Diego 
Terence EdenI did the `human.json` thing!
If you're a blogger who I've met in person, and you're not using AI or LLMs for your content, let me know and I'll add you to my list.
Dr David Mills
Emily_S@Edent me! I'll add yours to mine too.
Though I'm not seeing it show up on your blog yet? Waiting for a rebuild/cache invalidation?
@emily_s should be live now - let me know the URl you're trying if you're having issues.
@Edent you should be vouched for now from https://parsecsreach.org
@emily_s likewise from my end š
Esther Payne 
@Edent Me please. I need to look at doing that on my website as well.
We met at the Scottish Open Source Awards in Edinburgh. I remember we were talking about beer.
Owen Blacker@Edent Ooh, I like that. I should do the same
Antoine
Konstantin 
@Edent Great idea! š
The gallant knight@Edent Great, AI then know what's fresh and human made. No slop involved.
Stefano Marinelli@Edent Good idea!
Giacomo Tesio@[email protected]
Shouldn't human content be the default? Why should we concede such default?
@[email protected]
Shouldn't human content be the default? Why should we concede such default?
@[email protected]
Eric A. Meyer@Edent Ah, I see FOAF has returned once again.
Andy Piper@Meyerweb @Edent ha, yes, I referenced FOAF only 18 months ago... https://andypiper.co.uk/2024/08/29/the-web-made-by-humans/
@andypiper @Edent Have you looked through XFN at all? The goal there was admittedly to represent relationships in page source rather than metadata, but I imagine many of the general concepts could be of interest.
Pol Dellaiera@Edent Me, at fosdem :) https://not-a-number.io !
@Edent Just added you: https://not-a-number.io/human.json
kianryan āļøšš³ļøāš@Edent
Definitely no AI/LLM. Sometimes seen as human.
Definitely no AI/LLM. Sometimes seen as human.
Gustavo@Edent The design is human.
Meme aside, yeah, I find it weird the use of JSON (which many people say is more machine-friendly than human-friendly) for that. Reminds me of https://humanstxt.org/, that uses .txt which seems more human-friendly.
Also, the "web" thing reminds me a bit of Keybase/Keyoxide. I hope it's not a case of missing research of existent work. Reminds me of this https://fliptomato.wordpress.com/2007/03/19/medical-researcher-discovers-integration-gets-75-citations/. I wonder if this could be proposed as an extension of one the earlier projectsā¦
@qgustavor feel free to suggest that in the site's repo.
@qgustavor @Edent I had a go at contacting the humans.txt folks last year I think (maybe a bit before that), suggesting it was an idea whose time had come, but it was all very quiet there. Interested to see the human.json idea develop.
@andypiper @Edent I guess my main issue with the idea is that I don't like editing JSON files. I blame Crockford for not allowing comments. It could be JSON5 at least...
But FOSS collaboration is like that, it's how we got a file system API that allows streaming files to disk but doesn't allow streaming files to the download folder because Google singlehandedly made the API and everyone else rejected it as it looked unsafe as hell, and a speech recognition API that only works for searching, you can't use for voice control.
But FOSS collaboration is like that, it's how we got a file system API that allows streaming files to disk but doesn't allow streaming files to the download folder because Google singlehandedly made the API and everyone else rejected it as it looked unsafe as hell, and a speech recognition API that only works for searching, you can't use for voice control.
S. Lott@Edent Thanks for posting a reminder. https://slott56.github.io/2026-03-20_human_json.html
@andypiper @Edent itās not on every blog post, itās only on the top-most index page. Should it be present on every page?
@slott56 oh, you're right, I see it on the top-level page, but the browser extension is not picking it up. Hmm! how odd.
@andypiper good. Iāll tweak the template to put it on more pages.
@andypiper Not odd. Syntax error in the JSON.
I finally built a schema validator. JSON was bad.
Eternal Gratitude for noting that it didn't work and forcing me to debug it.
devsimsek
Erin Brown@Edent Oooh! Looking forward to reading this more carefully! ā¦ Since itās json, would it be feasible to add a property for including a public key? Perhaps this can further be reinforced through a cryptographic web of trust? š¤
@rinkside you can suggest it - but cryptographic webs are a usability nightmare. Every single one has failed.
Maybe the next one won't. But I don't have high hopes.
@Edent Will have to read more about that issue ā definitely not an area of expertise. Thank you!
These days Iām digging through old ā¦ ideas? ā¦ that may have been pushed aside in the rush to corporatize and centralize as we try building an alternate, more human network. š
@rinkside yeah, things like Keybase tried something similar - but they got acquihired by Zoom during the pandemic.
I think, pragmatically, If I've stuck something on the website I control, that should be "good enough".
@Edent Yes, the pgp key would reside on a self-controlled server. Isnāt there a feature with keys to build a web of trust through in-person key signing parties?
Host a human.json with a property pointing to a key that when inspected is bolstered by a human-verified web of trust. Thatās where my thinking is. š
@Edent you were on my human.json before you created your own, innit!
Daniel Siepmann@Edent Cool idea, I like it.
But I fear AI companies might again use it, once it have become adopted, to rank human content higher while training future models.
Bastian Greshake Tzovaras@Edent š (if zooms count š
)
Timo Tijhof
Gaelan Steele@Edent not a blogger per se, but there are enough miscellaneous pages on gaelan.me thatās itās probably worth including
(and Iām long overdue to properly index them all!)
Simon Cox 
Furbland's Very Cool Mastodonā¢> Protocol must match. https://example.com and http://example.com are different identities.
??? why
@GroupNebula563 I'm not the spec author. You should raise an issue on CodeBerg.
@Edent oh, that wasn't at you, that was just a general statement of confusion :P
@GroupNebula563 Different ports on a domain can be run by different people. For example, the contents of ftp://example.com might be owned by someone else.
@Edent fair enough :P
Patrik Wallstrƶm@Edent This is exactly what an AI would do.
Fazal Majid@Edent what is the opposite of "argumentum ad hominem", " argumentum ad robotem"?
Myles Lewando@Edent Great idea. I'm going to investigate further.
Nathan Hubbard@Edent I have been thinking about this a bit. Would this just give greater signal to the LLM content stealers that my site is ripe for original non-LLM collapse causing content? Iām leaning towards doing it anyway.
@n8foo My working assumption is they'll do that anyway. I'm not doing it for them.
ā Panda | ģķė¤@Edent Iāve added your blog to my newly created file (on https://gizmonaut.net).
It's been a very long time! Hope you're well.
tyil@Edent Interesting concept. I'm not sure yet if I like it, but I do agree that there is a strong desire for ways to verify real humans these days. A couple remarks I have, though...
I see a JSON Schema definition in the README, and if you want schemas to verify with, why not just use XML, which has first-class support for schemas and widely supported tooling to verify these?
Addditionally, I've been told all my life while telling people to use OpenPGP that "the web of trust doesn't work". This seems like basically the same thing but without the cryptography aspect of OpenPGP, what makes you think the web of trust will work properly here? In fact, if you want to make a web of trust, why not built something on top of OpenPGP? You get the trust aspect, while also promoting use of cryptography keys.
I see a JSON Schema definition in the README, and if you want schemas to verify with, why not just use XML, which has first-class support for schemas and widely supported tooling to verify these?
Addditionally, I've been told all my life while telling people to use OpenPGP that "the web of trust doesn't work". This seems like basically the same thing but without the cryptography aspect of OpenPGP, what makes you think the web of trust will work properly here? In fact, if you want to make a web of trust, why not built something on top of OpenPGP? You get the trust aspect, while also promoting use of cryptography keys.