Skip to content

Phil Katz and the Zip Format

Abstract

Almost every computer user has opened a .zip file; few know Phil Katz, the Milwaukee programmer who created the format. He made his name in the mid-1980s with faster versions of the ARC archiver that bulletin-board users relied on, was sued by ARC’s owner in 1988 and settled, and in February 1989 released PKZIP with a new format whose specification he made free for anyone to implement. That decision made zip the universal container for compressed files, and his DEFLATE method of 1993 became one of the most widely used compression algorithms, in gzip, PNG and the web’s HTTP compression. It also made his own program replaceable, and WinZip and Windows itself took over the market while Katz, who disliked Windows, shipped a Windows version only in 1996. His company prospered while his alcoholism isolated him from his family, his employees and the law. He was found dead in a Milwaukee hotel room in April 2000, aged 37.

Milwaukee and the Bulletin Boards

Phillip Walter Katz was born on 3 November 1962 in Milwaukee, Wisconsin, and grew up in suburban Glendale. School friends remembered a small, asthmatic boy nicknamed “the Elf”, good at chess and one of the “brains”. His father, Walter, with whom he played chess for hours, died after heart surgery when Phil was 19. Katz graduated from the University of Wisconsin–Milwaukee with a degree in computer science engineering in 1984, worked at two programming jobs, and lived with his mother, Hildegard, a nurse instructor, into his twenties.

At night he was on the bulletin board systems through which hobbyists exchanged programs and messages (The BBS Era). With modems running at 1,200 bits per second and long-distance calls charged by the minute, every kilobyte cost time and money, so compression utilities were among the most used programs on the boards. The standard was ARC, published by System Enhancement Associates (SEA) of Clifton, New Jersey, and written by Thom Henderson. It combined several files into one archive and compressed them.

Katz rewrote ARC’s routines in assembly language and from 1985 released his own compatible versions, PKXARC and PKARC, as shareware: free to try, with users asked to pay if they kept using it (Shareware and Public Domain). They were much faster than SEA’s program. He posted them on ExecPC, a large Milwaukee bulletin board run by Bob Mahoney. “He didn’t do it to get rich,” Mahoney recalled. “He did it for himself.” Payments arrived quickly; a colleague at the small software company where Katz then worked remembered him receiving two checks of $10,000 within two weeks, nearly a year’s salary, after which he decided he could quit. In 1986 he founded PKWARE, working from his mother’s kitchen table and giving her a share of the company; she left her job to run its administration.

The ARC War

SEA regarded PKARC as a copy of its program. In April 1988 it sued Katz and PKWARE for trademark infringement, for copying ARC’s “look and feel” and for lifting ARC code. On 2 August 1988 the parties announced a settlement: PKWARE paid SEA $62,500 under a confidential cross-license ($22,500 in back royalties and $40,000 in costs) and could distribute ARC-compatible programs only until 31 January 1989.

The bulletin-board community took Katz’s side. To many users, SEA had used lawyers against a better programmer who had improved a shared standard, and Katz’s admirers presented the case as David against Goliath. SEA won the lawsuit and lost the users: when Katz released a successor, sysops and users recompressed their archives in the new format, and ZIP replaced ARC on the boards.

PKZIP and the Open Format

Katz’s answer was a new program with a new file format that owed nothing to ARC. Mahoney suggested the name: “zip”, for moving at high speed. PKZIP 0.9 was released on 10 February 1989. With it came a text file, APPNOTE.TXT, describing the format in detail, and on 14 February 1989 PKWARE and Infinity Design Concepts announced that the .ZIP format was being placed in the public domain. Anyone could write programs that read and wrote zip files without paying PKWARE.

Others did. The Info-ZIP group, formed in 1990 around a mailing list, wrote free zip and unzip programs that ran on Unix and almost every other system. In December 1992 and February 1993 PKWARE released PKZIP 2.04 (versions 2.04c and 2.04g), whose new DEFLATE method combined LZ77 dictionary compression with Huffman coding (see Data Compression). Jean-loup Gailly and Mark Adler used the same method in gzip and zlib, and through them DEFLATE became the compression inside PNG images and HTTP, one of the most widely used compression algorithms in existence. The zip container spread just as far, into Java .jar files, the EPUB e-book format and, from 2007, Microsoft Office documents, which are zip archives of XML files.

Losing the Desktop

The open format helped zip and hurt PKZIP. When personal computing moved to Windows in the early 1990s, graphical programs built on the free specification, above all WinZip, became what “zip” meant to most users. Katz worked only in DOS, could not program for Windows and dismissed it as a fad; PKWARE did not ship a Windows product until 1996. In 2000 Microsoft built zip support into Windows ME itself as “compressed folders” (earlier available in the Plus! 98 add-on for Windows 98), using DynaZip, a commercial library derived from an early version of Info-ZIP’s code.

PKWARE stayed profitable in its niche. At its height it employed about three dozen people in an office in Brown Deer, north of Milwaukee, with annual revenue of around $5 million according to insiders, and big companies such as General Motors paid license fees for its tools. Employees remembered a relaxed office where Katz, who rarely used the president’s room, sat among the programmers, played Doom and argued about Star Trek. When PKWARE was inducted into the ShareWare Hall of Fame in 1997, Katz skipped the awards dinner.

The End

Rumors about Katz’s late nights and drinking circulated by the end of the 1980s. In 1990 he was charged with driving under the influence, and at least three more drunk-driving charges followed, along with warrants for driving with a suspended license and for bail jumping. He was sentenced to 45 days in jail for drunk driving in 1994 but never served it, first by appealing and then by failing to appear. In late 1994, after his mother asked the police to check on him, an officer found his condominium piled with garbage. Not long afterwards Katz left a letter instructing PKWARE’s chief operating officer to remove her from the company; she left and eventually accepted a buyout. From the end of 1995, fearing arrest, he stopped coming to the office and avoided his home, living in hotels and running the company by fax and e-mail. In 1997 the city of Mequon obtained a warrant to clean out his condominium after neighbors complained of the smell.

In February 1999 friends took him to a hospital for treatment, and he stayed sober for several months. He began drinking heavily again early in 2000. On 14 April 2000 a hotel maintenance worker on Milwaukee’s South Side found him dead on the floor of his room, holding an empty bottle of peppermint schnapps. The medical examiner recorded acute pancreatic bleeding caused by chronic alcoholism. He was 37. His mother told the medical examiner she had not seen him for five years.

PKWARE passed to his mother as next of kin and was sold in 2001 to a management team led by George Haddix and backed by the investment bank Grace Matthews. It turned to data security and compression for businesses, and it still maintains APPNOTE.TXT, the specification Katz set free.

Dead End

ARC is the clearest dead end in the story. SEA owned the leading archive format of the bulletin-board era and chose to defend it in court against the programmer who had made it faster. It won the case and lost its users to ZIP, because a format on the boards was worth only as much as the willingness of thousands of sysops to keep using it. Katz learned the opposite lesson: he gave the zip format away, which ensured that it outlived both ARC and, in practical terms, his own program.

📚 Sources