Switch branches/tags
Nothing to show
Commits on Dec 7, 2018
  1. ## Reorg

    wehlutyk committed Dec 7, 2018
    I've been pretty disorganised over the past month, and I've finally
    decided to reorganise my time, and future reorganisation processes with
    There are several aspects that came out of this.
    1) Types of activities. Important features I noted:
    * Some activities cannot be pre-timed, and risk absorbing my capacity to step back for several hours
    * Some activities must be done in the day (e.g. an email answer to respect someone)
    * Some activities must show overall steady progress (typically my postdoc work)
    * Some activities need focused attention and not too much context switching (anything a bit complex)
    * Some activities have deadlines
    * Some activities must be done regularly, even if not often (e.g. reorg)
    The best way I see to deal with this is to have a versatile calendaring
    technique that lets me visualise instantly, and reorganise easily.
    2) Planning activities
    * All planned activities must be operational (otherwise I just avoid
      them). I can plan a "make this operational" activity if needed.
    * I plan activities in a calendar, either digital or manual (postits on
      large surface)
    * The calendar has both repeated time-limited activities and sequentially arranged
      ("takes the time it takes") activities
    * The calendar can be reorganised at any time, especially the coming
    * I keep track of organisational ideas, and once I have about 3 of them,
      or when it feels necessary (like now),  I do a meta-reorg
    3) Following and evaluating
    * I keep track of activity timings using Hamster
    * I take breaks when Hamster notifies me (every half hour or so) to make
      sure I don't lose meta-capacity (this makes for a loose pomodoro)
    * I maintain a progress log along the day, each day, using this git repo
    * Every 2/3 weeks, I look at how much time I'm spending on each
      activity, how my planning is working (do I need to reorganise the
      calendar very often, and how much when I do so)
    In fact I think I'll do the manual version of the calendar, as it seems
    by far the best way to manage all this at a glance.
Commits on Nov 12, 2018
  1. feat/style: add instantclick

    wehlutyk committed Nov 12, 2018
  2. feat/tools: update

    wehlutyk committed Nov 12, 2018
  3. ## Absent lately

    wehlutyk committed Oct 18, 2018
    I've been swallowed by postdoc work lately, and am only recently coming
    back to my log, blog, and long-term research goals. Hey to that, and
    more to come shortly!
Commits on Oct 4, 2018
Commits on Sep 22, 2018
  1. ## Alan Kay: The Future Doesn't Have to be Incremental

    wehlutyk committed Sep 22, 2018
    People are, on dimension 1:
    95% instrumentalists (will only be interested in something if it serves
    their current goals, and are very conservative about changing their
    5% focused on ideas and tools (will change their goals if they see
    something that's just interesting)
    And on dimension 2:
    85% extroverts, who need people to agree with them (i.e. society's opinion
    is the most important)
    15% inner, self opinion most important
    To make money, get anthropologist Robert Brown's book on human
    universals, pick one and make a technqlqgical amplifier for one
    of them, that will make a product that's popular with people. Those are
    the things that hardly need to be marketed. at 22m35s.
    A way of imagining the future we would like is to stop looking at the
    part of the past that brought us to our current present (i.e. the past
    who's heritage we live in), and look at overlooked things in the past,
    and see what future they could yield if they were our current base.
    That's a way to do Problem Finding. This used to be funded in the US,
    but is not anymore.
    Another way is to imagine a thing that would be cool or important to
    have 30 years from now. Then imagine what that thing would look like 15
    years from now, i.e. when it would be in its very infancy. Then try to
    get the power of what that thing is running on, *today*. For computing,
    that means having a super computer and using it as if it was a laptop.
    The way to get good UIs is to do experiments with hundreds of users, and
    that's what they did at Xerox PARC.
Commits on Sep 19, 2018
  1. ## Recap of the past few weeks

