Why a network engineer should know this organisation

Three things a security practitioner takes for granted were, within living memory, either illegal, officially denied, or unavailable. Publishing cryptographic source code was regulated as arms export. The fifty-six-bit Data Encryption Standard (DES) was described by the government that mandated it as adequate, and the cost of breaking it as prohibitive. And an HTTPS (Hypertext Transfer Protocol Secure) certificate cost money and effort, so most of the web did not have one.

Each of those changed because of the same San Francisco nonprofit. The Electronic Frontier Foundation () is usually described as a digital-rights group, which is true and incomplete. It is also the organisation that built the first unclassified DES cracker, that took AT&T to court with a fibre splitter as evidence, and that co-founded the certificate authority which now issues most of the certificates on the internet. This article follows that line - from a lawsuit about a game company to a piece of software on every web server - because it is the part of the EFF's history that shows up in a practitioner's work every day.

The visit, and the game company

By John Perry Barlow's own account, the foundation was started by a visit from the FBI (Federal Bureau of Investigation). In late April 1990 a special agent came to see the Grateful Dead lyricist and Wyoming cattle rancher about stolen Apple source code, and understood so little of what he was investigating that Barlow spent the interview explaining the technology to him. Barlow wrote it up on the WELL, the bulletin board where he and Mitch Kapor, founder of Lotus, both spent their evenings. Kapor had had a similar experience. They agreed there was a need to defend civil liberties on a network that the people enforcing the law did not understand, and Kapor agreed to fund the legal fees. Steve Wozniak offered to match Kapor's contributions without limit; John Gilmore, Sun Microsystems employee number five, emailed a six-figure offer. The foundation was formally created on 10 July 1990.

The case that gave it purpose was Steve Jackson Games, an Austin publisher of tabletop role-playing books, whose premises the Secret Service had raided that March and whose computers and unpublished manuscripts it had taken, on the theory that a game book about hacking was a hacking manual. Jackson had tried and failed to find a civil-liberties group that understood the technology well enough to see the problem. The raid, and the unconnected sweeps of the same spring, are told in full in the Sundevil article; this one picks up where that one leaves off.

Code is speech: Bernstein, 1995-1999

In 1995 Daniel Bernstein, a mathematics graduate student at Berkeley, wanted to publish a paper about an encryption method he called Snuffle, together with its source code. He asked the State Department whether he needed a licence. The answer was that Snuffle, in either form, was a munition under the International Traffic in Arms Regulations - which meant that to publish his own idea he would have to submit it for review, register as an arms dealer, and obtain an export licence, with civil and criminal penalties for failure. The EFF sued on his behalf.

The government's argument was that source code is functional as well as expressive - it does something - and so falls outside the First Amendment. The district court found in 1996 that source code was protected speech and that the regulations were a prior restraint on it. When licensing authority moved from State to Commerce in December 1996, Bernstein amended the complaint and won again in August 1997. On 6 May 1999 a Ninth Circuit panel upheld him, two to one, finding that the export rules operated as a prepublication licensing scheme burdening scientific expression, vested boundless discretion in officials, and lacked procedural safeguards. That opinion was withdrawn in September 1999 for rehearing by the full court, and the government relaxed the regulations before the rehearing happened - so the panel's words were never the final word. The practical result was the same. Cryptographic source code has been published freely since, and the Sixth Circuit reached the same conclusion in 2000.

For a practitioner the point is not the doctrine. It is that every open-source cryptographic library in use today - the ones inside every TLS (Transport Layer Security) stack, every (virtual private network), every (Secure Shell) client - could be published only because someone established that publishing them was not arms dealing. The encryption family history covers the export-control era from the algorithm side; this is the side that got the algorithms out of the country.

The machine that ended DES: July 1998

The United States government had, for two decades, pressed industry to limit itself to DES and weaker ciphers, while declining to say how easy DES was to break. In 1998 the EFF answered the question with hardware. For less than 250,000 dollars it built the first unclassified DES-cracking machine - nicknamed Deep Crack, after IBM's chess computer - from custom chips and off-the-shelf parts, capable of trying tens of billions of keys a second. On 17 July 1998 it won Laboratories' DES Challenge II in fifty-six hours, shattering the previous record of thirty-nine days set by a network of tens of thousands of computers. The decrypted message read: it's time for those 128-, 192- and 256-bit keys.

Six months later, on 19 January 1999, Deep Crack working with distributed.net's network of nearly 100,000 PCs won DES Challenge III in twenty-two hours and fifteen minutes. The plaintext this time was an invitation to the second conference in Rome. The National Institute of Standards and Technology had already opened the competition to replace DES; the machine turned a theoretical objection into a demonstrated fact, and the design was published in full, as a book with O'Reilly, so that nobody could claim the result was special to one laboratory. Later that year the federal standard was reaffirmed with Triple DES recommended in its place.

