History

There were secrets to those IBM machines that had been painstakingly learned by some of the older people at MIT with access to the 704 and friends among the Priesthood. Amazingly, a few of these programmers, grad students working with McCarthy, had even written a program that utilized one of the rows of tiny lights: the lights would be lit in such an order that it looked like a little ball was being passed from right to left: if an operator hit a switch at just the right time, the motion of the lights could be reversed--Computer Ping-Pong! This obviously was the kind of thing that you'd show off to impress your peers, who would then take a look at the actual program you had written and see how it was done.

To top the program, someone else might try to do the same thing with fewer instructions--a worthy endeavor, since there was so little room in the small "memory" of the computers of those days that not many instructions could fit into them. John McCarthy had once noticed how his graduate students who loitered around the 704 would work over their computer programs to get the most out of the fewest instructions, and get the program compressed so that fewer cards would need to be fed to the machine. Shaving off an instruction or two was almost an obsession with them. McCarthy compared these students to ski bums. They got the same kind of primal thrill from "maximizing code" as fanatic skiers got from swooshing frantically down a hill. So the practice of taking a computer program and trying to cut off instructions without affecting the outcome came to be called "program bumming," and you would often hear people mumbling things like "Maybe I can bum a few instructions out and get the octal correction card loader down to three cards instead of four."

-- John McCarthy's grad students playing golf ("program bumming") in machine language on the IBM 704 computer in 1959 (from Hackers, Heroes of the Computer Revolution by Steven Levy)

I recently read an article where Knuth describes a game of golf on a 50's machine. The machine read cards, and one could put 8 machine instructions on a card. The goal was to write a program fitting on one card that would read in a number, and reverse it. Noone was able to do it, but Knuth once stunned his fellow students. He came in, put a card in the reader, entered the number 123456789, ran the program, and the program outputted 987654321. Of course, the program would always output 987654321, regardless of the number on input. Later, the machine got an extension. A whopping extra register, with front panel toggles. Which could be used to store an instruction as well. Knuth was delighted that that enabled him to write a single card program that reversed a number.

-- Abigail talks about Knuth on comp.lang.perl.misc 23 March 2001

The competent programmer is fully aware of the strictly limited size of his own skull; therefore he approaches the programming task in full humility, and among other things he avoids clever tricks like the plague. In the case of a well-known conversational programming language I have been told from various sides that as soon as a programming community is equipped with a terminal for it, a specific phenomenon occurs that even has a well-established name: it is called "the one-liners." It takes one of two different forms: one programmer places a one-line program on the desk of another and either he proudly tells what it does and adds the question, "Can you code this in less symbols?"---as if this were of any conceptual relevance!---or he just says, "Guess what it does!". From this observation we must conclude that this language as a tool is an open invitation for clever tricks; and while exactly this may be the explanation for some of its appeal, viz., to those who like to show how clever they are, I am sorry, but I must regard this as one of the most damning things that can be said about a programming language.

-- Edsger Wybe Dijkstra (1930-2002) describes golf being played in APL in his 1972 essay: The Humble Programmer (cited in On Golf)

As indicated by the quotes above, playing golf in computer languages started long before Perl did. It's just that it wasn't called golf back then.

Perl's main claim to fame in this golf business is that a Perl hacker, namely Greg Bacon, actually coined the term Perl Golf in a comp.lang.perl.misc thread on 28 May 1999. And the term golf has been applied to other computer languages since then. Ironically, Greg may have intended the term as derogatory -- just as Fred Hoyle coined the term "Big Bang" to mock that theory -- because Greg once dismissed golf as really, really boring.

As you might expect, the earliest exponents of Perl Golf were probably Larry and Randal. For example, here is an early golfish Larry post from March 1990. And in this one Larry implies that Randal was his golfing buddy in those early days.

In March 1995, Adam Back and Hal Finney caused a stir by implementing RSA encryption and decryption in Perl as a signature file. Later Clifford Adams and others joined in the fun.

