Skip to content

Comment on Ask HN: Starting an Anonymous Blog in 2019?

Comments

You have to blend in with the crowd to avoid sticking out. This means not buying a domain name (since domain registrations, even with contact information hiding, require real addresses in many cases for the registrar to process).

Then you use Tor to create a couple of ProtonMail and Tutanota addresses. Use these email addresses to create accounts on sites like GitHub while using Tor (make sure you link multiple addresses so that you have ways to get back in if one of them doesn’t work or you get kicked out).

Mirror all the writings on archive.org and another free site so that you have a backup to point people to (list the address of the other site in each site). Never trust any provider not to kick you out for “violating their terms and conditions” without telling you what you did or how you can fix it. When you get locked out, it’s usually with vague statements and no way to get back in. You’re at the mercy of bots and other people who may make it their mission to shut you down (depending on what you write).

Create multiple throwaway accounts without a lot of history to share the posts elsewhere.

Use Tor for everything related to the blog.

Edit: Building on what zelly said on translating from one language to another and back to reduce the chances of being identified by your writing. Somewhat similar in nature to hashing iterations, use one service to translate from X to Y, another to translate from Y to Z, and then yet another service to translate from Z to X. Then post X after making any necessary corrections. This could be automated with simple scripts. Using simple and short sentences could also help against any language analysis. Write, then feed it into something like Hemingwayapp, simplify it, then process it further with translation rounds.

It's not that hard to buy a domain name anonymously with Monero and providing fake information. Some don't even ask for anything, see for example njal.la.

Having to sacrifice style completely like this: that’s a drag. Are the style models accurate enough to id people already?

Seems like they’ve been using this with a good amount of success in certain cases. The Wikipedia page for Stylometry has more details. Bruce Schneier has also written about it a few times over the last decade.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.