    wehlutyk committed Sep 19, 2018
    I've been exploring more Dynamic Land things recently, reading the
    host of open tabs. Watched a couple Alan Kay conferences, explored the
    public HARC projects on GitLab.
    I've also been looking more seriously into testing freelancing. It seems
    worth a try to have the liberty of free time for my own projects.
    Funding a larger project (à la DL) is still not possible without a
    broader business model, or academic funding, or a good network of
    potential donors, in order to start and create partnerships with
    schools and universities.
    My previous plan is still the one to follow (see
  2. ## Conversation with Éric Guichard

    wehlutyk committed Sep 18, 2018
    Had a very interesting conversation with Éric Guichard on Monday 17/09.
    It was refreshing to see him listen intently to my cobbled up idea of
    enactive dynamic land, and to see him actually look for concrete ways in
    which I could fund this (asking how much it would cost). We connected on
    many levels: paths followed, frustrations, connections between social
    sciences and mathematically formalised sciences.
    He told me about Michel Lussault, Jack Goody which I should read,
    François Dagognet (a philosophe de la technique), Olaf Avenati (who
    looks really interesting!), and Jean Dhombres.
    He also thinks I should go to David Chavalarias about all this, or
    even to Jean-Pierre Nadal for what he's trying to do at the EHESS, which
    are things I had not imagined at all.
    Éric Guichard's drive, like Stéphane Grumbach's, seems to focus on
    the dysfunction of academia. I might be wrong, but it feels to me like
    they're okay publishing about what's wrong while staying inside the
    system. Which is fine, and even necessary, and it's great talking with
    them because they have insight, but it's not what I'm looking to do.
    ## Start of an Alan Kay deep dive
    Watched [Alan Kay at OOPSLA 1997 - The computer revolution hasn't
    happened yet](
    Points to:
    * [The Act of Creation](
    * the Lisp manual for its half-page reflective model written in Lisp itself
    * The Art of the Metaobject Protocol
    Watched [CrossRoads 2017 | Alan Kay keynote - Education
    That Takes Us To The 22nd Century - InfoSys Foundation
    Points to:
    * [Project GUTS - Growing Up Thinking
    * [Cynthia Solomon]( who
    is involved in OLPC
    * [Doreen Nelson who developped [Design-Based
    * [COLA](,
    which I can read about in Accessible Language-Based
    Environments of Recursive Theories. It's used in
    [OMeta](, itself underlying ometa-js
    which is used in [nl-datalog](, which
    led to [roomdb](
    * [Maru]( which seems to be used in
    Alan Kay's STEPS project
    Lots more at
    Had a look at VPRI research themes:
    * [Inventing Fundamental New Computing Technologies](
    * [Powerful Ideas Content and How to Represent it](
    * [Teaching and Learning Powerful Ideas](
    * [UIs That Aid Learning and Doing](
    Which I summarise in three points:
    * Locality (of construction) and control by oneself, understandibility of systems
    * Creating languages that let us scale and build systems (e.g. the internet)
    * Creating ways of communicating that go beyond words
    A few ideas that I jotted down while watching:
    * Progress in a fixed context ("pink") is almost always a form of optimisation
    * When you specialise you basically put yourself in state of mind where
    optimisation (in the pink context) is the only thing you can do. You
    have to learn lots of different kinds of things in order to have the
    start of these other contexts (e.g. a blue context).
    * If your blue background is biology, computers don't look fast or smart
    or complex at all. They're small, slow, simple, stupid.
    * Alan Kay's vision for OOP was about the processes between objects, not
    the objects themselves.
    * The systems aspects of things are deeply hidden almost everywhere,
    which is why it's so frustrating to talk in words.
    * The internet is the only example of an ecology (vs. mechanisms) in
    the computing world: it scales to 10-11 orders of magnitude, it doesn't
    * Current programming languages don't let you create systems, ecologies,
    where communication is negotiated. (Simpler example: hardly any
    programming language lets you express units for its numbers; you can't
    add 1cm to 2in.)
    * You could take all the compelling examples of ways that computers (or
    non) have been used to teach children ideas we previously thought they
    couldn't grasp at that age, put them on a wall, and start thinking from
    there: you'll see "well we don't have a programming language for this or
    * The whole point of education is not about letting them do *only* what
    they do when they're messing around. It's to learn things that are
    stronger than what we naturally grow up with. We don't want to torture
    them, we want them to have fun, but tough fun!
    * The future doesn't need to be incremental: it will be if you start
    thinking from the present and taking it as okay. But if you go to the
    future, you don't need to think that today is okay, you can just think
    about how the future should be, and bring that back. That, you can do to
    build up your long term project, while you're thinking about what to do
    next week.
    Thoughts after watching all this:
    What would a system that lets me combine objects and have them interact
    through enactive communication look like? Write objects that act
    as processes/have their own execution context, and message-pass to
    others. Write objects that continuously interact with others based on
    proximity. Particles and fields. With either code inside, or other
    continuous systems that make up the object inside. What is the base
    interaction? Objects can't be objects, they have to be made up of a soft
    boundary separating their internal workings from the external, and they
    communicate with other same-level objects through a mechanism that is
    orthogonal to (or at least emergent from)lower-level mechanisms.
    I've always gotten along by making most of my connections myself; very
    rarely has an actual conversation with someone (vs. watching a video or
    reading an article) brought me any progress in ideas (but it has brought
    progress in networking, other people with resources to connect with).
    One exception to that is Varun at the LSA summer school. So quitting
    academia should not be that much of a problem for the development of my
    ideas, since the environments I've found up to now have not grown my
    thought process as much as my own exploration. What has helped is summer
    schools, talking with people I found myself, and networking with people
    through connections.
Commits on Sep 11, 2018
  1. ## Working vs. looking for funding

