A technical guide to anonymous blogging – a very early draft

The Electronic Frontier Foundation posted an excellent guide to safe blogging a few days back. While the guide is quite rich in tips to ensure you don't reveal too much personal information while blogging, it doesn't look very closely at the technical issues associated with keeping a blog private. I decided to write a quick technical guide to anonymous blogging, trying to approach the problem from the perspective of a government whistleblower in a country with a less-than-transparent government. What follows is a first draft – I'll be posting it on the wiki as well (in a day or two) and will be grateful for comments, corrections and input.

Sarah works in a government office as an accountant. She becomes aware that her boss, the deputy minister, is stealing large amounts of money from the government. She wants to let the world know that a crime is taking place, but she's worried about losing her job. If she reports the matter to the Minister (if she could ever get an appointment!), she might get fired. She calls a reporter at the local newspaper, but he says he can't run a story without lots more information and documents proving her claims.

So Sarah decides to put up a weblog to tell the world what she knows about what's happening in the ministry. To protect herself, she wants to make sure no one can find out who she is based on her blog posts – she needs to blog anonymously.

There are two major ways a blogger can get caught when she is trying to blog anonymously. One is if she reveals her identity through the content she publishes – for instance, if Sarah says, “I'm the assistant chief compliance accountant to the Deputy Minister of Mines,” there's a good chance that someone reading her blog is going to figure out who she is pretty quickly. The Electronic Frontier Foundation's guide, “How to Blog Safely”, offers some great advice on how to avoid revealing your identity with the content of your blog.

The other way Sarah can get caught is if someone can determine her identity from information provided by their web browsers or email programs. Every computer attached to the internet has – or shares – an address called an IP address – it's a series of four numbers from 0-255, separated by dots – for instance: When Sarah uses her web browser to make a comment on the Minister's blog, the IP address she was using is included on her post.

With a little work, the Minister's computer technicians may be able to trace Sarah's identity from this IP address. If Sarah is using a computer in her home, dialing into an Internet Service Provider, the ISP likely has records of which IP address was assigned to which telephone number at a specific time. In some countries, the Minister might need a subpoena to obtain these records; in others (especially ones where the ISP is owned by the government!), the ISP might give out this information very easily, and Sarah might find herself in hot water.

There are a number of ways Sarah can hide her identity when using the Internet. As a general rule, the more secure Sarah wants to be, the more work she needs to do to hide her identity. Sarah – and anyone else hoping to blog anonymously – needs to consider just how paranoid she wants to be before deciding how hard she wants to work to protect her identity. As you will see, some of the strategies for protecting identity online require a great deal of technical knowledge and work.

Step one – psuedonyms

One easy way Sarah can hide her identity is to use a free webmail account and free blog host outside her native country. (Using a paid account for either email or webhosting is a poor idea, as the payment will link the account to a credit card, a checking account or paypal account that could be easily linked to Sarah.) She can create a new identity – a psuedonym – when she signs up for these accounts, and when the Minister finds her blog, he'll discover that it belongs to “A. N. Ymous”, with the email address anonymous.whistleblower@hotmail.com.

Some providers of free webmail accounts:

  • Hotmail
  • Yahoo
  • Hushmail – free webmail with support for strong cryptography

    Some providers of free weblog hosting:

  • Blogsome – free WordPress blogs
  • Blogger
  • Seo Blog
  • Here's the problem with this strategy – when Sarah signs up for an email service or a weblog, the webserver she's accessing logs her IP address. If that IP address can be traced to her – if she's using her computer at home, or her computer at work – and if the email or weblog company is forced to release that information, she could be found. It's not a simple matter to get most web service companies to reveal this information – to get Hotmail, for instance, to reveal the IP Sarah used to sign up for her account, the Minister would likely need to issue a subpoena, probably in cooperation with a US law enforcement agency. But Sarah may not want to take the risk that she could be found if her government can persuade her email and weblog host to reveal her identity.

