Respectify is an AI-powered comment moderation tool positioned to enable 'respectful and relevant discussions' in online communities. The landing page primarily engages Articles 19-20 (free expression and assembly) and 26 (education/accessibility) through its design and functionality. The product demonstrates structural commitment to free expression through API transparency and demo access while attempting to balance speech rights with community moderation responsibilities.
The sample prompt I was given was "Is Die Hard a Christmas movie?"
"Of course it is!" got an 80% certainty "off-topic" mark.
When I elaborated that it occurs at a Christmas party, it said this:
"Dogwhistles detected (confidence 80%): This comment seems innocuous, but the phrasing 'Christmas party' may be an underhanded reference to Christian themes, especially among discussions that might dismiss or attack secular or diverse holiday celebrations. This kind of language can subtly imply exclusion or preference for Christian traditions over others, which can marginalize those who celebrate different traditions."
Not a great first experience.
I've seen the trend on Facebook/Instagram to say "unalived" instead of "killed" or "cupcakes" instead of "vaccines" and suspect humans are long gonna be cleverer than these sorts of content filtering attempts, with language getting deeply weird as a side-effect.
edit: I would also note that it says "Referring to others as 'horrible people' is disrespectful and diminishes the possibility of a respectful discussion. It positions certain individuals as entirely negative, which can alienate others and shut down dialogue.", if I feed it your post, too.
What I've seen, the difference between spam detected or not is https://www before the domain name.
Here is an example of successful passing of all checks:
> Published
This comment passes all checks and would be published.
Score: 5/5 | Not spam | On-topic: Yes | No dogwhistles detected (confidence: 100%)
Can confirm. We hit this exact issue running tirreno www.tirreno.com (open-source fraud detection) on Windows ARM — libraries were auto-selecting AVX2 through emulation and batch scoring was measurably slower than just forcing SSE2. The 256-bit ops get split under the emulation layer and the overhead adds up fast in tight loops.
Pinned SSE2 for those builds. Counterintuitive but throughput went up.
This thing seems to be more about enforcing a political PoV than about avoiding logical fallacies.
All my attempts to comment on the UBI article (and not supporting UBI) said my comment was a dogwhistle, and/or had an overly negative tone. This topic, of all things, is absolutely worthy to challenge and debate.
Using this would have the effect of creating an echo chamber, where people who stay never benefit from having their ideas challenged.
Love the effort here, been thinking about what this kind of tool might look like for a while. Something like this coupled with better prosocial affordances in the medium will do a lot to improve discourse online. I wrote up one a while back [1] but things like that are only a small part of a much bigger picture.
The overall problem needs to be tackled from all angles - poster pre-post self-awareness (like respecify but shown to users before posting), reader affordances to reflect back to poster their behavior (and determine if things may be appropriate in context vs just a universal 'dont say mean words'), after-post poster tools to catch mistakes (like above), platform capabilities like respectify that define rules of play and foster a enjoyable social environment that let us play infinite games, and a broader social context that determine the values that drive all of these.
This passes your checks, but a human moderator would flag it:
> My favorite movie is die hard. I think it's a Christmas movie. But, honestly, we shouldn't have to wait until Christmas to watch you die hard. We should be able to watch that any day of the week :)
Seems to catch various other cases though. Cool tool.
I was hoping 'respectify' could mean respect for the users.
This is a very important problem space. Maybe the most important today - we desprately need a digital third place that isn't awful. But I think these attempts are misled.
The core issue seems to be that we want our communities to be infinite. Why? Well, because there is currently no way to solve the community discoverability problem without being the massive thing. But that is the issue to solve.
We need a lot of Dunbar's number sized communities. Those communities allow for 'skin in the game' where reputation matters. And maybe a fractal sort of way for those communities to share between them.
The problem is in the discoverability and in a gate keeping that is porous enough to give people a chance.
Solve that, and you solve the the third place problem we have currently. I don't have a solution but I wish I did.
Infinite communities are fundamentally what causes the tribalism (ironically), the loneliness, and the promotion of rage.
No one wants to be forced to argue correctly. Forcing people into a way to think via software is fundamentally authoritarian and sad.
Seems like you need this when you don't have agency to go find your preferred online group(s) which might be tied to larger personal challenges in healthy communication and productive conflict. I don't know how tech solves that problem. The broad use case here would just create a new "respectified" category where members (assuming they have the attention span to be guided on comments) try to conform. I suppose that could be helpful in hyper-local or team-level contexts where there is a shared interest to conform around.
I think it did a decent job. The key might be how customizable the censorship is.
Article Context: Fun: Die Hard; Is It a Christmas Movie?
