Mechanics Case Study: Twitter
Originally written in 2016 for the Quora design blog.
This is an extension and application of a previous post,
Why Mechanics Design.
Twitter occupies a unique and powerful position in our culture,
fostering both revolutionary change like the Arab Spring as well as
large-scale harassment campaigns like Gamergate. Both extremes are
made possible by the same product mechanics.
Twitter's stated mission is “to give everyone the power to create and
share ideas and information instantly, without barriers” . The key
distinguishing phrases here are instantly and
Here's how the product mechanics support the instantly piece:
140 character limit: This limitation means that the
content on Twitter is rapid to both read and write. Capture a
thought almost as you’re having it, and it can be read within
Chronological ranking: This means that what you see
in your timeline is the very latest in what people are writing or
sharing. If there's a conversation happening right now, you can open
Twitter and trust that you will see it, and that if you write
everyone who follows you that is checking Twitter will also see it.
Retweets: This makes it as low friction as possible
for users to rapidly spread information.
Hashtags: This allows people to follow a specific
event or conversation as its happening.
Push notifications: Twitter is more aggressive than
most when there's something happening right now that you might be
Periscope: Being able to embed Periscope streams
inside of tweets allows rapid distribution of truly live content,
and discussion around that content.
Live content integration: Embedding partner content
like NFL games directly into the app means people can both watch
live and comment live in the same space.
And here's how the mechanics support the
without barriers piece:
Public content: Most content on Twitter is
accessible to everyone, making it infinitely searchable and
Asymmetric following: Meaning both parties don't
need to mutually agree to see each other's content (contrast this
with how friending on Facebook works). This does a few things:
It lowers the friction for information to flow from one
individual to another — only one of them has to have the
interest and make the decision.
Group conversations can grow and shrink organically, rather than
requiring an all-in or all-out dynamic you see in other group
products with “membership” mechanics.
It allows for figures like celebrities and journalists to treat
Twitter as a one-way broadcast tool.
No account limitations: Pseudonyms, organizations,
publications — anyone can make an account, enabling content that
wouldn't make sense or be acceptable to share if it had to be
attached to an individual's real identity (as in Facebook, Quora).
@-mentions: This allows users to direct information
to other users, even if they aren't following each other. This
enables ad-hoc communication between strangers.
Hashtags: This allows people and content to
organize around a single topic with very little work, eliminating
the need for an existing social graph for distribution.
Moments: This creates opportunities for people to
follow events without having to build up a relevant graph or
repeatedly searching for the same event.
Embeds: Tweets can be easily embedded in other
webpages and products, allowing them to spread in a wide range of
contexts and networks.
Taken together, these mechanics make Twitter uniquely suited to
capitalize on their mission. If Facebook wanted to compete directly
with Twitter, they would likely have to adopt comparable mechanics,
which would be in tension with their existing goals like privacy and
intimacy among friends.
However, being the best at something often requires making extreme
trade-offs. These mechanics come with many downsides:
Harassment: Few limitations on account creation
combined with the multitude of ways to get people's attention makes
Twitter a potent vector for harassment and abuse. The instant
distribution also leaves no window for formal moderation or
effective community review.
Context collapse: The ability for content to be
distributed rapidly and unpredictably to many other people means
that users can never be totally confident about their audience
they're writing for.  Over time, users can become very
conservative with what they share on Twitter, effectively becoming
politicians who fear being taken out of context and so only say
Relevance: Chronological ranking means that the
first tweets you see have a lower chance of being the most
interesting tweets, as opposed to a system that takes many other
factors into account when deciding what to show you. They're
currently addressing this with ranked sections mixed in with
chronological tweets, with some success.
Onboarding curve: Many of these positive dynamics
depend on users having already built up good follow graphs, an
understanding of unique concepts like retweets, and checking at
times when interesting things are happening — all of which make it
hard for new people to get consistent value out of the product.
Conversations: Due to asymmetric following, reply
chains can be jumbles of different social graphs which makes them
very hard to parse and difficult to share.
Misinformation: The rapid spread of information
without any checks for quality or accuracy means rumors and lies can
go viral fast, and later corrections/retractions naturally tend not
to get as much attention.
Long-form: It's possible to string tweets together
to create long-form writing but it feels awkward and when individual
pieces get shared, context gets lost.
Evergreen: Old tweets that are still good tend to
disappear forever, due to the chronological ranking.
Attribution: The glut of anonymous accounts means
there's little accountability for theft of content.
Given how directly tied these issues are to the core mechanics of the
product, Twitter will have a tough time trying to address these
without compromising the many positive dynamics.
Bloomberg: The Future of Twitter: Q&A with Jack Dorsey