    Step two – public computers

    One additional step Sarah could take to hide her identity is to begin using computers to make her blogposts that are used by lots of other people. Rather than setting up her webmail and weblog accounts from her home or work computer, Sarah could set them up from a computer in a cybercafe, a library, or a university computer lab. When the Minister traces the IP used to post a comment or a post, he'll discover that the post was made from a cybercafe, where any number of people might have been using the computers.

    There are flaws in this strategy as well. If the cybercafe or computer lab keeps track of who is using what computer at what time, Sarah's identity could be compromised. She shouldn't try to post in the middle of the night when she's the only person in the computer lab – the geek on duty is likely to remember who she is. And she should change cybercafes often. If the Minister discovers that all the whistleblower's posts are coming from “Joe's Beer and Bits” on Main Street, he might stake someone out to watch the cybercafe and see who's posting to blogs in the hopes of catching Sarah.

    Step three – anonymous proxies.

    Sarah's getting sick of walking to Joe's cybercafe every time she wants to post to her blog. With some help from the neighborhood geek, she sets up her computer to access the web through an anonymous proxy. Now, when she uses her webmail and weblog services, she'll leave behind the IP address of the proxy server, not the address of her home machine… which will make it very hard for the Minister to find her.

    First, she finds a list of proxy servers online, by searching for “proxy server” on Google. She picks a proxy server from the publicproxyservers.com list, choosing a site marked “high anonymity”. She writes down the IP address of the proxy and the port listed on the proxy list.

    Some reliable lists of public proxies:

  • publicproxyservers.com – lists anonymous and non-anonymous proxies
  • Samair – only lists anonymous proxies, and includes information on proxies that support SSL
  • rosinstrument proxy database – searchable database of proxy servers
  • Then she opens the “preferences” section of her webbrowser. Under “general”, “network” or “security” (usually), she finds an option to set up a proxy to access the Internet. (On the Firefox browser, which I use, this option is found under Preferences – General – Connection Settings.)

    She turns on “manual proxy configuration”, enters the IP address of the proxy server and port into the fields for HTTP proxy and SSL proxy and saves her settings. She restarts her browser and starts surfing the web.

    She notices that her connection to the web seems to be a bit slower. That's because every page she requests from a webserver takes a detour. Instead of connecting directly to Hotmail.com, she connects to the proxy, which then connects to Hotmail. When Hotmail sends a page to her, it goes to the proxy first, then to her. She also notices that she has some difficulty accessing websites, especially sites that want her to log in. But at least her IP isn't being recorded by her weblog provider!

    A fun experiment with proxies:
    Visit noreply.org, a popular remailer website. The site will greet you by telling you what IP address you're coming from: “Hello pool-151-203-182-212.wma.east.verizon.net [], pleased to meet you.”

    Now go to anonymizer.com, a web service that allows you to view (some) webpages through an anonymous proxy. In the box on the top right of the anonymizer page, enter the URL for http://www.noreply.org (or just click this link.) You'll note that noreply.org now thinks you're coming from vortex.anonymizer.com. (Anonymizer is a nice way to test proxies without needing to change your browser settings, but it won't work with most sophisticated web services, like webmail or weblogging servers.)

    Finally, follow the instruction above to set up your webbrowser to use an anonymous proxy and then visit noreply.org to see where it thinks you're coming from.

    Alas, proxies aren't perfect either. If the nation Sarah lives in has restrictive internet laws, many websurfers may be using proxies to access sites blocked by the government. The government may respond by ordering certain popular proxies to be blocked. Surfers move to new proxies, the government blocks those proxies, and so the circle of life continues. All of this can become very time consuming.

    Sarah has another problem if she's one of very few people in the country using a proxy. If the comments on her blog can be traced to a single proxy server, and if the Minister can access logs from all the ISPs within a country, he might be able to discover that Sarah's computer was one of the very few that accessed a specific proxy server. He can't demonstrate that Sarah used the proxy to post to a weblog server, but he might conclude that the fact that the proxy was used to make a weblog post and that she was one of the few people in the nation to use that proxy constituted evidence that she made the post. Sarah would do well to use proxies that are popular locally and to switch proxies often.