In September 1995, in what was probably the first golf contest with money prizes, the surprising winner of this comp.programming.contests C golf contest was a Perl hacker (some might say the Perl hacker;-). The goal of this competition was to find the shortest C program that reverses all words in a file. I found this exchange amusing.

: >Well, if that's your goal, you're gonna have to work a little harde
+r.
: >I sent this out last week, but it never made it off my machine.  It
+'s
: >50 characters, counting the final newline.
: >
: >    main(){system("perl -pe 's/\\w+/reverse$&/eg'");}
: >
: >It's almost readable, too.
: >
: >Larry Wall
:
: Well, if we're allowed to write our own interpreters then yours is
: easy to beat!
:
:     main(){system("a");}
:
: 20 characters! (Not counting the final newline.)
:
: You haven't heard of my interpreter "a"?  Well it just so happens
: that, without any command line arguments, it filters stdin to stdout
: in a manner that is identical to any of the valid entries to this
: competition, like your entry above for example.  :-)
:
: John Ellson

Well, hey, I said you'd have to work a *little* harder, not a lot hard
+er.  :-)

Larry
[download]

Jumping forward to July 2000, Uri Guttman initiated the ill-fated Perl Golf Apocalypse scheduled to run, accompanied by witty commentary, in front of a live audience at TPC 4. Alas, a number of technical glitches caused the event to be cancelled, replaced by a Damian Conway talk. Given that Larry was in the audience, it's little wonder that Uri is still scarred by it today. To my knowledge, no one has been brave enough to attempt a live Perl Golf contest since. The Apocalypse questions formed part of a golf competition run by jmcnamara in PGA Golf, Round 1 and PGA Golf, Round 1 - Results here at the monastery. I note with interest that mtve found some improvements to the original contest solutions some years later (this seems to be a speciality of mtve and thospel).

During 2001, there was such a rush of golf posts here at the monastery, that chipmunk sent out a plea, Tips on Writing a Golf Challenge, to calm things down a bit. Despite Perl Golf - A proposal for a new section of PerlMonks.org PM never did get its own dedicated golf section.

Just before Christmas 2001, the fwp mailing list exploded with a flurry of golf threads, starting with the Santa Claus Golf Apocalypse won by Eugene van der Pijll. This tournament introduced an innovation: strict rules and a test program that all entries must pass before being accepted. A much improved version of this early test program can be found here (update: the original test program can now be found here). These golf contests spawned a new mailing list golf@perl.org, split off from fwp.

A new magazine, The Perl Review, launched early in 2002, sponsored a series of golf tournaments run at SourceForge by Dave Hoover and Jérôme Quelin. These tournaments were run monthly throughout 2002, the top five money winners at the end of season being:

1  Ton Hospel                 $ 4,384,000.00 (10/13)
2  Rick Klement               $ 3,712,000.00 (12/13)
3  Eugene van der Pijll       $ 3,540,000.00 (10/13)
4  Stephen Turner             $ 1,955,200.00 (11/13)
5  Mtv Europe                 $ 1,837,040.00 (9/13)
[download]

In 2003, the TPR tournaments moved to Terje Kristensen's minigolf site where Ton Hospel, Mtv Europe and Rick "tybalt89" Klement continued to dominate. Around this time, golf enjoyed a surge of popularity in Poland, courtesy of games run at the kernelpanic.pl web site and the Polish comp.lang.perl newsgroup. After two to three seasons, these sites became inactive and the game of Perl Golf transmogrified into Code Golf.

Serious golf competitions in Perl, Python, Ruby and PHP are nowadays run at Code Golf (update: now at code-golf.io with github login), while less serious ones (in 51 different languages) are played at shinh's golf site. Update: As indicated at The golf course looks great, my swing feels good, I like my chances (Part IV), code golf is much more popular in Python than Perl nowadays.

What is Perl Golf?

Is it an art, a sport, just a bit of fun, a waste of time, or a threat to sound coding practices? I don't want to dwell too much on this; it is different things to different people. Hopefully, as with obfuscation, people are sensible enough to not start golfing on production code. My personal view is that golf is mostly about fun and competition.

