6 June 1991Phil Zimmermann · CryptographyCurated

Phil Zimmermann: Why I Wrote PGP

Summary

In the PGP manual, Phil Zimmermann explains why everyone should encrypt their email, including people with nothing to hide. His essay “Why I Wrote PGP” appeared in 1991 with the first version and was extended in 1999 to cover the Clipper chip, wiretapping laws and the end of US export restrictions. This post summarises the arguments, checks the facts cited and shows what became of PGP and the Crypto Wars.

What the essay is about

“Why I Wrote PGP” is part of the original 1991 PGP User’s Guide; the version published today was updated by Phil Zimmermann in 1999. The text is therefore both a founding document and a retrospective: events such as the 1994 CALEA wiretapping law or the 1993 Clipper chip could not yet appear in the first version.

Zimmermann’s starting point: two hundred years ago every conversation was private, you simply went behind the barn. The right to a private conversation was then not only a philosophical right but a physical one. With the telephone and email, even the most intimate conversations are carried electronically and can be read without anyone noticing. Zimmermann’s central thesis: the only way to hold the line on privacy in the information age is strong cryptography.

Postcards, envelopes and driftnets

The best-known argument is aimed at “if you have nothing to hide …”: even law-abiding people do not send their mail on postcards. If only a minority used envelopes, that alone would make them suspicious. Because almost all letters are in envelopes, nobody stands out. In the same way, everyone should encrypt routinely, whether a message is innocent or not. Zimmermann calls this a form of solidarity.

The second image describes how surveillance changed. In the past, authorities had to steam open letters and listen to and transcribe phone calls; that was expensive and only worthwhile in individual cases, like catching one fish at a time with hook and line. Email, by contrast, can be scanned for keywords automatically, on a vast scale and without detection, like driftnet fishing. With growing computing power, Zimmermann writes, the same will become possible for voice.

The laws and plans Zimmermann reacted to

According to Zimmermann, the trigger for publication was Senate Bill 266 of 1991. A non-binding resolution buried in it demanded that communications systems provide the government with plain text when authorised by law. Zimmermann published PGP for free that same year; the measure failed after protests by civil libertarians and industry. We could not check the wording of the bill against the primary source because congress.gov blocks automated access.

The other claims can be checked. CALEA (Public Law 103-414 of 25 October 1994) requires telephone companies to build their equipment so that they can “expeditiously” isolate and intercept a subscriber’s communications for the government. The White House presented the Clipper chip on 16 April 1993: each device received keys at manufacture that were deposited with government agencies (key escrow). Zimmermann’s figure of “about a thousand” court-ordered wiretaps per year matches the US courts’ statistics: 856 in 1991 and 1,058 in 1995.

According to Zimmermann, in 1995 the FBI wanted the capacity to wiretap 1 percent of all phone calls in major US cities simultaneously, more than a thousandfold increase over previous levels. That could only be handled with automatic voice recognition. As a reason not to trust the state blindly, Zimmermann cites history: the FBI’s COINTELPRO programme against the anti-war and civil rights movements, the wiretapping of Martin Luther King, Nixon’s enemies list and Watergate.

How PGP got onto the net in 1991

On 5 June 1991 Zimmermann sent PGP 1.0 for MS-DOS to two friends: Allan Hoeltje posted it to Peacenet, a provider for peace and grassroots groups, and Kelly Goen to a Usenet group for source code and to many bulletin board systems. On 6 June it was on the internet. At the time Zimmermann thought the “US only” tag was a real distribution limit, but it was only advisory.

Zimmermann had missed five mortgage payments developing PGP in the first half of 1991. A week before the release he learned of Privacy Enhanced Mail (PEM), a standard backed by large companies. Zimmermann considered its 56-bit DES too weak and objected that PEM signed every message and exposed the signature outside the encryption. PGP 2.0 followed in September 1992 for several platforms in about ten languages, with the trust model (web of trust) that made PGP famous.