The lesson worth carrying is the one the EFF stated at the time: the government had exaggerated both the strength of the cipher and the cost of breaking it, and any well-funded organisation could have built the same machine in secret. A key length that a civil-liberties group can exhaust for a quarter of a million dollars is not a key length; it is a delay. The RSA Security entry in the industry catalogue records what the same government did, a few years later, to a random-number generator - the two stories are the same argument from opposite ends.

The room on Folsom Street: 2006-2022

In early 2006 a retired AT&T technician named Mark Klein walked into the EFF's office with documents. They showed that in AT&T's switching centre at 611 Folsom Street in San Francisco, a fibre-optic splitter had been installed on the backbone, copying internet traffic into a locked room - - to which only cleared personnel had access, and in which a Narus traffic-analysis system had been installed. The room had operated since 2003. Klein's documents indicated similar installations in other cities. He had, at real personal risk, brought the physical evidence of a mass-interception programme that until then had been rumour.

On 31 January 2006 the EFF filed Hepting v. AT&T on behalf of the company's customers, alleging that AT&T had collaborated with the National Security Agency in an illegal programme to wiretap and data-mine Americans' communications. That May a USA Today report confirmed the surveillance of call records, and more than fifty lawsuits followed, consolidated with the EFF as co-lead counsel. In July 2006 the court refused to dismiss the case on the government's assertion of state secrets. The Ninth Circuit heard argument in 2007 and never ruled: in July 2008 Congress passed the FISA Amendments Act, granting the telephone companies retroactive immunity for what they had done - a law written, the EFF has always said, in response to its victories in Hepting.

The foundation refiled against the government itself as Jewel v. NSA the same summer, with Klein's documents joined over the years by declarations from three NSA whistleblowers and, after 2013, by the published Snowden material describing the upstream collection from fibre cables that Room 641A was part of. In April 2019 the district court found that Klein could not establish the content or purpose of the room from his own knowledge; the Ninth Circuit ruled for the government on standing; and in June 2022 the Supreme Court declined to hear the case, ending sixteen years of litigation without a court ever ruling on whether the programme was lawful. Klein died in March 2025.

For the reader of this site the room matters as a fact about infrastructure. A splitter on a backbone fibre is invisible to every endpoint, every application and every certificate; it is defeated only by encryption that the tapping party cannot undo. Which is why the organisation that lost this case in the courts spent the following decade winning it in software.

From lawsuits to software: the Eckersley years

, an Australian computer scientist, joined the EFF in 2006 and became its chief computer scientist; he stayed until 2018 and has his own entry in this site's people section. Under him the foundation's technology projects turned the arguments it had been making in court into tools that made the arguments unnecessary.

Panopticlick, in 2010, showed that a browser could be identified without cookies, from the combination of its version, fonts, plugins and screen - the research that named browser fingerprinting as a problem before the advertising industry admitted using it. HTTPS Everywhere, built with the Project the same year, forced browsers onto encrypted connections wherever a site offered one, which at the time most did not. The SSL Observatory catalogued every certificate visible on the public internet and found how many were mis-issued. And Certbot automated the part of running a web server that most administrators got wrong: obtaining and renewing a certificate.

The largest of them began at the in 2012, when Eckersley met the researcher Alex Halderman and the two began planning a certificate authority that would issue certificates free, automatically, to anyone. Let's Encrypt, run by the Internet Security Research Group, in whose founding the EFF was a partner alongside Mozilla and others, opened to the public at the end of 2015 and became the largest certificate authority in the world. The reason the web is now encrypted by default - the reason a browser warns about a plain HTTP page rather than treating it as normal - is that the organisation which could not persuade a court to look inside Room 641A made the traffic passing through it unreadable instead.

What it is now

The EFF in 2025 had about 40,000 members, 125 staff, revenue of 26 million dollars and an endowment of 22 million; it funds legal defence, files amicus briefs, and runs the Pioneer Awards that have recognised, among many others, Klein himself. Its executive director is Nicole Ozer. It is criticised from the right as an obstacle to law enforcement and from the left as too close to the technology companies whose founders funded it, and both criticisms have been made continuously since 1990.

What the thread teaches

The through-line from the ranch kitchen to Let's Encrypt is a single idea applied four ways: that the people making rules about a technology should understand it, and that where they do not, the remedy is evidence. A lawsuit that establishes code is speech. A machine that establishes a key length is inadequate. A splitter, produced in court, that establishes a programme exists. And a certificate authority that establishes encryption need not cost anything. The industry catalogue on this site keeps returning to the question of who can compel a supplier - a government, a court, a record label, a lender. The EFF's history is the answer to the reciprocal question: who can compel the compellers, and with what. The answer, each time, was a fact that could not be argued with.

Sources