    wehlutyk committed Sep 11, 2018
    I once read that we should "fall in love with a problem, not a
    solution". Dynamic Land in a way is a solution, to the problem exposed
    in The Humane Representation of Thought. That's no secret! It just
    reminds me that I'd be better off spending time on my problem, which
    closely relates to Bret's problem, than spending time on trying to find
    funds for a Dynamic Land here. Bret has a knack for identifying the
    common thread that is woven through all the things here's interested in.
    He creates implementations that fascinate him, reacting to something
    that obviously annoys him, and after a few of those he connects them and
    finds the common theme.
    My problem revolves around "understanding each other", in all the subtle
    ways that may or may not yet be possible. I'm excited about it because
    I'm starting to glimpse how understanding works (thanks to the Enactive
    approach), and how it could be transformed, created in other ways, using
    the right interaction substrates and facilitators (of which Dynamic Land
    is one very good instance). It excites me also because it's tied with
    humane technology, which empowers and puts people control. This theme
    cuts through a number of solutions which I find fascinating: Dynamic
    Land, the Enactive approach to language and interaction, my Infuse
    project, locally produced electronics (possibly less powerful), openly
    understandable tech, control by the person.
    I've been putting a lot of effort into finding different models for
    funding a Dynamic Land of sorts, or independent research, or both and a
    team with that, but ultimately I'll get my problem solved by working on
    it, mulling over it a lot more, talking to people about it and finding
    other interested people, and having something to showcase when the
    opportunity arises (be it a foundation, people, a grant, whatever).
    This is the way to make progress both in actions and in my posing the
    problem. JM Mercier, my physics professor in prépa, always said that
    laying down the problem is by far the most important step in solving
    it. (This connects with Bret's notion of "problem finding", instead of
    "problem solving".) Right now I've been taking the academic approach of
    trying to find funding *before* dedicating myself to my problem. Things
    will flow a lot better by doing it the other way around, since I
    currently have the opportunity to do so. Once I have working things, and
    people I'm discussing this with, I can dream a lot more concretely about
    what I'd do if I had €5M over two years to fund a place and a team, and
    about where to find that money.
    This brings me to my current plan for progress:
    * make progress on my Infuse implementation
    * talk to many many more people about my problem, and Dynamic Land and
    Infuse (starting with yesterday's list of communities)
    * buy hardware to start creating Dynamic-landy things at home
Commits on Sep 10, 2018
  1. ## Funding (independent) research