The price: three years of investigation

Shortly after PGP 2.0, US Customs took an interest in the case. The US government treated strong cryptography as munitions; because PGP spread worldwide, it investigated for three years over a violation of the Arms Export Control Act. On 11 January 1996, Assistant US Attorney William Keane informed Zimmermann’s lawyer Philip Dubois that there would be no prosecution.

The dispute tended to help PGP. In 1995 the PGP source code was published as a book by MIT Press. The international version PGP 5.0i was created in 1997 on the same principle: printed source code was allowed to leave the US and was scanned back in in Europe. On 16 September 1999 the White House announced a new export policy; the Bureau of Export Administration rule of January 2000 allowed encryption to be exported to individuals, companies and other non-government users. Zimmermann wrote in 2021 that the export restrictions finally collapsed in 2000.

What remains of the arguments

Looking back on the 30th anniversary in 2021, Zimmermann lists what PGP was used for: the evacuation of 8,000 civilians during the Kosovo conflict, human rights groups in Guatemala and the Balkans, the resistance in Burma. Robert Morris Sr., formerly of the NSA, told Zimmermann in 2004 that the NSA had mainly feared the open source code would teach many people how to build strong public key cryptography.

Today strong cryptography is built into browsers, VPNs, banking apps and disk encryption. Nevertheless, Zimmermann warned in 2021 that governments in Australia, the UK and the US are again pushing back against end-to-end encryption, twenty years after the Crypto Wars were thought to be won. The 1999 “we have won” referred explicitly only to export controls.

Ideas

Encryption protects best when everyone uses it routinely.
People with nothing to hide still do not send postcards.
Automatic analysis turns targeted wiretapping into a driftnet.
Technical infrastructure outlasts the laws that prevent its abuse today.
Widespread encryption is politically harder to ban than rare encryption.
Printed source code was not covered by the export ban on software.

Insights

Zimmermann’s strongest argument is social rather than technical: privacy depends on its use being normal and not standing out.
The warning about permanent surveillance infrastructure under changing political conditions applies in any country: laws can change overnight, built-in wiretapping interfaces stay.
The export bans missed their goal: the investigation of Zimmermann made PGP better known, and printed source code legally bypassed the ban.

Facts

PGP 1.0 for MS-DOS reached the net via Peacenet, Usenet and bulletin boards on 5 and 6 June 1991.
PGP 2.0 was released in September 1992 with the web of trust model.
The investigation of Zimmermann for violating the Arms Export Control Act ended on 11 January 1996 without charges.
The White House presented the Clipper chip with escrowed keys on 16 April 1993.
US courts authorised 856 wiretaps in 1991 and 1,058 in 1995.

References

Critique

The solidarity argument assumes that everyone is able to encrypt. In 1999 Whitten and Tygar showed that only a third of twelve experienced email users could correctly sign and encrypt a message with PGP 5.0 within 90 minutes; a quarter accidentally sent the secret in plain text.
The essay equates privacy with encrypted content. PGP, however, does not hide who is communicating with whom and offers no forward secrecy: anyone who later obtains the key can read all old messages. Latacora raised these points against PGP in 2019, and EFAIL showed in 2018 how email clients could leak the plain text of encrypted messages.

Remarks

We report the wording of Senate Bill 266 and the 1995 FBI plan as described by Zimmermann; congress.gov and the New York Times blocked automated retrieval.
Since 2004 Zimmermann has worked on encrypted internet telephony (ZRTP, Zfone, Silent Phone) and is Associate Professor Emeritus of cybersecurity at Delft University of Technology.

Recommendations

Read the essay in the original: it is short and contains the most quoted arguments of the Crypto Wars to this day.
For confidential conversations today, use a messenger with end-to-end encryption and forward secrecy.
If you use OpenPGP, use current implementations following RFC 9580 and keep email clients up to date after EFAIL.

Read the essay

Search the Web Archive