Restoring the Web's original promise for meaningful connection

We curate and research evidence-based design solutions to bring out the best in human nature online.

Browse interventions below
Slack logo

Join our Slack!

We're building a community of academics connecting with designers.
Sign up now

Browse interventions

How do our ratings work?
Suggest intervention
A mockup of an intervention in which all posts have a delay built into them.

Delay-by-Default Posting

Gives users a chance to reconsider their words before they're public

Inference

A mockup of a prompt that reminds users that "Headlines Don't Tell The Full Story".

Disallow Sharing of Unclicked Links

Cuts down on the spread of misinformation

Tentative

A mockup of an intervention, with a smirking figurine asking "You didn't get hacked did you?" as though to suggest that, in fact, they're well aware that the "hacked" person was actually being a troll.

Email to trolls suggesting their account's been hacked

Cuts down on trolling behavior without shaming users

Inference

An expanded modal box, which conveys the concept of listening to a recording of someone else's voice.

Give audio option for user input

Humanizes users in others' eyes

Inference

A mockup of a prompt asking users to rate the accuracy of a headline on a scale of 1-to-5.

Headline Rating Interstitial

Reduce sharing of misinformation

Convincing

A mockup of an intervention. An image is blurred out behind a warning, which says the image in question could be questionable or inflammatory.

Hide edge-case content behind warning

Create middle ground between promoting content and banning it

Inference

A mockup of a comment. Next to the user's profile picture, there's a rounded token that calls attention to its being a "Consensus Post".

Highlight Consensus Posts

Depolarizes conversations

Inference

A mockup illustrating two posts in a feed: one is labelled as "misleading" whereas the one beneath it has a less severe label of "Stay informed".

Label misleading content; add links to reliable, related content

Reduces spread of true but deliberately misleading information

Convincing

A user types out 'Romani Evnt Domus' and a moderator corrects his Latin grammar. This is an easter egg from a Monty Python film.

Let mods suggest edits to users

Cut down unnecessary tension and confusion

Inference

A mockup of an over-the-top news headline that reads "ALIENS!". In front of it is a reminder: "If shocking claims in a headline sound unbelievable, then they probably are."

List tips for checking accuracy of shared headlines

Reduces spread of mis- and disinformation

Convincing

A prompt telling users that trolls may be there, but they are uncommon.

Message that trolls are present but outnumbered

Make new users more comfortable in community and more resilient to trolling

Likely

A mockup of a phone screen with this intervention, explaining that "You may encounter unverified claims on this topic." The icon is of an ear on a shield, representing the idea of 'protecting your ears from rumors.'

Messages prebunking misinformation on controversial topics

Innoculates users against misinformation

Validated

A mockup of a prompt telling a user that "Your Comment Is Likely To Be Hurtful To Others" with an option to either edit the post or post it anyway.

Preliminary Flagging Before Posting

Reduces online harrassment

Convincing

A commenter posts a lot of content, to which an authoritative finger is raised in objection to that commenter's not letting others contribute.

Prompt suggesting letting others post

Opens up the discussion to more people

Inference

A mockup of an intervention showing a Twitter comment, but it has a new button along the bottom that shows how many people have viewed the tweet.

Show Content Reach (# of Views)

More conscientious posts, reduced misconceptions of people unlike themselves

Emergent

A wireframe of how a slowdown intervention would, with a timer, show users how long they must wait to post again.

Slowdown Modes

Slows users down so their followup comments are more thoughtful

Inference

A mockup of a react button. The platform's 'Thanks' icon is emphasized.

Thank You Button

Increases retention & contagious prosocial behaviors

Likely

A mockup of a prompt telling users that their posts will be subject to peer review. The user is invited to either edit their post or proceed to posting it.

Warning that Comments Will be Reviewed by Peers Before Publication

Reduces toxicity of comments

Emergent

A mockup of a comment. It has a prompt on top of it that says 'heads up, topics like this can get heated'.

Warning that topic is sensitive

Reduces likelihood of sharing triggering content

Inference

Validated

We reserve Validated Grades for interventions that are both peer-reviewed and replicated. This is our highest evidentiary grade.

A mockup of a phone screen with this intervention, explaining that "You may encounter unverified claims on this topic." The icon is of an ear on a shield, representing the idea of 'protecting your ears from rumors.'

Messages prebunking misinformation on controversial topics

Innoculates users against misinformation

Validated

Convincing

Convincing Grades are for interventions with peer-reviewed evidence. They have yet to face attempted replication. This is our penultimate grade.

A mockup of a prompt asking users to rate the accuracy of a headline on a scale of 1-to-5.

Headline Rating Interstitial

Reduce sharing of misinformation

Convincing
A mockup illustrating two posts in a feed: one is labelled as "misleading" whereas the one beneath it has a less severe label of "Stay informed".

Label misleading content; add links to reliable, related content

Reduces spread of true but deliberately misleading information

Convincing
A mockup of an over-the-top news headline that reads "ALIENS!". In front of it is a reminder: "If shocking claims in a headline sound unbelievable, then they probably are."

List tips for checking accuracy of shared headlines

Reduces spread of mis- and disinformation

Convincing
A mockup of a prompt telling a user that "Your Comment Is Likely To Be Hurtful To Others" with an option to either edit the post or post it anyway.

Preliminary Flagging Before Posting

Reduces online harrassment

Convincing

Likely

Likely Grades are for interventions with preprinted evidence. Preprints are academic manuscripts awaiting formal peer review.

A prompt telling users that trolls may be there, but they are uncommon.

Message that trolls are present but outnumbered

Make new users more comfortable in community and more resilient to trolling

Likely
A mockup of a react button. The platform's 'Thanks' icon is emphasized.

Thank You Button

Increases retention & contagious prosocial behaviors

Likely

Tentative

Tentative Grades occur when randomized controls or independence from vested interests are absent. It also happens to Convincing interventions if they fail to replicate.

A mockup of a prompt that reminds users that "Headlines Don't Tell The Full Story".

Disallow Sharing of Unclicked Links

Cuts down on the spread of misinformation

Tentative

Emergent

Emergent Grades are for interventions with only qualitative evidence.

Inference

Inference Grades are for interventions that lack direct evidence. Yet analogous studies, expert opinions, or first principles, suggest that they might work.

Prompt suggesting letting others post

Opens up the discussion to more people

Inference

Let mods suggest edits to users

Cut down unnecessary tension and confusion

Inference

Warning that topic is sensitive

Reduces likelihood of sharing triggering content

Inference

Slowdown Modes

Slows users down so their followup comments are more thoughtful

Inference

Give audio option for user input

Humanizes users in others' eyes

Inference

Email to trolls suggesting their account's been hacked

Cuts down on trolling behavior without shaming users

Inference

Highlight Consensus Posts

Depolarizes conversations

Inference

Delay-by-Default Posting

Gives users a chance to reconsider their words before they're public

Inference

Hide edge-case content behind warning

Create middle ground between promoting content and banning it

Inference

Have an intervention for our library?

Suggest intervention