    wehlutyk committed Sep 10, 2018
    I really liked the [Independent
    research]( post by Nadia
    Eghbal. It connected with a lof of options I've been reviewing lately
    in order to make research work for me. The fact that several (many?)
    famous researchers in the 18th and 19th century were either self-funded
    (wealthy) or funded by mécénat. Darwin for instance (something I had
    alread come across, but had forgotten). She talks about the challenge
    of external validation (which is harder if you don't have a famous
    institution backing you), but the real problem is funding I find.
    Another part that impacted me is YCR's emphasis on output, and not
    publications or conference participation, as a metric. Seems like it's
    very necessary, and I very much resonate with that.
    Then popped up in my exploration
    of the Future of Coding slack (see below), and wham it resonated again.
    Bret's and Alan's thought on how to fund and what worked in the past
    are very insightful, reinforcing the impression that academia as it
    exists today is not a good place to make progress in questions that can
    interest me. (I especially like the emphasis on "problem finding", not
    "problem solving"). Last night I was thinking about how research could
    be funded through a variety of channels (for me as a person), and that
    I don't need to wed myself to any particular one: consulting "gigs",
    regular grants, donations, savings after a high-pay job, whatever. But
    reading about the conditions that existed at PARC, ARPA, and the like,
    it does feel like what I have been thinking about up to now is rather
    meagre: great breakthroughs take a team, and by that I mean not only
    internet-or-publication-mediated collaboration, but a team that meets
    regularly in person and works together in a shared space ("much of
    the fundamental creative work in computing requires 8 or more people
    working together"). The options I've been thinking about don't really
    get anywhere close to what's needed to fund a team.
    This is making me think of expanding the "freelance funding" model:
    imagine a group of people who have the skills to be hired at high-pay
    jobs (maybe not every person in the group, but enough to sutain what
    comes next), and decide to take turns at funding each other. A structure
    where trust is strong enough that you feel comfortable working for
    n days at a job 1) that is not necessarily your dream job, and 2)
    for which you share the pay in order to fund the part of the team
    that's not working. Then you get to do research for a month without
    any constraints. It sounds like it would be a lot of unwanted work to
    sustain a group, and it requires a huge level of trust (that others
    won't leave after having benefited, that you feel okay working while
    others are doing what they want, etc.). But it could be worth running
    some numbers, and compare to the time spent raising funds in academia.
    The plus side is that it would be people-centred, not project-centred.
    Kind of a research coop, or a research solidarity commitment. Might be
    stupid, but I'd like to see the details.
    Seeing the amounts that Dynamic Land is looking for to fund its initial
    years is a bit daunting, but could be a good benchmark to evaluate
    these funding ideas (it includes research staff, rent, and equipment
    ## Discovered new communities around Dynamic Land ideas
    I decided to have a look at the [Future of Coding
    slack](, and something clicked.
    The conversations are pretty interesting, especially those that mention
    Dynamic Land or point to other communities. I was super happy to
    stumble upon [](, started by another
    BV fan who dreamt about (and is still dreaming about it?) working
    at HARC. There is a great thread on DL which I want to get into,
    especially since it mentions the publicity and people-interactions
    aspects of DL (versus the computing capabilities). Another community is
    [SomethingNew]( I particularly
    like this: "We thus discourage trying to agree on "one way" of doing
    things, and want to try to recognize and foster all possible views,
    values, approaches, etc.". Not one project, but a forum. I like the
    .md) and
    d) pages, and there is much more to read in that repo. Started watching
    it, so I'll get updates from there.
    This leaves me with some reading, and a number of people to write to in
    order to move on:
    * Laurent Guerby, about his story for starting and funding
    * Judicael, to see if he's interested in all this
    * tbd DL thread, touching both on the publicity of DL, and on the people
    interaction aspect
    * future of coding, simply to introduce myself
    * xmca list, to answer the replies I got about DL
    * Alix Ducros
    * iCAP
  2. ## Graph Embedding Day in La Doua

