Configuring the Plugin

This page covers the options available in Settings → Respectify. If you haven't connected the plugin yet, see Installation first.

The settings page has three sections:

• API Credentials: your Email and API Key. Click Test at any time to check they work; the Account Status panel then shows your credit balance. Every Respectify feature is available on every account, and usage is billed against your credit. Click Save Changes after a successful test.
• Comment Handling: what Respectify checks, and what happens to a comment that falls short. Most of this page is about this section.
• Rarely Used Settings: hidden until you click Display rarely used settings. See Advanced Settings.

What Respectify Checks

There are four checks, each with its own checkbox in the Comment Handling section, and all of them are on by default:

• Assess Comment Health
• Check Topic Relevance (off-topic comments and undesired topics)
• Check for Dogwhistles
• Check for Spam

Unticking a checkbox means that check plays no part in what happens to a comment.

Comment Health

This is the core of what Respectify does: analysing comments to assess the quality of the reply, neutrally. A "healthy" comment is one that contributes to the conversation: it engages with the topic, understands who it's replying to, and communicates thoughtfully. Someone can disagree strongly and still do so in a healthy way; health is about how views are expressed, not what those views are.

Tick Assess Comment Health to turn it on. The Comment Health setting below it then has one drop-down per item: nine kinds of problem, plus the comment's quality.

Item	Levels offered
Threats of violence or harm	Standard, Strict, Zero tolerance
Identity attacks and slurs	Standard, Strict, Zero tolerance
Sexual content	Standard, Strict, Zero tolerance
Harassment	Standard, Strict, Zero tolerance
Insults and put-downs	Off, Permissive, Relaxed, Standard, Strict, Zero tolerance
Profanity	Off, Permissive, Relaxed, Standard, Strict, Zero tolerance
Dismissive or sarcastic tone	Off, Permissive, Relaxed, Standard, Strict, Zero tolerance
Logical fallacies	Off, Permissive, Relaxed, Standard, Strict, Zero tolerance
Low effort posts	Off, Permissive, Relaxed, Standard, Strict, Zero tolerance
Comment quality	Off, Permissive, Relaxed, Standard, Strict, Zero tolerance

Every item defaults to Standard. For the problem items, the level says how much of that problem your site accepts:

• Off: not checked.
• Permissive: filters only the very worst. Only extreme problems (such as a slur combined with a threat, or a credible threat against a named person) are sent back.
• Relaxed: severe problems are sent back; minor and clear ones are accepted.
• Standard (recommended): minor problems, such as a stray swear word or a mild jab, are accepted; clear ones are sent back.
• Strict: any problem is sent back.
• Zero tolerance: any problem is sent back, and any hint of one gets Respectify's full review.

For Comment quality, the level sets the quality a comment needs: Permissive and Relaxed accept most comments, Standard accepts normal good-quality comments, and Strict and Zero tolerance accept only outstanding ones. Off means quality is not judged.

Threats, identity attacks, sexual content and harassment can be made stricter, but not relaxed or turned off, so their drop-downs offer only Standard, Strict and Zero tolerance. At every one of those levels, even a minor instance is sent back. If a disallowed level is ever submitted for one of them, it is not saved: that item keeps its previous level and the settings page shows an error.

When a comment falls short of your site's levels, it is sent back to the commenter with feedback so they can revise it. Respectify makes that call for each comment (including when a comment is highly toxic overall), and the plugin follows it. Comment health never deletes a comment.

The plugin sends all ten levels with every comment it checks. Nothing is saved on your API key, so the levels on this page are the ones that apply. See Comment Health for exactly what each level accepts, the item scores, and the severity of each finding.

Topic Relevance

Sometimes comment sections drift off-topic. With Check Topic Relevance ticked, Respectify checks whether comments relate to what the post is actually about. The same checkbox also turns on Undesired Topics, which lets you keep specific topics out of your comments: useful for handling known bad-actor themes you've seen in your community.

Off Topic Comments: choose what happens to a comment that doesn't relate to the post:

• Are Okay (the default): off-topic comments are published, for freely flowing conversation
• Delete
• Give Opportunity to Revise: the comment is sent back with a note that it doesn't seem to relate to the article

Undesired Topics: list the topics you don't want to see in your comments, one per line. Be descriptive: instead of "politics", try "US politics after 1970". This avoids over-blocking legitimate discussion.

