is very powerful and yeah we are trying to think about how to
01:37:34
get backlinks but it always needs something that monitors the world system and i think uh the the the when notetaking will be
01:37:45
closer to the system then the tools we have to monitor reference integrity will be available more easily so our experience in this was that when
01:38:02
we had a uniform site maps we added links to the sitemaps in addition to pages and then it becomes natural to
01:38:14
say if you're looking at five sites in one of our things if that's your neighborhood and you ask for backlinks you'll get backlinks in those five sites and it it's computationally efficient which is important to us
01:38:28
because you know we're running on the user's computer but uh i think that that and the point was made about externalizing links but it doesn't have to links don't have
01:38:39
to be born externalized they just have to be externalizable and the context controllable a common effect of ours is is you'll be browsing along and there's a little
01:38:51
counter in the bottom of how many pages are close at hand and it'll go from a hundred to a thousand to maybe ten thousand and when it gets up past ten thousand then all of a sudden a lot of these things sort of start
01:39:04
breaking down you say oh my god i got two people too many people in my neighborhood just throw them all away and start over and you know you're focused again you know it's it's uh go away don't bother me sort of effect but
01:39:17
it's uh it's very real thanks for it kind of gets back to another question that just came up what are the drawbacks to bi-directional links you kind of hinted at that a little bit which is of course if they're
01:39:34
too broad coming from too many sources it's kind of a spammy um um environment and so um you know you really only want backlinks from the folks that you care about um and but
01:39:46
sometimes you don't know what you're gonna care about so um there's a almost a reputation you know kind of an awareness attention uh dynamic that comes if the system expands
01:39:59
too too large um one of the reasons maybe the the web didn't implement them you know some people might think that's a good idea that it never did um
01:40:13
um any we're we're really couple minutes from ending here i think i'll just throw it open back to the panelists and uh if anybody wants to to make uh kind of a comment or an
01:40:26
observation in closing um um maybe even around interoperability thank you i will i will surely start it uh perhaps by
01:40:37
saying that um it is just my personal community and that of uh the people i work with uh to do what we can to make it easy for people to
01:40:50
uh to adopt uh the aura or essentially the the the conventions that we are depending on that we believe are essentially the cheapest ones we have we know and the most generally usable and if you
01:41:03
point us in the direction of a better convention we'll adopt it and we actually planted up as many conventions as it takes so that you know people in general because of the inclusivity aspect so i just want to do i i guess i i guess
01:41:16
offer the uh offer this offered to um to essentially uh work with you to reach better standards um and of course just like
01:41:27
uh i'll follow your way right uh because you have been working on this for so for so long many of you uh so yeah thank you for your time and uh for letting me share this thanks for bro
01:41:43
anybody else connor um i'll go i'll go last i'll i'll help him okay um anybody else want to jump in a few words yeah uh just a quick word thank you
01:42:10
again uh i think there is a burst of new solutions new problems and i very much appreciated the discussion and the attention paid to
01:42:23
the longevity of some solutions that we already have and i think the more experimentation there is uh i've i've seen a lot of things that i want to explore now
01:42:34
uh the mx looks great read-wise and all that stuff so the more solutions we have uh the the more discussions like this one we need to have so thanks again you're welcome so dan are you gonna
01:42:52
invite us all next year to see where we go yeah we'll just do a little we'll call it the the near the the new new new future of uh note taking
01:43:03
um just to well thanks for bringing us together yeah you're more than welcome thanks for coming yeah i i i collected my thoughts for a second um it is funny thing that i think the
01:43:16
last time i saw you dan was at this conference maybe two years ago um it's a it's been a crazy couple years um i like particularly want to shout out
01:43:29
or like have a thank you while i'm on the call with bastian because org mode was one of a few systems that i was using as a proto-rome um before i you know
01:43:40
um that and vault and a few others and um toward because the um conceptual model of the federated wiki and his sort of thinking about um dissent and
01:43:54
you know the problems of like a sort of forced consensus um were really really big for me and then also i mean i don't wanna like now now it looks bad who i don't name but uh
01:44:06
and then i i should also shout out daniel though because he gave me the how to take smart notes um book uh which uh ended up really helping i i had only zero casting had already been a big influence but i did not realize somebody
01:44:18