    wehlutyk committed Sep 7, 2018
    Quite a lot of interesting things happened here. I met several people
    that I'm happy to know now, including Rémy Cazabet, Aleksandar
    Bojchevski and Anton Tsitsulin. And I made it alive to the end of the
    ### About feature-network embeddings
    The talks and conversations with Aleksandar and Anton finished
    convincing me that our current take at nw2vec might by ill-posed, in
    that we're mixing representations that are already vectors that could be
    used separately (node2vec + node features). In what cases this is useful
    is not clear. It would also be more satisfying to work on principles
    and theory, i.e. a proper way of formulating problems, and not only on
    empirical implementations. Mikhail, for instance, always asks where are
    graphs a fundamental structure (other than in social networks) that is
    prior to any other measure: you don't really need graphs when you're
    converting other data into a graph structure.
    This is a recurrent theme in research I find: you start a project, and
    it turns out it's not the silver bullet it seemed to be at the
    beginning, and ambitions drastically drop all of a sudden. I think in
    this case there are several possible outcomes:
    * find out that you're wrong, i.e. it *is* a silver bullet
    * reorient the question
    * extract the value of the work that was already done in understanding
    concepts that are being used in the literature
    At any rate my exploratory work is still very useful in understanding
    how GCN & Co behave in an auto-encoder. And there's no point in deciding
    the issue of this project now, it should continue its exploration and
    include this new insight. Also, the genrative aspect of VAE could come
    in useful. This points to the importance of going back to the literature
    and revise world views regularly. Finally, it just occurred to me that
    in the case where there *is* some level of correlation between features
    and network, a combined representation should be able to deduplicate
    compared to separate representations (which might be what Marton was
    getting at from the beginning).
    ### iCAP
    Rémy also told me a bit more about the [Learning Lab
    Lyon](, which looks like the
    anti-model for Dynamic Land. In spite of that, the people who initiated
    that might be interesting to talk to. It's mostly
    [iCAP]( apparently, and I think it would be
    good to write to them. I'm starting a draft to remind me of that for my
    next Dynamic Land session.
    ### DSRT and others
    Other conversations led me to [DSRT](, a company in
    Lyon which is developping [](, i.e. making a
    living by creating bibliograpy services to researchers. The company
    was hard to find, which is not a very good sign, and I'm always turned
    down by the poor web implementation and design skills demonstrated by
    French companies, but I think it's very much worth having coffee with
    the people behind this.
    Another discovery was [IX-labs]( and [Sylvain
    Peyronnet](, a private research lab
    (whose topics I'm not interested in, but it's interesting that it
    ### General thoughts
    I'm getting more and more psyched about an Enactive project around a
    Dynamic Land-ish place. My mind was blown over the end of August, and it
    still is: the Future of Coding #26 with Glench was a good way of gaining
    perspective, and confirmed that any of this only makes sense in the
    context of a stronger research project on Dynamic Media and
    communication. This feeling is paralleled by wanting to reduce my time
    isolated in front of a computer while not relinquishing all interaction
    with the computing world. I think I'll spend some real writing/email
    time on this on Monday, it's really something I find exciting.
Commits on Sep 5, 2018
  1. ## Notes from 5/09

    wehlutyk committed Sep 5, 2018
    My plan for today had to be readjusted again, as we had a long meeting
    this morning with Marton and Jacobo. The full detail of what I do on
    nw2vec doesn't really need to go in this log, which should be more dense
    and about the high-level ideas. So I'll go back to putting that on paper
    for now.
    Academia has a way of taking over your time without warning which I'm
    really not a fan of. I didn't get around to write my Ezequiel draft, nor
    spend some time on Dynamic Land, nor start my slides for Friday (though
    today's work does go into that, it's not concrete slide-making).
    Finished listening to which was
    really interesting! Especially Glen's comments about how, if his
    previous projects succeeded big time, people would not necessarily be
    any happier in the world; a thought which made him turn to
    psychotherapy. It's great to hear that is it puts back all the projects
    I've been reading about into perspective. Food for thought.
    ## Plan for 6/09
    Spend the whole day on my slides for Friday, from the first minute until
  2. ## Log for 4/09