As for the usefulness of Perl golf, apart from being a fun way to explore TMTOWTDI, notice that Perl golf, by its nature, pushes the language to its limits, often exercising boundary conditions. It should come as no surprise therefore that quite a few perl bugs have been unearthed during golf tournaments, especially by Ton Hospel. Update: and much later by Discipulus.

Golf Competitions

I think Felix (the contest judge) made a bad judgement, because the winning program was not nearly as confusing as mine---it was actually very simple.

-- MJD on TPJ OPC 5 judging at perl.plover.com

Competitions requiring judges, such as Olympic gymnastics and obfuscated Perl competitions, are easily spoiled by poor judging, as alluded to above by MJD. The wonderful thing about Perl Golf competitions compared to obfuscated ones is that golf is typically judged simply by counting up the number of characters in your solution. I've never seen any ugly disputes or complaints about who should have won. Update: long after this node was written, I did see one post-game dispute ... though it seemed to be just one newbie competitor complaining (see Re^4: Perl Golf Ethics for the gory details).

Of course, the winning golf solution may be horribly ugly or highly inefficient. So several people have suggested alternative types of competitions, judged by different criteria. Perhaps the most commonly suggested is the 100 metre dash where each solution is timed and the fastest wins. A novel twist on this idea is to measure not the CPU or wall clock time, but the number of Perl ops. For more information on Perl ops see Nicholas Clark's excellent paper "when perl is not quite fast enough". Yet another alternative is MJD's Perl Quiz of the Week where human judges comment on the "quality" of a solution after a 60 hour grace period. Update 2019: As mentioned at Weekly Challenge, manwar has started a Perl Weekly Challenge, where, similar to MJD's Perl Quiz of the Week, humans comment on interesting solutions submitted. Also, bliako suggested Perl Monks host coding contests: Coding challenges to PM.

The story goes that Marshall discovered this variation 9 years before he played it, and avoided defending with the Ruy Lopez for 9 years until he had a chance to spring it on Capablanca.

-- Quote taken from note by Bill Wall (any relation to Larry?)

It's a mark of Capablanca's genius that he was able to refute over the board a variation his grandmaster opponent had been studying and preparing for nine years.

In similar style to Capablanca, during the TPR tournaments four expert golfers would sometimes cook up a problem, study it in consultation for a month, and conclude the best solution possible was, say, 120 strokes. Yet within a couple of hours, the remarkable Ton Hospel would then go and post a score of around 100!

In many ways, Perl Golf tournaments resemble a fiercely contested chess tournament.

Some Classic Golf Games

One golf that seems to come up again and again is the famous 99 bottles of beer song. It's been discussed here at the monastery in: 99 bottles, 2 lines, 99 Bottles Of Beer (can't we do better), 99 bottles of beer on the Wall, Golf: Back to those damn beer bottles again... and Bottle Golf. For the record, I believe the shortest known solution is not in any of those nodes, but was concocted by thospel and mtve as follows:

sub
b{[@b=(abs||No,bottle."s"x!!++$_,of,beer),on,the,wall]}print
"@{+b},\n@b,\nTake one down, pass it around,\n@{+b}.\n"
for-pop||-99..-1
[download]

This version takes a command line argument specifying the number of bottles (default 99). If you find a shorter one, respond away! May 2011 Update: Shorter bottle golf solutions are now described in Drunk on golf: 99 Bottles of Beer.

The most interesting golf games occur when several solutions, all with a similar score, are possible. To illustrate, consider TPR(0,5a), an infix to postfix converter. This problem could be solved with: an operator stack; via overload; a recursive regex; top-down parsing; or Eugene's winning Mad Dutch algorithm (invented for this game). Sometimes, a truly astonishing approach is found, as when Jukka Suomela in TPR(0,4b) observed that perl itself has a built-in topological sorter: the garbage collector. Chris Dolan's explanation can be found here.

Magic Formulae

There's just no sane regularity in this thing. But in "random" mappings with a very small result set like this, the shortest solution is often to make up some magic formula that has no particular meaning, but just happens to give the wanted result.

