Internet Filtering – How It Works, How to Work Around It

With Australian Internet users up in arms about the prospect of mandatory Internet filtering by the government, we thought this would be a timely subject. The purpose of this article is to show what Internet filtering is, and also why it isn't the best solution to the problem of illegal content online. It is not intended as an argument in favor of doing anything illegal.

The Basics

Filtering is based on very simple technologies that we've already been using for years:

- Website blocking. The most basic kind. Any site which is deemed unsuitable can be added to a "blacklist". At any level, including the web browser, the operating system, the gateway or router (modem), or the Internet Service Provider (ISP), a request to visit this site can be simply denied. All that's needed is plain text in a file or some configuration settings; many web administrator programs have a feature built-in for this purpose.

- Keyword blocking. A little more advanced; here we use a program that will scan the website when you request it but before it is shown to you. Any words found on the "blacklist" which are on the website will cause the website to become blocked.

- Search result filtering. This occurs with a doctored web browser or other program-level function, which intercepts a search query. The results returned from the search engine are first scrutinized for blacklisted terms, whether by site or keyword, and then the offending links are removed before you are shown the search results.

The points at which filtering can be done are as follows:

- The User's Computer. With software installed and configured from an admin account. This is the most easily subverted.

- Private network. Such as the "intranet" provided by a school, government office, or corporate workplace. That's handled on servers locked away in a room somewhere.

- ISPs. the commercial provider of Internet service can control all of the data going through their service.

- National Internet backbone. Going even over the heads of the ISPs, a country's government can institute a filtering system on the very infrastructure.

How to Break It

All of these methods rely on one kind of technology: plain, ordinary text. This is one of the factors that make it so easy to bypass. Let us use an example to show a few means whereby it can be easily circumvented: Say you're looking for the "Dead Parrot Sketch" by Monty Python.

Web Proxy Server: In the event that "www.MontyPython.com" is blocked, you may visit a web proxy server; here is a list of some. Picking one, you go there and type in "www.MontyPython.com", and the target site comes up in the web browser, while the server logs only record that you visited a proxy site. These are so simple, you could write one yourself in about ten minutes. They're laos frequently discovered and blocked.

Proxy Software: In the same case of blocking the website, proxy software can be installed on the user's machine or a connected server to automatically use proxies to view the web. These are much more secure and harder to detect; a list of available software is here. Most of them require some technical knowledge, however.

Privacy Software: In the event that you can access "www.MontyPython.com", but authorities are watching to see if you went there, there are various varieties of privacy software which is easy to install and use. The most popular varieties include Java Anon Proxy, Psiphon, and Tor. These also work against various specific cases of blocking and censorship.

Encryption: Taking it to the next level, Internet activity can be encrypted at many levels, from direct web access to sending and receiving files. In this case, a friend would have to know that you wanted the script to the Parrot Sketch; that friend (outside the censored zone) would then visit "www.MontyPython.com", download the script, archive it in a zip file, encrypt it, and send it to you. Email attachments are good for this. Many web-based email services offer encryption as a routine privacy service already.

Text munging: This method has already been deployed for years in the arms race between spammers and spam filters. In this example, someone who knew the the domain "MontyPython" was being keyword-blocked could post a link with the text "M0N7YPY7H0N", using letter and number substitution very much like that used in "leetspeak". It's only somewhat reliable, and of couse, filters can be trained to look for all permutations of the banned term. A Firefox add-on called 'leetkey' exists to do this automatically, and it can also translate to and from other unofficial dialects and partial encryption.

Image/Video Hiding: No text filter in the world can search an image or a video, at least not yet! That's why CAPTCHAs still work. Just display the copy of the Dead Parrot sketch, take a screenshot of it, save the image file as "MyPet.jpg", and post it on Flickr, then spread the word through private email. A video can display the text a page at a time (remember that videos can be paused), or even read aloud, and again given a misleading name. Unfortunately, if you don't know about the image or video under the code-word, you'll never find it.

"Dark Web": These are various unofficial loopholes to be found in some edge cases of filtering. A server which blocks "www.MontyPython.com" might not block BitTorrent, so a torrent of the Dead Parrot sketch from the original Monty Python TV series can be downloaded. BitTorrent is both an encrypted and anonymous protocol. Another example might be an old-fashioned BBS (Bulletin Board System) or Gopher server with plain text files of the Dead Parrot script. Yet another example is to find a copy of the file on Usenet or get it sent to you through IRC chat. It's not so much that these methods are made to avoid detection, as it is that these are the places where censors might not think of looking.

Sneakernet!: The final straw, you can simply have a friend save a copy of the Dead Parrot sketch onto their USB thumb drive, then physically walk that thumb drive over to you. Much in the same way that physical contraband is smuggled, with the same risks involved. International groups such as Information Without Borders are attempting to implement a complete protocol for a working "sneakernet".

In conclusion: The primary protest against Internet censoring and filtering is that it does very little to hamper actual criminals, such as child pornographers, while it greatly restricts the rights of ordinary, law-abiding citizens in the course of performing everyday, legal tasks on the Internet. We hope we have demonstrated some of the facts behind this case.

Filed Under: Jobs in Technology • Mobile Computing • Mobile Internet • Security Technology • The Internetz