    wehlutyk committed Sep 4, 2018
    Gosh, I spent the whole day debugging this BlogCatalog problem, and got
    to the bottom of it. Unfortunately that means I didn't get my other
    items done, so it's bleeding into tomorrow. Still, this involved:
    * Implement the simpler Gaussian decoder without `u`, and launch a new
    BlogCatalog run
    I also made some progress on Infuse networking, as the CEO of a
    company in Paris might be interested by the project, and we could be
    talking about it more in a few days. Yay!
    ## Plan for 5/09
    * Fix my previous commit which has shifted dates
    * Look at results from the night's BlogCatalog run
    * Do some cleaning up of the nw2vec runs
    * Spend at most 30 minutes resuming the GCN reproductions
    * Concretely start my Friday presentation, making slides
    * Have lunch
    * Continue with the slides
    * At 3pm, spend one hour on the Ezequiel draft
    * At 4pm, spend the remaining time on Dynamic Land
  3. ## Notes from 3/09 morning

    wehlutyk committed Sep 3, 2018
    Going back to work majority-time on nw2vec, so I might be losing a bit
    of the context of these past two weeks' thoughts. This is the difficulty
    with multi-contexting things, as I have decided to do it, without having
    hard commitments that force to switch back and forth when it's time. So
    it's important that even if today's nw2vec plan bleeds into the
    afternoon, I spend at least an hour on my other two tasks: writng to
    Ezequiel and writing to Dynamic Land.
    It's funny how headphones have become so popular among programming jobs
    recently: everywhere I see people programming, they have big
    high-quality headphones on. It occurs to me that this lets programmers
    exercice their ear, i.e. another human skill than staring at a tiny
    rectangle and moving their fingers.
    ## Dynamic Land thoughts of the day
    Reduce the amount of backstage code as much as possible, i.e. have the
    greatest percentage of actual running code *on pages in the room*. Maybe
    run on top of STEPS? So use a few libraries as possible, if we can bake
    things denser/terser.
    ## Log for 3/09
    * Resumed nw2vec with some cleaning, rounding up issues, relaunching
    the long-running minibatch exploration -- the detail is in the repo
    * Started making a minimal plan for my presentation on Friday
    ## Plan for 4/09
    * Look at results from the night's BlogCatalog run
    * Implement the simpler Gaussian decoder without `u`, and launch a new
    BlogCatalog run
    * Spend at most 30 minutes resuming the GCN reproductions
    * Concretely start my Friday presentation, making slides
    * Have lunch
    * Continue with the slides
    * At 3pm, spend one hour on the Ezequiel draft
    * At 4pm, spend the remaining time on Dynamic Land
Commits on Aug 31, 2018
  1. ## Log for 31/08

    wehlutyk committed Aug 31, 2018
    Got through all my notes on Enactive language + DM, which is cool, and
    finally started my draft for Ezequiel in earnest.
    Couldn't resume any nw2vec computation as the grunch server is under
    Started listening to which is
    absolutely awesome! I have to finish my email to the DynamicLand people,
    especially asking what's their policy regarding other places starting
    similar experiments.
    Also got all my other tidbits done (looking at future language and math
    classes, investigating bébés nageurs, and music structures for children
    à la GAM).
    I didn't get around to doing the following:
    * Finish reading Cuffari et al. 2014
    * Start planning nw2vec presentation
    ## Plan for 3/09
    * Start on my nw2vec presentation for 7/09; spend the whole morning on
    this if necessary
    * Have lunch
    * Answer a couple other important emails
    * Continue my Ezequiel draft (I may not need to finish Cuffari et al.
    for this, but could skim through it to see if it's necessary)
    * Write to the DynamicLand people (cc Omar and Steve) to ask about their
    policy for other experiments in other places, explaining how much I got
    abouth the current system and where my research is headed. The idea is
    to have a green light for collaboration with them (or parallelisation), before
    inviting other people in Lyon on Twitter to start a group.
  2. ## Plan for 31/08