    Step four – MixMaster, Invisiblog and GPG

    Sarah starts to wonder what happens if the proxy servers she's using get compromised? What if the Minister convinces the operator of a proxy server – either through legal means or through bribery – to keep records and see whether anyone from his country is using the proxy, and what sites they're using. She's relying on the proxy administrator to protect her, and she doesn't even know who the administrator is!

    Spending quite a long time with the local geek this time, she explores a new option: Invisiblog. Run by an anonymous group of Australians called vigilant.tv, Invisiblog is a site designed for and by the truly paranoid. You can't post to Invisiblog via the web, as you do with most blog servers. You post to it using specially formatted email, sent through the MixMaster remailer system, signed cryptographically.

    It took Sarah a few tries to understand that last sentence. Eventually, she set up GPG – the GNU implementation of Pretty Good Privacy, a public-key encryption system. (In two sentences: Public-key encryption is a technique that allows you to send messages to a person that only she can read, without her needing to share a secret key with you that would let you read messages other people send to her. Public key encryption also allows people to “sign” documents with a digital signature that is almost impossible to forge.) She generates a keypair that she will use to post to the blog – by signing a post with her “private key”, the blog server will be able to use her “public key” to check that a post is coming from her, and then put it on the blog.

    She then sets up MixMaster, a mailing system designed to obscure the origins of an email message. MixMaster uses a chain of anonymous remailers – computer programs that strip all identifying information off an email and send it to its destination – to send email messages with a high degree of anonymity. By using a chain of 2 to 20 remailers, the message is very difficult to trace, even if one or more of the remailers is “compromised” and is recording sender information. She has to “build” MixMaster by compiling its source code, a project that requires a great deal of geek assistance.

    She sends a first MixMaster message to Invisiblog, which includes her public key. Invisiblog uses this to set up a new blog, with the catchy name “invisiblog.com/ac4589d7001ac238″ – the long string is the last 16 bytes of her GPG key. Then she sends future posts to invisiblog, by writing a text message, signing it with her public key and sending it via MixMaster.

    It's not nearly as fast as her old style of blogging. The misdirection of MixMaster mailers means that it takes anywhere from two hours to two days for her message to reach the servers. And she has to be very careful about looking at the blog – if she looks at it too often, her IP address will appear in the blog's log frequently, signalling that she'l likely to be the blog author. But she's reassured by the fact that the owners of invisiblog have no idea who she is… and she now knows more about open-source software than she'd ever wanted to know.

    The main problem with the Invisiblog system is the fact that it's incredibly difficult for mere mortals to use. Most people find GPG a challenge to set up, and have difficulty understanding the complexities of public and private keys. More userfriendly crypto tools, like Ciphire, have been set up to help the less geeky of us, but even they can be tricky to use. As a result, very few people – including people who might really need it – use encryption for most of their email.

    MixMaster is a true technical challenge for most users. Windows users can use an early DOS version of the program by just downloading it. I downloaded and tested it yesterday, and it doesn't appear to work… or perhaps my email is still being remailed back and forth between remailers. Anyone wanting to use the newer version, or wanting to use the program on Linux or Mac, needs to be able to compile the program themselves, a task beyond many expert users. It's possible that Invisiblog would become more useful if it accepted messages from web-accessible remailers, like riot.eu.org – for now, I can't see it as being particularly helpful for the people who need it most.

    There's an additional set of problems with strong encryption in repressive nations. If Sarah's computer is seized by the government and her private key is found, it would constitute strong evidence that Sarah had authored the controversial blog posts. And, in countries where encryption is not widely used, simply sending out MixMaster messages – mail messages wrapped in strong encryption – might be enough to cause Sarah's internet usage to be watched closely.

    Is Sarah's solution – learning enough about cryptography and software to use MixMaster – your solution? Or is some combination of steps 1,2 and 3 sufficient to let you blog anonymously? There's no one answer – any path towards anonymity needs to consider local conditions, your own technical competence and your level of paranoia. If you have reason to be worried that what you're posting could endanger your safety, a combination of steps 1, 2 and 3 is probably not a bad idea.

