todo
No content on this page itself — see linked references below.
9 linked references
★ Redesigning Twitter
- Elaborate on this thought:
- "But there are like 4 non-existing relationships tweets could have" (beyond replies & QTs)
- Attempting to list the relationships tweets could have:
- ☑ Reply → linear flow; after that tweet, read this one
- ☑ QT → commentary, before that tweet, read this one
- ☒
- Attempt 2 - focusing on the meaning, not the feature or app affordance
(noticing after writing a fair bit here... I appear to implicitly use numbers for reply-esque relationships and x for QT-esque relationships)- tweets A1,A2,...,A10 form a coherent train of thought (= thread of replies)
- tweet B' is a footnote to tweet B (where B is probably in a longer thread, or in a conversation)
- sort of like a footnote, this could be an elaboration, a definition, a citation, or all sorts of other things
- (this can be sort of done now, by tweeting a whole thread then later add a parenthetical. I have a few examples, maybe in my first McGilchrist thread, or in that thread where I was quoting Romeo talking about woo)
- what can't be done now is clearly indicating the existence of such a thing. it shows up just as the fact that that tweet has 2+ replies rather than 1
- this also is super borky in relation to thread ownership
- tweet C2 is a tangent from tweet C1
- this is very fucking similar to the B parenthetical above. maybe they're identical, but I wanted to explore if they would need different interfaces
- one way that this currently gets implemented, actually, is that C2 is a new thread that QTs C1
- and this would actually be basically fine, if it weren't for the fact that there's no natural way to get from C1 to C2!
- related concept - tweet C2* is a parenthetical after tweet C2 before tweet C3
- the distinguishing feature of a parenthetical being that it can be omitted without losing meaning
- one thing I'm wanting to note is that you could very visually distinctly have a twitter thread
- tweet D (+thread) is a new thought, inspired by tweet Dx (ie Dx is the context for thought D)
- again, not sure how this differs from the above, but something about "this thread was inspired by this tweet" seems like a thing
- OH! I know part of how this is different, maybe!
- it's for when you want to have your own top-level thread that you own (though this concept is confused; better would just be the ability to reply & hoist)
- and yet you're also implicitly replying to somebody else (including your past self)
- blogs distinguish this via comments vs "a post on my own blog as a reply"
- Medium handles this in a curious way, hm... #2integrate
- so you might want to actually say "first, read this tweet" or "first, read this storm" and this gets funny because QT-embeds are below the main tweet
- which on one level is fine, you make a tweet that says "first, read this, then keep reading for my thoughts <embed>" then another series of tweets with thoughts
- tweet E embeds tweet Ex in another thread to make a point
- (difference between embed-with-commentary and embed-as-whole-node in thread)
- tweet F* is a correction/update to tweet F
- there are a whole bunch of ways to implement this, that don't currently exist
- people have talked about it; I probably have links to this elsewhere in this workflowy
- tweet G brings tweets Ga,Gb,...,Gg together
- basically this functions as "this tweet is a reply to ALL of these N tweets"
- if you're only tracking replies, this is where twitter becomes a graph, not a tree
- here's a crappily implemented example with links & screenshots
- here's another related thing, although less awkward here since 3/4 of the things being juxtaposed were already book excerpts
- what is the point of this?
- tweet G would show up in the replies list of Ga, Gb, etc
- if you were looking at tweet G, you could see tweets Ga, Gb, etc
- on reflection, there is one sense in which this could be achieved by something like
- you can QT multiple tweets at once
- (maybe this needs a whole new name, call it a collection or something)
- QTs finally have backlinks
- but that doesn't quite capture what I'm looking at here, & it relates to this implicit framing:
- a reply is a way of saying "now that you've read that, read this"
- a QT is a way of saying "now that you've read this, read that"
- (Have now made this into a thread: twitter.com/Malcolm_Ocean/status/1151949360621608969)
- Also this thread is relevant
- tweet-tree (collapsible & zoomable)
- H1
- H1.1
- H1.2
- H2
- H2.1
- H2.1.1
- H2.1.2
- [structurally this is what workflowy is except its nodes have some differences from tweets]
- diffs: in twitter
- each node has one clear author
- each node is non-editable and has a timestamp
- nodes may embed various other things including other nodes
- 280char limit, line breaks allowed, no main-vs-note distinction
- [stuff about prose, now in malcolmocean roam]
roamresearch.com/#/app/malcolmocean/page/fDzXwsNW3 - Got some inspiration while reading this article by KK about books and thinking about what prose is. One thing it inherently is is linear.
kk.org/thetechnium/what-books-will/- Multimedia prose is still prose in this sense.
- ie text with stuff embedded (stuff = images, tweets, threads, videos, polls, flashcards, &+++)
- because there's still a clear order in which to read it
- also most poems are basically prose in the sense I'm using it here
- (although something with a repeating refrain could be understood as a nonlinear piece of text!)
- which is then making me think of musical notation, repeat signs, and da capo
- Here's an interesting meta-piece:
- I find myself wanting to add the "ie text with stuff embedded" as a parenthetical after "multimedia prose"
- but that becomes so long it ruins the clarity of the simple sentence "Multimedia prose is still prose in this sense."
- which then makes me want a feature that lets me insert a caret or something ^ and have the parenthetical float outside the linear flow of the text
- tooltip-notes (like WBW & meaningness) kind of achieve this role, but it's annoying to have to actually do the hover, so in practice people often don't do it
- So this is part of what is powerful about a thread structure. It occupies an amazing space between pure prose (100% linear) and fucking graphs that are a mess and really hard to navigate.