Then choose when a comment counts as mentioning one of them:

• Any mention at all (the default)
• The comment can mention something, but only a little bit: use the slider to set how much of a comment may be about an undesired topic, from 10% to 100% (10% by default)

And what happens to it: Give Opportunity to Revise (the default) or Delete.

Undesired topics only apply when the list has at least one topic in it.

Dogwhistle Detection

Dogwhistles are coded language that appears innocent but carries hidden meanings, often used to signal harmful views while maintaining deniability. With Check for Dogwhistles ticked, Respectify looks for them.

Respectify detects dogwhistles automatically, but if this is relevant to your community you can help it understand your community's context by listing, one per line:

• Sensitive Topics: topics in your community where dogwhistles might appear, such as nationalism, racial issues, LGBTQ+ rights or conspiracy theories
• Dogwhistle Examples: specific coded language you've noticed in your community's comments

When Dogwhistles Detected sets what happens to the comment:

• Give Opportunity to Revise (the default): treats the commenter as writing in good faith. They see that their comment contains language that may carry unintended meanings, with Respectify's reasoning and, where available, the terms it detected.
• Delete
• Allow (Publish): publishes the comment anyway (though that defeats the purpose)

See Dogwhistle Detection for more on how detection works.

Spam Detection

With Check for Spam ticked, Respectify checks for spam: promotional content, bot-generated text, link spam and the like. Every comment gets one of three outcomes:

• Spam: handled by the How to Handle Spam setting. Delete (the default) refuses the comment. Give Opportunity to Revise sends it back like any other comment that needs revision, though this is rarely useful for spam. A spam comment is not checked for anything else.
• Suspected spam: a comment that might be spam, but is not certainly spam, is never deleted. It is sent back to the commenter with a suggestion, such as "This looks like spam. How about removing the link?" If the comment also has other feedback, such as comment health feedback, the commenter sees both together. A real person can fix the comment and post again; a spammer won't bother. This happens whatever How to Handle Spam is set to, and there is no setting for it.
• Not spam: the comment carries on to your other checks as normal. A comment that Respectify could not screen for spam is treated as not spam, rather than being deleted without a verdict.

See Spam Detection for how each outcome is decided.

How This All Works in Practice

Here's what happens when someone submits a comment on your site:

1. Before the comment is posted, it's sent to Respectify, together with the checks you have turned on and your comment health levels. When comment health, topic relevance or dogwhistle checks are on, the first comment on a post also sends Respectify the post's content, so comments can be judged against what the post is about. The plugin remembers this for the post, and sends it again only when needed, such as after you change your email or API key.
2. Each check that finds a problem asks for an outcome: the outcome you chose in its setting, or, for comment health and suspected spam, a revision. The most severe outcome wins:

   • Delete it: if any check asks for deletion. The comment is not saved, and the commenter sees a message explaining why.

   • Request revision: otherwise, if any check asks for a revision. The comment is not saved yet: the commenter sees Respectify's feedback explaining how their comment could be improved, and can edit it and submit it again.

     Over time, users who read this feedback become more aware of how they communicate: thinking more clearly, expressing themselves better. Respectify is a tool for you to moderate your website, yes, but our purpose in the world and which we hope you also find worthwhile is larger: one of edification, not moderation per se.

   • Publish it: if no check finds a problem, or every check that did is set to allow it. The comment appears immediately.

The commenter sees feedback for the most important issue: spam first, then a suspected-spam suggestion (alongside any other feedback), then comment health, dogwhistles, off-topic comments and finally undesired topics. Comment health feedback quotes the problem phrases with suggested rewrites where Respectify has them.

This happens in real time. There's no separate moderation queue to check. The one exception is when Respectify can't be reached or returns an error: then the comment is held in your WordPress moderation queue rather than lost, the commenter is told it couldn't be processed, and you see a notice in your WordPress admin (and get an email, at most once a day) so you can check your settings.

Each comment the plugin saves keeps a summary of its comment health result (each item's score, the minimum your level accepts, and whether it was acceptable) in the comment's respectify_health meta. It holds no comment text.

Advanced Settings

Click Display rarely used settings to see the Base URL and API Version settings, for a custom Respectify server. Leave both blank for the normal Respectify service. You won't need these unless Respectify support has specifically asked you to change them (for example, to access a beta version).