The History of the ZIP File — A Lawsuit, a Format, and a Tragedy

Open almost any ZIP file in a hex editor and the first two bytes read PK. They are a man's initials. The format exists because he was sued, and the way he responded to that lawsuit is the reason ZIP is now inside every Word document, every Android app and every ebook on the planet.

Before ZIP

In the mid-1980s, home computers connected over modems at a few hundred bits per second, and bulletin board systems charged by connection time. Compression was not a convenience; it was the difference between a download taking twenty minutes and taking an hour.

The dominant tool was ARC, from System Enhancement Associates, released in 1985. It did two things at once — bundling several files into one and compressing them — and it became the standard format for software distribution on BBSes.

ARC was shareware. The format was not documented for others to implement, and SEA treated it as proprietary.

Phil Katz

Katz was a programmer in Milwaukee, and in 1986 he wrote PKARC — a tool that read and wrote ARC files considerably faster than ARC itself, with the compression routines hand-written in assembly language.

It was, by all accounts, dramatically better. On the hardware of the day the speed difference was not marginal; PKARC displaced ARC across the BBS world within months. Katz founded PKWARE to sell it.

The lawsuit

In 1988, SEA sued Katz for copyright and trademark infringement over PKARC.

The case settled. Katz paid damages, agreed to stop distributing ARC-compatible tools, and the matter closed — but not before becoming one of the first genuinely public disputes in the shareware community about who owns a file format. Opinion divided sharply, and SEA's reputation among BBS users suffered badly for pursuing a developer whose product was simply better.

What matters is what Katz did next.

The reasoning is straightforward in hindsight. A format that anyone can implement gets implemented by everyone, and a format everyone implements becomes the one users expect. Katz sold the best tool for a format that had become the standard — a better position than owning a format nobody else could support.

SEA continued to treat ARC as proprietary and faded from relevance within a few years.

The design

ZIP's technical structure reflects the constraints of its era and has aged remarkably well.

Why the index is at the end

This is the design's most consequential choice, and it buys three things:

• Append-only updates. Adding a file means writing it and rewriting only the directory — not the whole archive. On a floppy disk in 1989 that mattered enormously.
• Self-extracting archives. Because readers seek to the end to find the index, a ZIP can be appended to an executable and both remain valid. The program runs; the archive is still readable.
• Streaming creation. Files can be compressed and written one at a time without knowing the total size in advance.

The cost is the failure mode anyone who has downloaded a large archive has met: lose the last few bytes and the entire file is unreadable, even though nearly all the data is present. A truncated ZIP is not partially recoverable by default — the map is gone.

Compression

ZIP separates the container from the compression method, which is why it has outlasted its original algorithms. Early versions used methods called Shrink and Implode; DEFLATE, introduced in 1993, became the standard and still is.

DEFLATE combines LZ77 — replacing repeated sequences with a reference to their earlier occurrence — with Huffman coding, which gives shorter bit patterns to more frequent values. It is not the most efficient algorithm available, and it is fast, well understood, patent-free and implemented everywhere. It is also used by gzip, PNG and HTTP compression.

How ZIP ended up inside everything

The open specification produced a second-order effect Katz cannot have anticipated. Because every platform already had a ZIP implementation, it became the obvious choice whenever anyone needed a container for structured files.

Each of these teams faced the same problem — bundle several files, compress them, guarantee every platform can read the result — and reached the same conclusion. Designing a new container would mean writing readers for every language and operating system. Using ZIP meant it worked immediately, everywhere.

What happened to Phil Katz

The story does not end well, and it is worth telling honestly rather than as a footnote.

Katz struggled with alcoholism through the 1990s. PKWARE remained a successful business but he withdrew progressively from it, and the last years of his life involved repeated legal trouble and periods of homelessness despite considerable wealth. He died in a motel room in Milwaukee in April 2000, aged 37, of complications from chronic alcoholism.

He was 37 years old and had written a file format that, twenty-six years later, is in essentially every computing device in existence. Every Word document opens with his initials.

Zip bombs

One consequence of ZIP's design deserves a mention because it is still an active security concern.

Compression ratios have no upper bound. A file of a billion identical bytes compresses to almost nothing, so a small archive can expand to something enormous.

The classic example is 42.zip — roughly 42 kilobytes containing nested layers of archives that, fully expanded, produce about 4.5 petabytes. Any system that decompressed recursively without limits would fill its disk and fall over.

The countermeasure was to limit recursion depth, and more recent work defeats that too: by overlapping compressed streams so that a single layer of files shares the same compressed data, a modern zip bomb reaches ratios above 28 million to one without nesting at all.

The lesson in the story

ZIP is a case study in a pattern that has repeated many times since. Two parties had a format. One tried to control it and kept the value close. The other published the specification and competed on implementation.

The controlled format disappeared. The open one became infrastructure — so thoroughly that its inventor's initials are now in the first two bytes of billions of files created daily by people who have never heard of him.

Katz kept the commercial value where it was defensible: in a fast, well-engineered implementation. He gave away the part that only had value if everyone used it. That is a better bargain than it looks, and the same reasoning explains why PDF, HTML and UTF-8 succeeded where their proprietary contemporaries did not.

Summary

• PK is Phil Katz, and those bytes begin every ZIP file.
• ZIP exists because of a 1988 lawsuit over an ARC-compatible tool.
• Katz published the specification openly and sold the implementation — the decision that made ZIP universal.
• The index sits at the end, which enables appending and self-extracting archives, and makes truncation fatal.
• DEFLATE has been the standard method since 1993 and is also used by gzip, PNG and HTTP.
• DOCX, APK, JAR, EPUB and more are ZIP archives — which makes them openable and often recoverable.
• Zip bombs exploit unbounded compression ratios. Limit output, not input.
• Katz died in 2000, aged 37. His initials open billions of files a day.

Frequently Asked Questions