- ie it has a default linearity to it, meaning you're still sharing a coherent train of thought
- but then also because everything is nodular
- individual links in that chain can be reused in other thought-trains later
- this reusability is what comes from the magic of the embed/QT, which by default a messy-ass graph doesn't have either
- replies/comments can branch off from specific nodes
- Then, there are multiple ways to relate prose & nonlinearity.
- ONE is the wikipedia model. Hypertext books like meaningness are other examples. Here, there is long-form prose, just like a book or newsletter, except individual words can be linked to other long-form prose items. Some deeplinking is possible via section headings.
- TWO is the google docs model. There is one linear flow of prose, and ALSO there's commentary on the side. This commentary can tag people or even assign some edit to someone.
- THREE is the twitter model. Replies & embeds (both other tweets as QTs and media like youtube) allow short-form prose to compose a series of nodes. Importantly, the reification* of threading means that this isn't just a crazy graph.
- (aside: better word than reification? affordance? function, groove, default, weight... I'm trying to point at how by replying tweets together, they kind of stick as a thread)
- the tweets don't change order... I can link to a thread and know that it'll continue to be the thread that it is
- the point is, that I can make a thread and say "follow these thoughts from start to end" and then nothing can interrupt that thread. people can comment on any node in the thread, but that doesn't break the flow
- this breaks down, however, because sometimes A makes a thread and B makes a sort of thread-of-replies but then if A replies to the first message in B's thread, then it breaks the flow of what B was saying in the main view (because A owns the thread) and there's no way to get it back. This makes it really hard to clearly see what's happening in dialogues.
- ideas for ways to improve this (all of these could be implemented poorly)
- give the Reply UI a threading feature and make those subthreads uninterruptible (ie owned by their creator)
- whenever someone replies to themselves, treat that differently than dialogue
- value order-of-reply somewhat more than thread ownership
- ~ when there are replies, show the avatars & first lines of those replies under the main tweet
- see how many total replies are under a given
workflowy.com/#/c56379c9c080
- FOUR? is tables, eg like I have my blog posts in airtable.
- not sure how this relates exactly but it feels important
- actually, maybe even a better example than my blog posts is the harvesting models/incidents etc table that benj & I worked on summer '16
- FIVE? is progressive summarization. A cool UI to try would be one where you can actually collapse-away the different layers
- eg in this image, being able to hide what's not highlighted; to somehow more densely interlink the summary & remix with the original i0.wp.com/cdn-images-1.medium.com/max/800/1*cs4WrYZDSERqIfHJBZoflQ.jpe
- in a twitter context, you could imagine this as the ability to flag tweets and then collapse thread thread to only the flagged tweets visible
- you can already do this just by pulling them out, but it's ugly, partly because of truncation
twitter.com/Malcolm_Ocean/status/1151938479074549760
- oh shit! this is actually kind of how workflowy works!
- it doesn't do ProgSum and its collapsing isn't like sublime where you can instantly collapse to a certain level
- but it does have the ability to type something that reads linearly like prose and yet also can be collapsed to view its structure
- SEVEN: choose your own adventure
- largely is a thing with fiction, but the format can be (and surely has been) adapted for nonfiction
- books have a tiny bit of nonlinearity
- in the form of the table of contents and the index. and footnotes
- but these are all terribly implemented!
- kindle footnotes should use a zoom gesture or something and should shot up as a modal atop the page, not jump me to the fucking end of the book like it's made of paper and needs to be stored at the end to be kept out of the way)
- Metamitya notes: knovigator would do this
- "it would open a footnote as its own first-class citizen"
- this seems pretty great to me
- Some tweets about this:
twitter.com/Malcolm_Ocean/status/1156423522836766721- What I really want to have is a few really good editors for:
- trees (workflowy!)
- tables (airtable!)
- prose (google docs!)
- threads (twitter!)
- ??
- PLUS the ability to embed any of these anywhere within any of the others. Boom.
- So I could write a blog post within workflowy, or embed a workflowy tree in a twitter thread, or make a twitter thread in an airtable, or have a table within a google doc.
- @NotionHQ seems like maybe it's trying to be this but so far it's mediocre at trees/prose/tables and doesn't support threads at all.
- aside: there's also what relation a tweet and a user can have
- which is currently
- from:USER
- to:USER
- others that could exist
- mentions:USER
- (I think this is currently confused with to:, but it's conceptually different. there is an obvious obvious difference between mentioning someone and addressing them)
- I'm not sure what this would actually look like; maybe too complex to implement?
- quotes:USER
- ??? feels like there could be more but maybe it's that simple
Examples on various levels of abstraction › Stuff that exists to try out (hacky tools & more) › Existing power tools › quicksearches for tweets from self/friends/following
- #todo import from twitter
- #todo make tweetstorm about how I avoid distractions on Twitter
- userstyles
- tweetonly
- hide notification count
- block promoted tweets
- Focus Twitter Permalink
.PermalinkOverlay-modal {
top: 0 !important;
left: 0 !important;
margin-left: 0 !important;
right: 0 !important;
width: 100%;
height: 100%;
}
.PermalinkOverlay-with-background {
background: black;
}
div.PermalinkOverlay-content {
position: absolute;
top: 60px !important;
left: 50% !important;
width: 640px !important;
margin-left: -320px !important;
right: auto !important;
}
not all positive-sum outcomes are win-wins \ omniwin
- #todo QT "BATNAs are underrated"
omniwin training game
- #todo get list of words used
- stretched
slack2roam
Thinking out loud May 19th, 2020 as I work on this. › First step: explore slack export data and see what it loo… › And each JSON file is just an array... › the rich text formatting is... complex! it's basically su…
- #todo check if someone's written something to turn that into HTML?