-- Ton Hospel explaining his original decimal-to-roman magic formula

A crucial skill for serious golfers is the dark art of unearthing magic formulae. Though he didn't call them "magic formulae" back then, Ton Hospel invented them back in 2002. In 2006, he further popularized them by unleashing a mind-bogglingly complex Decimal-to-Roman magic formula to win the $350 cash prize on offer in the 2006 Fonality Christmas golf challenge. Incredibly, primo found a one-stroke improvement to Ton's masterpiece six years later. Proving this was no fluke, primo made a massive improvement to another ancient Ton Hospel winning golf solution.

Inspired by Ton, and emphasizing their importance in golf, I concocted four different magic formulae in four different languages (Perl, Ruby, Python, PHP), to win the follow-up Roman-to-Decimal golf challenge at the code golf web site. That is, in all four languages, the key component was not the code, but the magic formula.

This program might contain the fastest known existing implementation of full forward crypt

-- Ton Hospel sensationally wins a Perl golf tournament by writing the fastest C program

Perversely, searching for a magic formula can sometimes transmogrify a golf challenge from writing the shortest program to writing the fastest one! Examples of this can be found in the Ton Hospel quote above and in The 10**21 Problem (Part I) and at Re^2: The golf course looks great, my swing feels good, I like my chances (Part III) and Re^2: The golf course looks great, my swing feels good, I like my chances (Part III).

Compression in Golf

The first golfers to employ code compression techniques in golf were Marko Nippula of Finland and Stefan 'Sec' Zehl of Germany in 2002. In the TPR(0,4) "Interlinear Counts" golf competition, played during the first week of June 2002, Marko submitted a 169-stroke "pack u" compressed solution and 'Sec' a 244-stroke one, both well behind Ton Hospel's (uncompressed) winning score of 150. This controversial technique was banned in subsequent Perl golf competitions in 2002, but is widely used in code golf competitions today, as detailed in Compression in Golf: Part I.

Some Standard Golfing Techniques

Common operations should be "Huffman coded." That is, frequently used operators should be shorter than infrequently used ones.

-- Larry Wall in Apocalypse 3

: OK.  Still, it'd be nice to get anonymous scalars somehow.

The Huffman encoding of that is do{\my$x}.
[download]

-- Larry Wall on p5p, 30 April 1999

An excellent introduction to the basics of shortening your Perl code be found in Perl Golf 101.

As discussed in this thread (update: and this thread and here), some well-known golfing techniques are:

Golfing Technique                     Inventor           Year
-----------------                     --------           ----
@{[]}  aka baby cart                  The Larry or       1994
                                      The Schwartz
}{     aka eskimo greeting            The Abigail        late 1990s
~~     aka inchworm                   ???
~-     aka inchworm-on-a-stick        The Hospel         2002
$_ x= boolean expression              The Larry          early 1990s
y///c aka Abigail's Length Horror     The Hall           1996
stuff value into $\ for printing      The van der Pijll  2001
}for(...){ variation of eskimo        The Hospel         2001
--$| magical flip-flop                The Hospel         2002
\$h{X} is one less than ++$h{X}
       aka Thelen's Device            The Thelen         2002
-i and $^I for data value             The Sperling       2002
Control @- via regex parens aka       The ySas           2008
ySas' device, see [id://851128]
[download]

Update: As described in the "Inchworm on a stick" section at perlsecret, the high precedence decrement ~-$x inchworm-on-a-stick only works for positive $x while the high precedence increment -~$x inchworm-on-a-stick only works for negative $x ... though with use integer they work for all $x (as they do in Ruby and Python golf).