    wehlutyk committed Aug 30, 2018
    * Morning off (other meetings)
    * Have lunch
    * Re-read initial markdown notes, and full Dynamic Land a3 sheet, for
    * Start the Ezequiel draft
    * Finish reading Cuffari et al. 2014
    * Launch a nw2vec computation
    * Start planning nw2vec presentation
    Other emails will wait for next week, none of them are as urgent as the
Commits on Aug 30, 2018
  1. ## Projecting oneself into the future

    wehlutyk committed Aug 30, 2018
    There is also no such ideal in research in France I feel: I don't
    have the ideal of "becoming a researcher paid by CNRS". I could have
    the ideal of finding this or that, or revolutionasing a field, but
    "becoming a researcher in France" implies such a crummy path to get
    to such doubtful conditions, with no garantee that I would be able to
    revolutionise something on the way, that there is no way it attracts me
    like an ideal (even though there are many positive sides).
    He also mentioned a number of books or things to read or look at:
    * [La gouvernance par les
    nombres](, by Alain Supiot
    (both a book and a film)
    * [La part maudite](, an essay by Georges Bataille
    * A book on the Lawyers in the Weimar republic, and how they were used
    to develop the new law by having them work on the technical legal
    problems, but not on the political and moral problems
    * [The Schuman Declarationa](
    aration) in 1950
  2. ## Wrapping up deep dive on Dynamic Land

    wehlutyk committed Aug 29, 2018
    For the past 2 weeks I've been exploring all aspects possible of Dynamic
    Land and Bret Victor's work. It's been fascinating, and having shared
    that content with XMCA makes me hope the conversation and reflection
    will continue. I'm seriously considering starting a place like this
    * I have a large a3 page of notes about what Dynamic Land inspired for
    * I've registered to a number of meetups where I could find like-minded
    people, found thanks to the scraping of the Twitter followers of
    * I started a detailed pro-con list to inform a decision on what to do
    next: either try to continue in research or start a part-time job in the
    private sector to fund free-form research the other half of the time; or
    try to combine both (i.e. don't dropout entirely, or, if continuing in
    research, carve out time to start these projects).
    It feels very good to have done this: I feel I have infused in those
    ideas for two weeks, and I'm starting to feel that I've wrapped my head
    around the subject.
    Now it's time to step back, acknowledge what I haven't
    read, and move on with:
    * dayjob
    * a concrete project around the enactive approach to language
    * deciding (or combining) whether the next step is that project or
    diving into creating a Dynamic Land here. This is all in my a3 paper on
    what's next, and I have a bunch of people to talk with about this,
    starting with Stéphane Grumbach and Marton.
    I have a lot of thoughts about how Realtalk OS could be implemented,
    after reading a bunch of pages of code from the twitter feed. The
    important pictures of code are in `content/static/dynamicland`, along
    with the zine, BV's 2014 research agenda, and a shot of BV's library
    (supposedly), which could be good inspiration for what to read next once
    I'm through Mindstorms and Ingold's "The Perception of the Environment".
    The notes about how to implement are on a3 paper with me.
    I also stumbled upon [Michael Nielsen]( who wrote the
    [Neural Networks and Deep
    Learning]( book. There are a
    lot of good ideas and posts there, especially the stuff about ChalkTalk
    and reimplementing it with a tinge of neural networks to make it more
    evolvable. The other good point is that he doesn't write too many posts,
    so it's possible to go through his ideas in dense forms by clicking the
    10-or-so links on his home page. I like that. His work seems to be in
    the ballpark of Mindstorms, Ken Perlin (ChalkTalk), and Bret Victor. He
    has [an interesting paper on Distill](
    which illustrates that.
    ## Original plan for 29/08/2018 afternoon
    * Have a break
    * Read through BV's research agenda and jot down notes and
    criticisms. This should be very useful to kickstart merging
    these ideas with the enactive approach to language and thought.
    * Re-read my initial markdown notes for the Ezequiel draft
    * Re-read the a3 paper on Dynamic Land
    * Jot down anything else that occurrs to me at that moment
    * If time allows, start writing the Ezequiel draft from those notes
    But I bumped into Stéphane Grumbach and seized the opportunity for a
    chat with him, which lasted most of the afternoon, moving most of the
    things to spill-over for another day. See below:
    ## Long conversation with [Stéphane Grumbach](, [Clément Renaud](, and their intern (whose name I forgot)
    Talked about:
    * The future of universities, creating a university or a school
    * The publication system, Elsevier and Google Scholar moving to
    paid recommendation once OA will be legally required
    ### Spill-over for another day
    * Read through BV's research agenda and jot down notes and
    criticisms. This should be very useful to kickstart merging
    these ideas with the enactive approach to language and thought.
    * Re-read my initial markdown notes for the Ezequiel draft
    * Re-read the a3 paper on Dynamic Land
    * Jot down anything else that occurrs to me at that moment
    * If time allows, start writing the Ezequiel draft from those notes
    * Finish re-reading the Cuffari et al. 2014 article on language
    * Find out if I can use a working version of STEPS (the "universal
    document sounds like something that might be useful) so as not to have
    to read through the whole report. But don't spend more than 30 minutes
    on this.
Commits on Aug 29, 2018
  1. ## More thoughts after reading commits and thoughts of Steve Krouse

