Open Letter to Twitter re Conversational Design

aims

collective intelligence

Twitter is about "what's happening?" - ahead of newspapers

not just about now — QTs & search allow deep history

Those are the sorts of metrics that should be twitter's priorities. I know they're hard to measure. This thing has the potential to be the best dating app (for some people) and a way better place for finding your dream job than linkedin (for most people).

There were people who were consistently at the forefront of covid news in 2020 on twitter, often in ways that contradicted the what the WHO was saying at the time but later turned out to be correct.

good summary by Visa

First, I want to express profound appreciation for:

Visakan Veerasamy, one of the people at the very forefront of what's possible with this platform, expressed in 2019 that quote tweets are a killer feature and needed more prominance—which, impressively, Twitter recognized!

> Quote-tweeting is the ultimate value proposition of the internet.

To be able to mark up every concept, every saying, every image—with endless levels of commentary and meta-commentary...

Is the great Talmudic dream.

Currently there are 2 modes of QT:

(Links to other websites are currently somewhere in between, where the link previews when you paste them but they still take up characters)

The obvious direction to go is for QT functionality to become more like the first kind, even when you're adding them mid-thread. This continues Twitter's long-time pattern of making leading-edge behaviors easy for everyone to do. When writing a long thread with a lot of QTs, it's easy to forget exactly which tweets you're quoting in each tweet.

On one level, the preview feature is more important than getting the full 280 characters, but removing QTs from the tweet length makes space for the possibility of something much more powerful: multiquote tweet

Each tweet is an atomic unit of thought—a digital utterance. That's not to say each tweet contains only one idea, but that each tweet is sort of self-contained. Context is important, but when someone likes or retweets or quotes a tweet, the whole tweet is what they are interacting with.

When someone makes a quote-tweet, they're essentially saying "I have some thoughts I'd like to share, that relate to the tweet here". This might be a critique of the quoted tweet/thread, or it might be using the quoted material as a sort of footnote of supportive evidence or further reading or ironic contrast. This meta-commentary is very powerful, whether it's used by someone reflected "I think what I really meant to say here was" or someone framing a thread they just read as an answer to a particular question they and their followers might care about.

Currently, however, it's impossible to QT two or more tweets at once. This means that in the natural ontology of Twitter, there is no way to properly compare or contrast or relate different thoughts. This contributes, I think, to the divergent quality of thinking on Twitter: the structure of the app prohibits expressing convergent thoughts. You can have a meta-thread that pulls a bunch of things together, but each tweet in that thread is still only referencing one other tweet. You can use screenshots, but then all context & interlinking & copy-pastability is destroyed.

The amount of utterances that need to connect two other utterances is huge. Thoughts shaped like:

It's not 100% clear to me what the ideal composer for this would look like, but we can get some inspiration from how pasting in images produces first one image, then two side-by side, then 1 & 2, then a 2×2 grid. From my vantage point, double-QT, side by side, is an obvious win. It seems likely to me that up to 4 would be good, aside from the added challenge of designing an intuitive interface for it.

But here's an interface I think would be workable-enough, which I thought of in an hour. So it's surely doable.

Follower management tools

Friendcatcher / mutual finder

Why does search matter?

How do we take it to the next level?

Why do notifications matter?

How can we transform the game here?

I would imagine that the Twitter team could come up with lots of great improvements here, but there's been no attention to it. Still, this matters.

Ability to 2-click DM everyone in a thread.

Packy's article has some good stuff on this.

This is not high-priority at all but it would just take a few minutes: add a 5th option so we can do 2-by-2s with a "show answer" button

Related to polls and to views and to other things—a way to aggregate

< stuff about >

thoughts are the core unit—not people,

all accounts are fundamentally the same type of entity whether celebs or fans, creators or consumers, bots or companies

multiple identities are encouraged,

and whistleblowers can anonymously share critical information with the world.

multiquote tweet: up to 4 QTs, like with images

ability to see which tweets you've seen before & search by tweets you've seen before

detect QT-dunks and dramatically deemphasize them in the algorithm

R > why, when I land on the timeline, do I just see random stuff?

friends of theirs

people in their field

non-famous people they admire like bloggers or small-time authors

replying to other users (effectively) is the main way that small accounts get any attention

detect the kinds of tweets a user likes to reply to and show them more of those

say the obvious

new users often awkwardly ask questions in the way they see bigger accounts doing

R > the collective intelligence stuff

R > timelines are left to machine learning to maximize click through rate

requires monetizing not through ads

monetize the API ??

conversation converging

R > do you have a story / use-case / anecdote of when multiquote tweet would be a gamechanger?

instead, what you need to do is incentivize & create crossover hits—if something is unusually liked/RT'd by people from different groups that usually don't like/RT each others' tweets, that's a sign that someone has successfully written a tweet that connects those groups (unless it's just a cute cat or something, but you could detect this in most cases)

The top priority is to untangle your incentives. As I understand it, you had to cripple the Twitter API back in <year> because it hurt ad revenue. My sense is that twitter wants to be a platform, and so maybe the API is actually where to do the monetization! Twitter itself would be free, but if you want to link your account(s) with more than N platform tools, you have to pay a subscription. Part of that could even go to the apps—you could be a whole app store.

The other priority is to pay attention to what's working. Many people, on and off twitter, think of it as a shouting fest, and parts of it are. And... on the same app, with the same features and some different cultural assumptions, there are pockets where people are falling in love and starting businesses together and sharing their learning processes with each other. Those are the sorts of metrics that should be twitter's priorities. I know they're hard to measure. This thing has the potential to be the best dating app (for some people) and way better than linkedin (for most people)

(This was scheduled on 2021-09-26 as a showtime for 2021-10-26. I sure hope I write more than what's above before this goes live, because I don't think what's written above is actually very useful, and I do think I have something useful to say.)