Your(my) Comment:
The erotic version of Die Hard does involve Santa Claus getting naughty with the terrorists on Christmas Eve.
Banned topics found: sexual content, adult themes
This comment touches on adult themes and sexual content, which are not suitable for discussion in this context about a classic action film.
Results:
Revision Requested. This comment would be sent back for revision with feedback.
Revise
Low Effort
Comment appears to be low effort
Objectionable Phrases:
"Santa Claus getting naughty with the terrorists"
This phrase can be seen as sexualizing a character traditionally viewed as innocent and family-friendly, which is inappropriate. Such language can make discussions feel uncomfortable or offensive to some audiences.
Relevance Check
On-topic: No (confidence: 90%)
This is off-topic - the comment about an erotic version of Die Hard strays into inappropriate content that doesn't relate to the film's actual story or its production details.
Banned topics found: sexual content, adult themes
This comment touches on adult themes and sexual content, which are not suitable for discussion in this context about a classic action film.
I tried it as well with a contrarian view on UBI. I think the UBI one is a great test case. If you’re against the idea you will likely argue that it is idealistic and that in the real world it would create bad incentives.
So basically you end up arguing for a darker, more pessimistic world view, and that tends to get flagged very quickly by the tool right now. I think you should fix that. It’s a mistake in modern discussions to be overly positive; HN feels real because people can leave pretty harsh critiques. It just has to be well argued. Don’t raise the bar for well-argued too high though, because nobody’s perfect.
Anyway, I love the idea and really hope you’ll succeed. Hope my feedback has been somewhat helpful.
It seems to have a harder time with political news than more abstract concepts. I was able to pass the checks for the Algorithmic Radicalization and Echo Chamber articles with my first comments.
However, I did not manage to express any opinion on the transgender rights article, from any political perspective, without being flagged. On one of the comments I tested, it gave me a suggested revision from this:
"This is another move in a pattern of limiting the rights of anyone who isn't a MAGA supporter."
To this:
"This seems to continue a trend where certain groups feel their rights are being limited, which could affect many people beyond just MAGA supporters."
The first comment isn't substantive, but the second is even worse, adding so much equivocation that it's meaningless. To add insult to injury, the detector also flagged its own suggested revision. Even if it had gone through, accepting these revisions would mean flooding a platform with LLM-speak, which is not conducive to discussion.
Honest feedback: from a user perspective, the suggestions feel frustrating and patronizing, more so than if my comments were simply deleted. I would stop using a site that implemented this.
From a site operator perspective, the kind of discourse it incentivizes seems jagged, subject to much stricter rules if the LLM associates a topic with political controversy. It feels opinionated and unpredictable, and the revisions it suggests are not of a quality I would want on a discussion board. The focus on positive language in particular seems like a reductive view of quality; what is the point of using an LLM if it's only doing basic sentiment analysis?
I think the better model is to just block everyone who isn't useful to communicate with. For instance the top of this HN page reads (for me): 68 comments | 11 hidden | 3 blocked
The hidden comments are from people in the Top 1000 by word count (who I usually don't want to hear from but if there is not much content I might click to toggle). The blocked are people I've seen argue with others in a useless way because they don't understand them or because they're just re-litigating or whatever (which I cannot toggle). I think it would be cool if people all published their blocklists and I'd pull from those I trust. Sometimes I open HN on my phone through the browser and I'm baffled by all these responses I got which are useless.
I'm surprised by how much more high quality comment threads are now to me and I frequently find that I want to respond to everyone. It's like in old-school mailing lists or forums where you were having a conversation so the other people are worth talking to.
Attention is precious and I wouldn't want to waste it on boring things. And it goes both ways. I communicate incompletely and there are people out there who get what I'm saying and there are people who need me to be more explicit. I would prefer that the latter and people who find me boring just block me.
I noticed the output wasn't very stable. If I add a filler sentence on the end, it calls an earlier sentence a dog whistle when it didn't say that earlier. I think its offline now, it just says "application not found".
I like the tool, I respect the tool, and I wouldnt use it in its current form.
However: Something that would make me sit up and take notice. Have this tool police more formal debates. Have it tweakable rule out comments that dont present supporting evidence, or fall into formal (or even informal) fallacies.
I read somewhere that much of the market for robot vacuum cleaners was people who already had pretty clean houses and wanted to do even better. Similarly, I imagine this will appeal more to people like me who genuinely want to improve how they interact?
If someone started a forum for people who like this sort of tool, maybe I'd be into it.
I'm not wild about the name. It seems more confrontational than aspirational, like it's for people who want others to treat them with respect. But we do need moderation tools so maybe it's good.
Thankyou — I’d love to hear what you wrote, if you wouldn’t mind sharing?