    wehlutyk committed Aug 29, 2018
    I really like the design of his website, especially the super simple
    colour palette. Compared to mine, which is a bit too grayscale, it feels
    quite ventilated.
    Second, and this is similar to what I had felt when writing a lot in
    RedNotebook, I'm probably not going to come back very often on these
    notes. If it helps me, it's to get ideas out of me, and help me with
    playing around with them, *at the moment of writing*. Writing creates a
    slower process of thought which lets me come back on things I thought a
    few seconds to a few minutes ago, and lets me make a better map of the
    possibilities that I was exploring.
  2. ## Thoughts on using git for thoughts

    wehlutyk committed Aug 29, 2018
    [Steve Krouse]( has an
    amazing way of writing down all his thoughts in git commits. Makes me
    think I could try out doing that for a while, as a replacement for
    RedNotebook. It looks more portable and exportable, though I'm not sure
    everyday usability will be better.
    He also makes pretty fine-grained plans for his days, which feels like
    an effective way of keeping track of what you spend time on, and a way
    to avoid too many distractions. Reminds me of how I tried to plan my
    days during the end of the PhD, then stopped doing it when we were in
    the chaos of starting life here in Lyon.
    Writing feels good, I'm thinking I'll try to restart this habit.
    I feel like most of my time is spent reading, and trying to manage the
    never-stopping branches of thought and more things to read that appear
    all the time. Seems like making more detailed plans for the day could
    help with this.
Commits on Jun 20, 2018
  1. feat/research: update page

    wehlutyk committed Jun 20, 2018
Commits on Apr 29, 2018
  1. fix/research: no emdash

    wehlutyk committed Apr 29, 2018
  2. style/research: phrasing

    wehlutyk committed Apr 29, 2018
  3. feat/research: add HBES talk

    wehlutyk committed Apr 29, 2018
Commits on Apr 24, 2018
  1. style/index: phrasing

    wehlutyk committed Apr 24, 2018
Commits on Apr 23, 2018
  1. style/index: phrasing

    wehlutyk committed Apr 23, 2018
  2. feat/index: update intro

    wehlutyk committed Apr 23, 2018