had written a book about it until daniel gave it to me and um that has been really important for our community um i think they uh the curses and views are live you know like you like and i think
01:44:33
a lot of this um there's a lot of ways to focus on things that are not like core i guess right um uh
01:44:45
and and not and yeah i think my main my main thesis is like solve problems in the order that you have to solve them um in order to get a feedback loop going i
01:44:57
similarly was once a like um you know like everything must always be open source i i think there's probably a world where like rome ends up being open source at some point um but like for me there's a complexity
01:45:11
budget and there's like you know you minimize the variables you've got and it's very possible for open source systems to end up with perverse incentives based on where their funding comes from and focusing on things that are not
01:45:23
necessarily um uh pushing the frontier so and yeah i think that's my main point is that um right now we're still at such an exploratory space and the problem is so
01:45:37
like there's uh like the problem is so weakly explored of how do we represent knowledge to ourselves and to others and how do we figure out how to think better together
01:45:48
um that you know adding additional unnecessary constraints while you're trying to figure this out i'm very happy that there's so many open source clones of rome that are focused on interoperability those things because like if
01:46:01
if i you know screw this up there will still be people who are building on you know the last five years of work um but like there's so like so many problems that are still open and if you simplify your system
01:46:14
sometimes imperfectly at first right um you can you can uh you can have a chance at like figuring out what the solution might look like so um you know like the one of the main things you were mentioning dan of like
01:46:26
multiplayer backlinks that's the problem that we've been working on right how do you how do you figure out how to filter for that um based on like users within your trust network the internet didn't start with the social network built on to it and like eventually you know adding
01:46:40
decentralization and is is a whole other layer on top of it but um we'll see you know i'm we're going to be working on this for a really long time so it's just about the sequence of the problems so nice i can only echo this um
01:46:54
like we have been definitely a victim of over optimization and idealism in the past and um made it very difficult for us to build a product that actually can provide the value it needs to um
01:47:07
yep and uh so that that was also one of those insights or like one of those learnings that led to uh the presentation i made before where you really you don't need to be an
01:47:20
idealistic interoperable product where you adhere to every standard that is available right now which introduces a lot of complexities slowness of iteration etc but rather go for these very particular
01:47:33
interoperable integrations between the services that are crucial for your users to complete their workflows and um yeah and ultimately i feel this is this is maybe the commitment that will be really
01:47:46
helpful for all of us together is is understanding that the biggest blocker to collective intelligence right now is nevertheless non-interoperability it's it's it's the unability to connect the
01:47:58
knowledge across different sources of information and move information across different systems as the user evolves their own knowledge workflows and as the user is capable of executing um better technical
01:48:12
implementations for the heuristics they have in in processing information and um interoperability can certainly help and but we don't need to be idealistic about it i've i've got to run
01:48:24
everybody but thank you guys so much thanks dan for hosting um and thank everybody cheers thanks connor and daniel let me kick it yeah i mean um we've been in this space for
01:48:39
about five years now and it definitely feels like you know as of a year or two ago there was definitely an inflection point where it feels like there's a lot more innovation a lot more people focused on this a lot of
01:48:51
people coming in so it just feels like an exciting time to be working on these on these problems we all call it note taking but i imagine in the not too distant future there'll be a different term
01:49:03
for whatever this space is we're in i'm not sure there is an agreed upon one yet you know personal knowledge management or tools for thought or something like that but yeah i appreciate being invited to
01:49:15
this and being able to participate a lot of exciting times ahead indeed um i think we've was that everybody yeah um well thanks i just want to say
01:49:28
thanks to everybody for joining um this was really um a session i've been looking to forward to for a while um really appreciate you guys making time and for coming from all different time zones around the
01:49:40
world um junior had to drop off because it was probably three o'clock in the morning there isn't um but um uh it's a pleasure to be with you guys um you know in this space kind of making stuff
01:49:52
together and uh look forward to more of it in the future so thank you all thanks to everybody in
01:49:59
the audience
End of transcript