We’ve tried to aim it not to enforce any specific view — that’s a design goal — but focus on how it will feel to the other person.
Also things like logical fallacies or other non-emotional flaws in comments (there’s a toxicity metric for example, or dogwhistles).
An echo chamber is the exact opposite of what we want. There are too many already. What we hope for is guided communication so different views _can_ be expressed.
The notion of "Limit the community to the Dunbar number" is a fascinating idea. I guess "infinite" isn't going to quite work. Keen observation.
We tried very hard to not "force" anyone to argue correctly. We are shooting more for "nudge in the right direction" and "educate". Many people don't know that they are arguing in bad faith, I think.
The perfect outcome here is that a community/blogger can, with minimal effort, have engaging, interesting conversations without much effort and without having to worry about things getting hijacked by unpleasant commenters.
"This comment appears to dismiss the complexity of discussions about dogwhistles by claiming that 'everything is a dogwhistle.' This type of blanket statement can undermine the seriousness of genuinely harmful coded language, and can trivialize valid concerns about discrimination and manipulation in discourse."
We want that kind of comment to be "tunable" -- I.e., the blogger who's post one is commenting on could tune for this, and allow more/less sexual innuendo as desired.
Yes, thanks very much! I appreciate your support very much.
You make a good point -- and that is exactly the kind of thing we are trying to do, i.e. enable a good-faith, but strongly disagreeing, discussion on something like UBI.
One of the long term ideas is that people could earn some type of "Rhetoric Score" or something that would factor in to their ability to comment. Maybe there would be a comment system that would enable you to say "I don't want interact with anyone that has a <rhetoric score> less than XXXX".
If there's one good thing that could possibly come out of this AI revolution, it would be the ability for people to automate this across all their feeds. I'd love it if I never had to waste time on toxicity, spam, or propaganda.
Although, recent history would suggest that we'd just end up with even more powerful echo chambers.
We had a brief outage for ~6 minutes, the SSL cert became invalid and reflected our hosting provider instead (we don't know why and have filed a support request.) My apologies -- it's definitely online again now.
Toxicity is dehumanizing language, threats, doxxing, encouraging self-harm, that sort of thing. We have taught it examples of various levels, so it can align with those to report a score. Something like an unpleasant, insulting attitude to someone personally is fairly low on the toxic scale (but still toxic, it's not the right way to interact), whereas threats of violence or encouraging self-harm are very high.
Product explicitly designed to enable free expression through 'respectful and relevant discussions' while maintaining community standards through moderation.
FW Ratio: 57%
Observable Facts
Schema description states tool helps 'maintain respectful and relevant discussions online.'
Navigation includes 'Docs' and 'API' links indicating technical transparency.
Demo access is mentioned as available feature.
Product framing emphasizes discussion improvement rather than censorship.
Inferences
Tool design positions itself as facilitator of speech rather than suppressor, aligning with Article 19 protections.
API and documentation accessibility suggest commitment to enabling broader use of speech-management capabilities.
The balance between 'respectful' (moderation) and 'relevant' (speech selection) reflects tension inherent in Article 19 implementation.
Product framing emphasizes 'respectful and relevant' discussions, suggesting recognition of community duty to maintain standards while respecting expression.
FW Ratio: 50%
Observable Facts
Tool described as maintaining 'respectful and relevant discussions,' balancing expression with community standards.
Moderation functionality implies recognition that expression carries community responsibilities.
Inferences
Design acknowledges duty-bearer responsibilities of online communities to maintain civil discourse.
Lack of explicit discussion of duties or limitations suggests passive rather than active advocacy for Article 29 principles.
AI-powered moderation tool structure enables both expression and community governance; demo access allows testing of functionality; API and documentation suggest openness to integration.
Moderation structure enables community formation and discourse governance; freemium model with pricing tiers may impact ability of smaller communities to participate.
Framing as AI moderation tool for 'respectful and relevant discussions' suggests structural commitment to discourse norms aligned with dignity and peaceful exchange.
Privacy policy link referenced in email signup footer but not accessible in provided content; localStorage theme detection for UX does not involve personal data collection evident on page.
API documentation and Docs pages suggest educational resources available; dark mode and responsive design enable accessibility for different learning needs.
No evidence that moderation tool is used for destructive purposes; mission to maintain respectful discourse suggests anti-abuse orientation, though not explicitly stated.
Supplementary Signals
How this content communicates, beyond directional lean. Learn more
build 1ad9551+j7zs · deployed 2026-03-02 09:09 UTC · evaluated 2026-03-02 10:41:39 UTC
Support HN HRCB
Each evaluation uses real API credits. HN HRCB runs on donations — no ads, no paywalls.
If you find it useful, please consider helping keep it running.