Hacker News Reader

Comment by john_strinlai | Hacker News Reader

indeed, the actual case report (https://www.documentcloud.org/documents/28831278-govuscourts...) states:

"technical and network-activity records from fly.io included:

SSH session logs (CSV format) listing multiple "New SSH session" entries for <redacted> with verified=true between approximately 2026-03-26 and 2026-04-01, timestamped in UTC.

outbound-connection records (CSV format) reflecting numerous TCP port 443 connections originating from internal fly.io addresses <redacted> to external IP addresses attributed to AMAZON-02 (Amazon Web Services) in the US and Cloudflarenet (Cloudflare)".

View on HN

Yea, I thought the whole Lavabit episode made it clear that privacy oriented solutions based in the US were going to have a hard time. Most Proton’s whole selling point is their location.

And that they'll let you register even when connected to tor.

I tried and failed to get working account.

Huh, I though they did. But maybe I was wrong or they changed that.

I think they have a daily rate limit for free Tor accounts. You can make a paid account through Tor with Bitcoin, which of course you obtained by some anonymous method.

However on page 8 the document states that these "session logs" don't record commands, they're just audit logs of when, where, by whom the sessions were initiated.

for some reason, the above comment wont let me edit it (despite it being a minute old). but i was going to add:

"i wonder if fly.io bothers to vet/challenge records requests or if they just hand data over".

(this one can be edited, though. very strange)

All of your comments are coming up as [dead]. I vouched and upvoted your first one shortly after you posted it, I think that's probably why you couldn't edit it anymore.

huh... shadowban of some sort because they aren't appearing as [dead] on my side. not sure what rule i broke, but okay.

thanks dang/tomhow.

fwiw I'm seeing these comments.

yeah, they were vouched for. but, for example, others were not and are still hidden (https://news.ycombinator.com/item?id=50273618). and apparently this one, too.

the most annoying part is that i got no indication that i was suddenly shadowbanned. if someone didn't tell me, it would have probably been a day or two at least of wondering why i had no engagement on my comments.

no notification, no appeal system outside of email and pray. very cool system.

Informing the subject they are shadowbanned would kind of defeat the purpose of the shadowban. Might as well just issue a regular ban then.

It's nice that this place maintains an email endpoint that's actually staffed by humans who can intervene if needed. That's more than can be said about most places these days.

>Informing the subject they are shadowbanned would kind of defeat the purpose of the shadowban. Might as well just issue a regular ban then.

shadow bans here have no purpose. i would have found out in a couple days even if no one told me. so yes, it might as well just be a regular ban, correct. at least that way i would know what rule i broke.

Definitely some weirdness going on in this whole thread in general. It went from front page to 5th page in an instant.

Some of your comments were killed because our software classified the text as genai. That's not allowed on HN - see https://news.ycombinator.com/newsguidelines.html#generated and https://news.ycombinator.com/item?id=47340079.

none of my text is genai. ever. and i dont appreciate your insinuation that it is.

the comment that started this was quoting directly from the HSI document.

this comment was one that was killed: https://news.ycombinator.com/item?id=50273618 can you please tell me how i can write it so your bot doesn't think i am a bot?

same for this one https://news.ycombinator.com/item?id=50273525 or this one https://news.ycombinator.com/item?id=50274615. i'll take advice on how to rewrite the original one too (https://news.ycombinator.com/item?id=50272749) because im at a loss for words how it was classified as genai.

edit: i just ran all of my recent comments through pangram and they came back 100% human (which i knew, of course, considering i wrote them), so i am really not clear on how you want me to write my comments. apparently using my own words isn't enough for you or your bot.

OK, well there seem to have been a couple of false positives and perhaps a glitch in one case, which we'll look into. Sorry about that. In the case of the quote from the HSI document, perhaps that text you copied/pasted was LLM-enhanced (the comment was classified as genai according to all the methods we use, but of course no approach is 100% accurate).

False positives are going to happen, no matter how much we try to make our systems infallible, and you're welcome to email us if any comments of yours get killed and aren't quickly vouched by other users.

I can still confirm that your comments weren't being killed via any deliberate moderator intervention.

alright, that's fine, and thanks for saying it. false positives are expected, but your initial reply was not.

Thanks for understanding. I should have read the comments more closely and normally do, but I was on the defensive after the comment seemed to accuse us of killing comments, or at least make assumptions that we did or would do that.

Fly.io really needs to explain what happened here.

Reply on news.ycombinator.com