References

• What: golf and benchmarks, Where: snippets, When: from now on, Why: for fun and glory!
• Golf Do-s and Don't-s
• Another golf style?
• On Golf
• Masem home node with many PM golf links
• Tips on Writing a Golf Challenge
• PGA Golf, Round 1
• PGA Golf, Round 1 - Results
• Book Idea: The Golfer's Handbook
• Fun With Perl: Golf
• Golfing and Sprinting
• MoleSter update - Golf
• Perl Golf 101
• USD $350 Cash First Prize for Perl Golf Competition
• Perl Golf Ethics
• Dueling Flamingos: The Story of the Fonality Christmas Golf Challenge
• Golf: Magic Formula for Roman Numerals
• The golf course looks great, my swing feels good, I like my chances (Part I)
• The golf course looks great, my swing feels good, I like my chances (Part II)
• The golf course looks great, my swing feels good, I like my chances (Part III)
• The golf course looks great, my swing feels good, I like my chances (Part IV)
• The golf course looks great, my swing feels good, I like my chances (Part V)
• The golf course looks great, my swing feels good, I like my chances (Part VI)
• Drunk on golf: 99 Bottles of Beer
• Compression in Golf: Part I
• The 10**21 Problem (Part I)
• ySas Device
• Larry plays golf with himself (and Randal)
• Larry golfing again
• Original Perl RSA program posted to cypherpunks list 9 Mar 1995 by Adam Back, response by Hal Finney
• RSA signature file in Perl history
• Larry wins money prize in comp.programming.contests C golf contest, Sep 11 1995
• Greg Bacon coins the term Perl Golf
• ircnet #perl golf challenge, Christmas 2001
• Ton Hospel's generic golf tester
• TPR Golf Contests
• terje's mini golf site
• Terje/mtv pdf book about Perl Golf
• Polish Perl Golf site
• perl golf mailing list archive
• Final TPR Career Money Leader List
• Perl Golf Apocalypse Invite
• First ever Perl 6 golf post cited in this 2002 Perl 6 Summary
• c2.com PerlGolf entry
• Shortest Wiki Contest
• Shortest Sudoku Solver in many different languages by Mark Byers
• Golf competitions in Perl, Ruby, Python or PHP
• PHP golf site
• Golf competitions in Vim
• Golf games played in 51 different languages (under construction by shinh)
• stackexchange codegolf and programming puzzles question and answer
• codegolf books (see especially hallvabo answer which mentions Japanese "Short Coding" book by Ozy)
• Sphere online judge (SPOJ) programming competitions
• Contest Coding with human judge (not golf)
• Golf competitions in C (beta)
• MJD's Perl Quiz of the Week
• Larry Wall Articles and Interviews
• Damian Conway interviewed by Builder AU
• Interview with Dennis Ritchie, Bjarne Stroustrup, and James Gosling
• Some quotes about programming languages
• history.perl.org Perl Timeline
• Python golf contest
• Nicholas Clark YAPC::Eu paper: when perl is not quite fast enough
• Edsger Dijkstra manuscript archive
• Hackers, Heroes of the Computer Revolution by Steven Levy
• Code golf (wikipedia)

References Added Later

• New codegolf site (codegolf-io, github login)
• I crashed 99 bottles.. the dot was the point (multiconcat)
• Weekly Challenge
• Coding challenges to PM
• Obfuscated Perl Contest run by Jon Orwant and Felix Gallo at The Perl Journal (wikipedia)
• Perl Golf - A proposal for a new section of PerlMonks.org
• Re: What does your old Perl code look like? (discusses the 2001 Santa Claus Golf Apocalypse)
• code golf stackexchange (shortest game of life)
• code.golf (new golf site, see thospel and primo)
• Re^2: How to test for empty hash? (Perl Secret Operators)

Other Articles in This Series

• The Lighter Side of Perl Culture (Part I): Introduction
• The Lighter Side of Perl Culture (Part II): JAPH
• The Lighter Side of Perl Culture (Part III): Obfu
• The Lighter Side of Perl Culture (Part V): Poetry
• The Lighter Side of Perl Culture (Part VI): April Fools

Updated Aug 2011: Added John McCarthy (1959) and RSA (1995) reference; Dec 2013: added ySas' device; Oct 2014: Added new "Magic Formulae" and "Compression in Golf" sections. April 2019: Added reference to Weekly Challenge. Added note that Python golf has become more popular than Perl golf.