    Oh, and remember not to sign your blog posts with your real name!


    • Small Bits of Chaos: Video, Anonymous Blogs, Real ID Act dead

      This New York Times article on Videos Challenge Accounts of Convention Unrest covers the fascinating conflict between the video and human memories of an event; the issues raised by transparent video editing, and other issues. Worth reading. During a…

    • I’m already getting good feedback on this, including the valid question: why no mention of TOR (The Onion Router). The simple answer – I’m still experimenting with TOR and wasn’t ready to write knowledgeably about it yet. I’ll revise this document with information on Ciphire and TOR in the next week or two.

    • Ari

      Looks like a solid article, and though I’m not much of a techie, it seems like it’s got some with some good practical advice for those who need it.

      More philosophically, I’m opposed to the notion of anonymous blogging where-ever it is possible to avoid it. Anonymity reduces the credibility of what is being said, and opens up the opportunity for scandal and muckracking, for example by people deliberately spreading false information using an anonymous blog. Instead, where-ever possible bloggers should correctly disclose their ID, at least in the case of opinion/commentry blogs. Obviously anonymity is essential in some cases, notable in non-democratic states and where one’s job might be on the line, but as a general trend in blogging, I don’t think it’s a healthy one.

    • […] ; Ethan @ 6:30 pm

      I received excellent feedback on the first draft of the technical guide to internet anonymity I wrote a few days back for Global Voices. One of the ques […]

    • I generally write about proxy servers from the perspective of censorship circumvention, but I believe my comments hold true for people seeking anonymity.

      With open proxy servers, it is not explicitly known if open proxy servers have been set-up as a public service or if they have been just badly configured to inadvertently allow public access. Depending on the interpretation of local law, the use of open proxy server may be viewed as ‘unauthorized access’ and open proxy users may subject to legal penalties.

      The anonymity provided by using public proxies is simply to sheild your IP address from the host you are connecting to (such as your blog host provider). there is also no guarantee that the information concerning anonymity level of the proxy servers is accurate. Always double check that identifying information is not being forwarded by the proxy server.

      All the communication between you and the proxy server is almost always in plain text, thus easily intercepted by anyone sniffing traffic upstream. And any information passing through the proxy server can be intercepted by the owner of the proxy server.

      When using an open proxy server your anonymity is entirely in the hands of an unknown, untrusted intermediary. Therefore the use of open proxies is not a prefered option and should only be used by persons in low security risk environments who require temporary or ad-hoc anonymity needs and who do not need to transmit sensitive information.

    • […]

      Anonymous Blogging
      Tuesday 3 May 2005 @ 7:37 pm

      From Global Voices, a technical accompaniment to the recent EFF post on How to Blog Safely (Anonymously).


    • […] hlessly alarmist article on blogging safely Canwest’s Tips for Blogging About Work Global Voices on blogging anonymously

      | Posted in Weblogs, Employment |


    • […] zuckerman @ 6:21 am

      I posted, some weeks back, the first draft of a technical guide to anonymous blogging. I’ve gotten great feedback from fo […]

    • http://www.martus.org/

      “””Martus is a software tool that allows users to document incidents of abuse by creating bulletins, uploading them at the earliest opportunity, and storing them on redundant servers located around the world.”””

      “””Martus™ software – on the client side – will incorporate encryption capabilities to maintain the integrity and security of the bulletins created and reported by NGOs. This encryption takes place at the initial stage of bulletin creation, before it is transmitted over the Internet.”””

      “””Martus tools are designed as free/open source software programs, built on top of open source tools.”””

    • Thanks, Jean. I know Martus pretty well and think it, and the group behind it – Benetech – are terrific. They’re not a publishing platform, though. They’re really designed to provide secure backup and storage for data, not to publish anonymously on a regular basis…

    Join the conversation

    Authors, please log in »


    • All comments are reviewed by a moderator. Do not submit your comment more than once or it may be identified as spam.
    • Please treat others with respect. Comments containing hate speech, obscenity, and personal attacks will not be approved.